Diary of a Technocratic Anarchist

2011-12-17T09:22:00.000-08:00

Solving a nagging question about print adhension

2011-12-16T17:20:00.000-08:00

Unlike most of you, I don't use an electrically heated print surface. Some time ago I bought a Rapman 3.1, which used an acrylic 3 mm print table. I soon discovered that 3 mm was far too thin and quickly warped beyond use. Switching to 10 mm solved that problem.

After a long time of successful prints, I noticed that with winter causing colder temperatures in the print room I was having more and more trouble getting my prints to stick to the acrylic. I tried cleaning it and sanding it with little avail. Electrically heated print tables were just coming available but insofar as printing was concerned, I thought that things were already complicated enough without adding that sort of equipment to my Rapman.

I had an IR heat lamp in the lab, detritus of another experiment, and discovered that using it on a tripod to raise the temperature of the acrylic print table above 40 degrees Celsius measured with a non-contact IR thermometer gave me consistent adhesion. I soon discovered that I could turn off the IR lamp after 4-5 print layers with no ill effects. It was not needed for the rest of the print.

The rig looked a bit like this...

Note that the lamp is placed at a 45 degree angle to the acrylic print table.

I soon noticed that adhesion at the near side of the print table was much less firm than that at the back and less firm at the left side than the right. I attributed this to various things, uneven heating being one of the possibilities. While the left/right difference made sense the front/back difference didn't seeing as the IR lamp was aligned with the left/right axis.

Cranking the terminal heating temperature before starting a print to about 50 degrees solved most of the problem for the center of the table and I was able to print along the front/back axis with reliable success. Unfortunately, the back side of the print area seemed to have the print pad melting into the acrylic while the front side would separate easily.

It made no sense. I thought for a while that it had something to do with the acrylic plate and rotated it with no effect. Swapping ends and sides always left the back side of the print table very firmly attached to the print pad. While that wasn't a horrible situation it was annoying, because it meant that processing the printed objects after separation became more time consuming.

A few weeks ago, I purchased a FLIR E30 thermal imaging camera with the intention of learning more about what was happening with prints as they were being laid down, the ultimate goal being building in advanced heuristics into my Slice and Dice app which converts STL files into Gcode. I also had hopes about eventually doing some research into what actually happens thermally with extruder hot ends with the notion that I might be able to design a better one.

Yesterday, the E30 arrived and I decided that a good beginning exercise might be to look at the distribution of heat on my acrylic print table when I used the IR lamp in its standard configuration to heat it. The results were quite unexpected.

>

The lamp put down a marked hot spot at the upper right rather than at the right as I expected. The upper right was exactly where I had the most trouble with print pad melting. Obviously, the IR lamp did not give even heating when tilted but overheated in on the upper right.

This was nasty. I had previously thought about using several smaller IR lamps at the corners of the Sampo printer that I have been developing. If the smaller lamps behaved like my single, large one, however, this might not be a good idea at all.

I then got to thinking about how IR lamps are actually used in food heating cabinets. They are almost always placed point straight down. I rearranged my tripod to place the lamp almost vertically over the acrylic print table.

That sorted out the temperature distribution problem...

A new acquisition...

2011-12-09T09:40:00.000-08:00

Details

Sampo's touch screen begins to work...

2011-08-16T18:39:00.000-07:00

Adriaan has been working hard learning the TFT programming protocols and has his first touch screen menu working on the Sampo controller board.

Bogdan makes a measurement suggestion...

2011-07-27T23:16:00.000-07:00

After I got the x and y axes operating independently, Bogdan suggested that I measure the steps between the limits switches to see how much difference in measurements might be attributable to the mechanical microswitches that I am using. I had already been recording that with the y axis, so I decided to extend the monitoring code a bit.

With the y-axis, I had been simply writing the number of pulses to an SD card whenever a limit switch was tripped. In that I was running the axes at 65 mm/sec, I was getting a rather substantial thump whenever a switch was encountered. When I thought about it, I began to suspect that the impact was a result of the time it was taking the write to the SD card to happen in that I was doing that immediately after the switch was tripped. I changed the code to record a set number of triggering events for each axis and then exit the stepper loop and print the whole set of measurements at one time. That reduced the noise of switch triggering on the x axis to almost nothing. It also reduced the noise from triggering events on the y axis, but not as much. Considering the y-axis is shifting the whole weight of the x axis assembly, the extra momentum generated thereby is probably causing the larger thump.

I first took a set of 50 triggering events running at 65 mm/sec.

You can see that the two limits switches on serving the x axis trigger with slightly different sensitivities, one triggering about 0.3 mm greater than the other {transition is running at 0.89 mm/step for both axes}. The both y axis limits switches trigger at the same place except that occasionally one gets moody and triggers 8 steps {~0.6-0.7 mm} longer than the first.

I then took another set of measurements at 32.5 mm/sec. The NEMA 23s were near resonance frequencies at this speed raising the noise level of the printer considerably. I will have to see about damping this.

What you can see is that the variation on the x-axis stayed about the same while the moodiness of the one limit switch on the y axis disappeared. Notice also that the steps between switches are down.

From there, I took a set of measurements at my usual printing speed for Rapman at 22 mm/sec.

Decreasing the transition velocity got us further away from the resonance frequencies of the NEMA 23s. It must be said, however, that the printer was still louder than when I was running it at 65 mm/sec. You can notice here that the variation in limit switch triggering has dropped to 1-2 steps.

This has been an interesting exercise. One thing that is obvious now is that to control noise levels I should be controlling the stepper speed both by the delays between steps and by adjusting the level of microstepping that I am using.

X & Y Axes operational from the controller

2011-07-27T14:14:00.000-07:00

I was finally able to get time to integrate the full anti-bounce board with the x and y axis limit switches.

The two axes are playing ping pong and running at a speed of 65 mm/sec with no slippage and no heating of either the steppers or the driver chips. I've run them all morning with no mishaps.

Now I am going to have to see to writing a gcode interpreter and taking a shot at the TFT 320x240 graphics touch screen for system control.

Leveraging Bogdan's anti-bounce circuit for Sampo...

2011-07-20T22:33:00.000-07:00

In developing the Darwin-derivative, Rapman-derivative Sampo 3D printer project as a kaizen exercise I utilized the same sort of microswitches for limits checking as are specified in the Rapman design. I soon discovered that the switches have a formidable electronic bounce. I was able to control that using the button function in my firmware compiler for the y-axis. The computations taken for a firmware fix, however, were going to put a terrific drag of my MCU that I didn't want to have to deal with.

Enter Bogdan Kecman with helpful suggestions on how to put together an antibounce circuit for the limits switches.

I had last built an antibounce circuit in 1981, so his help was greatly appreciated. I built a lashup of the circuit to check to see that the component values were right and then went on to design a board to handle all six limts switches. I wanted six instead of Rapman's three because a lot of problems that I'd had with Rapman stemmed from the fact that it has limits switches only on one end of its axes. When things went bad one could find steppers trying to skate off of the unchecked far end of axes. As well, Rapman limits checking only seems to be done when one is resetting the axes at the beginning of a print. I want to do better than that.

I bought components and dug out my stripboard and had a go at the design. Some time before I put together a stripboard design program after having had no luck with the ones I was able to access on the web. Eventually, I evolved this board.

Frontside...

Backside...

It has been some time since I built a board, so I found putting this one together quite frustrating. I was about to give up this evening after making a bunch of mistakes and then got angry to the point of rage. The adrenalin let me get the @#$#@$ thing finished.

Tomorrow I will drill out the breaks in the strips, check the board for continuity and, if I have enough time, try to rig it into Sampo and extend the firmware to utilize it. I suspect that will have to wait till the weekend, however.

Mendel z-axis stepper mounts done

2011-07-13T17:44:00.000-07:00

I managed to get some hours together to do some more work on my son's Prusa Mendel, the z-axis stepper mounts this time.

I printed the z-axis stepper mounts at a 45 degree angle to minimize parts preparation time and avoid warping.

I'm now working on the x-axis stepper and idler mounts. My son processed the stepper mount last night and I did the idler mount this morning. They're printing this evening.

Mendel frame takes shape...

2011-07-10T20:36:00.000-07:00

I had a major crash of the Rapman 3.0 printer and for several hours I thought I was really knackered. One of the leads to a phase of the extruder stepper parted because of fatigue from thousands of hours of vibration. The system shut down and reset. At first I thought I had a simple static discharge event, the first in many months. I fired the system back up and discovered that the extruder stepper would dance around but wouldn't pump filament.

After serious prayers that the stepper driver chip for the extruder hadn't fried I rewired the connector and got the extruder pumping ABS again only to discover that the intense vibration from the stepper before the reset had actually shaken apart the Arcol extruder hot end that I had bought from László Krekács in Hungary.

This sounds worse than it was. I simply screwed it all back together and cleaned the extruder end and it worked perfectly again. Unfortunately, I will have to recalibrate Rapman now. That should take a few hours that I didn't have available today.

In any case, I was able to cobble the Prusa Mendel frame together this morning.

It's a dinky little thing, but interesting all the same.

Printing a Mendel derivative

2011-07-09T22:52:00.000-07:00

After talking with my son recently, I concluded that he needs to start printing a lot faster than I'm going to have Sampo debugged and duplicated. I have a lot of 8 mm linear shafting and linear bearings in stock, so I decided to build him a Mendel derivative using Sampo firmware and controllers. That should get him printing a lot faster than would otherwise be the case.

I started printing parts for a Prusa Mendel yesterday. So far, so good. Got the gantries finished and am printing the rest of the parts now.

5/16th inch threaded rod is pretty much a one on one replacement for M8, #8 machine screws replace M4s and #4 machine screws replace M3s. I couldn't see much point in printing the SAE Mendel. I'm going to have to redesign the extruder carriage to seat linear bearings instead of those strange PLA things it ordinarily uses.

Picking up speed

2011-07-05T12:09:00.000-07:00

The MIPS core that PIC32 uses is a very high performance CPU that was used in high end Windows workstations in the early to mid-1990s. It doesn't behave much like the 8 and 16 bit PIC chips, so it's taken me a while to get down the learning curve. The button function in the Mikroelektronika compiler library works, but requires about 10 msec to filter out the bounce when a limits switch is encountered.

Processing a button function for each step when I was running at half step slowed the y-axis down to 12-15 mm/sec. To get around that problem in firmware I wrote a smart limits switch routine that runs slow until it finds the first limits switch and then kicks the stepper motor up to full speed until it nears the other limits switch.

Using this approach lets me increase the maximum transition speed for the y-axis from 15 mm/sec to about 65 mm/sec for my firmware testing as you can see in the video clip.

It's worth noting that the Allegro driver chip has a maximum rating of 0.75 amps and the NEMA 23 is a six wire model drawing 0.5 amps per phase. I've wired it in series which brings that amperage down considerably. In spite of this I'm getting 65 mm/sec and both the stepper and the NEMA 23 are running quite cool. The driver chip requires no heat sink or fan.

Y-axis test firmware operational

2011-07-04T23:19:00.000-07:00

After a delightful chat with Bogdan this morning about the PIC32's MIPS core processor, I was able to get the y-axis test firmware working. Bogdan has done a lot of work with the PIC32 and is very generous with his knowledge. A few hours later, I had the y-axis responding to the limits switches.

Right now with sampling from the limits switches in the same loop that runs the stepper I'm getting 15 mm/sec. On its own the stepper can do about 52 mm/sec. I suspect that the speed of the axis will be getting a lot closer to that upper limit once I get the limits switches into an interrupt loop. :-)

Printing flexible cable guides...

2011-06-29T19:32:00.000-07:00

There is not much to say about this. Once I got two of the links printed and assembled so that I knew everything fit together, I bought a few hundred #4-40 3/4 inch machine screws and nuts to hold the parts together. I'd designed the parts to perfectly seat a 3/4 inch machine screw and nut.

When I got home with my trove of fasteners from my stockist I discovered that his Chinese supplier had been making a little extra money by trimming his 3/4 inch screws (0.75 inch) down to 0.714. What that meant was that the screws went all the way through the guide assembly but didn't emerge on the other side to allow the nut to be seated on the end of the machine screw.

My stockist is getting me some 7/8 inch machine screws as replacements and writing a hot note to the warehouse. Quality assurance at the Chinese plant needs a bit of a rework, I think.

Interestingly, I had designed the holes for the #4 machine screws so that the threads engaged the sides of the holes, so actually nuts weren't required. With that in mind I went ahead and assembled the flexible cable guide for the x-axis. It seems to work perfectly.

I get a tight turn like I'd hoped with no clashing. Right now I am up to 16 inches of a 24 inch assembly for the x-axis. When I get the full 24 inches printed and assembled I will design and print the end mounts.

It will be interestingly how many hours of operation this kind of flexible cable guide will handle before something wears out.

Flex cable carrier

2011-06-26T21:34:00.000-07:00

I've never been happy with the way that Rapman handles axis and extruder cabling, so I decided to print my own flex cable carrier system. I saw several possibilities in Thingiverse. Most of them were knockoffs of existing injection molded parts, however, and printed very poorly.

A few looked as if they were designed specifically for a 3D printer like this one...

I didn't much like this one largely because of the large turning radius for the flex. I wanted something more like this...

Since my wiring was considerably more modest, however, I wanted something with a bit sharper turning radius still. After several hours of trying out alternatives in Art of Illusion, I came up with this as a first try.

It is shown here with #4 bolts 1.5 inches long. It can work interchangeably with #4 UTS/SAE - 3/4 inch or M3 - 20 mm bolts. With fasteners it costs about $1.50/ft. Commercially available alternatives average about $12.50/ft.

The system can make a 180 degree flex within 50 mm. The next move is to get a couple of packets of #4 bolts and nuts and print a few feet of this to try with the x-axis cabling.

A "string wars" approach to the y-axis

2011-06-14T20:45:00.000-07:00

Darwin and it's direct derivatives uses two belt loops driven by a shaft connected to a NEMA 23 stepper.

In Sampo, that dual loop arrangement has been replaced with a single large loop

The belt guides are equipped with standard 608 skateboard bearings with printed fenders.

Topologically, the y-axis is simply an attenuated version of the x-axis.

Print table installed

2011-05-31T08:56:00.000-07:00

I had gone to considerable trouble to make sure that I sited the bolt holes in the MDF {medium density fibreboard} print table properly. To that end I designed a template for each of the brackets that would show me where the guide holes should be put.

Once I had the table suitably aligned, I was able to drill the guide holes with my Dremel tool quite easily.

I then removed the table and took it to the workshop to drill out the holes to #8 bolt diameter. When I returned and tried to mount the board I discovered that the holes didn't line up. I had neglected to mark the lower left corner of the print table and had no way of knowing which side our orientation matched my drilled holes. We are talking about a few mm here, mind. There were 16 possible orientations and that was complicated by the fact that the brackets were able to rotate around the z-axis linear shafts in the xy plane. After about the tenth possible orientation, I found one that fit 3 of the four brackets and just redrilled both the MDF and the bracket. Mercifully, solid ABS is very amenable to drilling so other than having the lower right corner showing an extra set of bolt holes, it all worked out quite well.

Hopefully, I will remember to mark the lower left corner the next time I build one of these. The print table moves quite freely as the video will demonstrate.

Now all that remains is for me to reprint the z-axis cable grippers and mount them and the z-axis will be complete.

Ready to install the print table

2011-05-30T16:45:00.000-07:00

I got the last of the y-axis brackets printed and light mounted.

It looks like the print table with be 420x420 mm. I've designed a bolt-on template that seats the table on the brackets and shows me where to site the guide holes for the mounting bolts.

And here is the first drill template mounted on a z-axis bracket.

Got it right on the second try! :-)

Stepping into firmware

2011-05-29T11:52:00.000-07:00

Having had good experience with the Rapman's PIC32-based controller, I decided to stick with that MCU for my new printer. As I mentioned earlier, rather than buy a $1k+ C compiler from Microchip, I bought a much less expensive, full-featured BASIC compiler for the PIC32 {they also offer C and Pascal compilers} and a full development board from Mikroelektronika in Belgrade. Friday night, with the last of the z-axis brackets being printed on my Rapman and the 19 June exhibition in San Francisco coming up, I decided that I'd better get cracking on the firmware.

When I bought the PIC32 development board from Mikroelektronika, I also picked up a little stepper controller board from them.

It uses an Allegro A3967 driver chip rated at .75 Amps. Now ordinarily I wouldn't have considered getting such a thing, but in this case it seemed reasonable to have a ready made stepper tester board that I knew worked with my development board and I knew I had a number of small stepper motors that I could use with it. Nothing I'd consider using on the printer, mind, but all the same useful in the learning process.

Like Darwin and Rapman, I'd decided to be conservative and use NEMA 23 steppers. The price on these has dropped dramatically since we were building Darwins several years ago. While shopping for NEMA 23 steppers, I happened across this little jewel.

Oddly enough, this 6 wire NEMA only drew 0.4 amps but produced a lot of torque. I bought it, too, just so I could have have a NEMA the right size that I didn't necessarily want to use on the printer. At that time I was looking at using the much heavier capacity Pololu stepper drivers that have recently proved so popular with Mendel electronics.

Now here is where serendipitous good fortune intruded. The firm selling this stepper also had a special on 24 volt power supplies.

When we began with Reprap about the only reasonably priced power supply we could lay hands on was a salvaged 5-12v ATX box out of old PCs. 24 volt supplies at the time were quite dear. We knew very well that we could get a lot better performance out of steppers if we used 24 volts, but nobody wanted to invest in a 24 volt supply. This bad boy put out 6.5 amps at 24 volts for $19. The economics of that were hard to argue with given that my development board power conditioning circuit would eat anything up to 30 v DC.

Friday night and Saturday I spent the necessary hours skating down the learning curve of the Mikroelektronika development board and compiler IDE. This took longer than it should have in that the PIC32 boards and compiler are very new to Mikroelektronika. As a result, while they had code samples in BASIC for their stepper controller board, they were for 8 bit PIC chips and development boards, not their new 32 bit offerings. I don't know why firms don't offer extremely simple sample code patches. Instead, they always clutter it up with nonsense that runs LCD boards and makes LEDs flash on and off prettily. Of course, how that works on an 8 bit board is very different than it is for a 32 bit board.

By Saturday afternoon, I'd managed to unclutter and migrate their code to PIC32 and had the stepper controller connected to the NEMA 23 working properly.

I knew there was a lot of friction in my cable z-axis system, so I did an initial gear design of 3.5:1 to insure that I got plenty of torque. I had rigged the 6 wire stepper in series at Bogdan's suggestion so that the amperage pull was considerably below 0.4 amps. Imagine my surprise when I discovered that there was ample torque even at half-step to happily push that stiff z-axis lead screw collar back and forth under serious load at 660 pps, a step rate just short of the resonance speed of the stepper. Even under those loads the controller chip never got above about 50 C even after several hours under load. That means that no heat sink is necessary.

When you translate that pulse rate out to an MXL belt driven x or y axis powered by an 18 groove pulley you get a calculated top speed of about 60 mm/sec. That's about three times the head velocity that I print at. It would appear that running a stepper with 24 volt power makes a very big performance difference.

Here you can see the stepper controller attached to the NEMA 23 and the PIC32 development board.

I've been thinking about that cool controller chip and that NEMA 23 and wondering about the possibility of driving a Wade extruder design with a NEMA 23. The technology and economics are certainly attractive.

I'm going to buy some more of those controllers and also a relay card so that I can control the hot ends and heat lamps on the printer.

Mikroelektronika certainly has a very big toolbox of accessory boards that let you prototype just about anything without having to build up circuitry from scratch. They're not as cheap as you could build from scratch, but if you count the time and cost of building up purpose made boards while you are developing a printer and not sure of everything you want in it, they're very cost effective.

Winding up the String Wars

2011-05-24T11:42:00.000-07:00

I was finally able to finish making the connection between the z-axis lead screw and the cable turnbuckle for the z-axis positioning system. Given that the 3/8-24 threaded rod moves 0.945 mm per full turn and that the 32:12 gear reduction coupled with the 200 step/full turn for the NEMA 23 we get something like 533.33 steps per turn or 0.00177 mm movement per step.

You can see the general layout of the z-axis lead screw with this pic...

I've circled, from left to right, the thrust collar nut, the cable turnbuckle and the linear bearings that make up the elements to transfer power from the stepper motor to the cable.

Here you can see the three connected...

Here is a detail of one of the four cable-driven lifts for the print table with the bracket in place...

I plan on having one fixed joint between the brackets and the table and two sliding joints constrained in the x and y-axes in the two brackets adjacent to it and a sliding joint in the xy plane opposite. I hope that will be stable enough.

Next, though, I have to see if I can finish the design and printing of the BfB hot end adapted Wade extruder derivative. If that takes too long I will fall back on the two full BfB extruders that I have in stock.

NEMA 23 connected to the Z-axis lead screw

2011-05-23T14:34:00.000-07:00

With the completion of the gear pair last night, the connection between the NEMA 23 and the Z-axis lead screw is complete.

That milestone meant that I had to move Sampo into the front room so that firmware development could begin.

A Z-axis gear set for the Sampo 3D printer

2011-05-22T22:23:00.000-07:00

Few experiences are more satisfying than a creating a useful device of great intrinsic beauty

Another battle in the string wars: cabled z-axis working

2011-05-21T20:13:00.000-07:00

It took me 10 hours and fifty minutes to print the first print table bracket and I got a few measurements wrong, but it appears that the cabled z-axis concept is going to work.

Since the speed of the z-axis is not particularly critical, I intend to use a fairly high gear ratio between the NEMA 23 stepper and the lead screw that drives the cabling to insure that I have adequate force to overcome friction in the system.

Refighting the String Wars.

2011-05-15T21:45:00.000-07:00

Back at the beginning of 2006, before there was even a Darwin, eD Sells at the University of Bath was designing Darwin's predecessor, ARNIE. Confronting the problem of designing a z-axis, eD adapted the kinematics that were used on old wire cable parallel bars found on drafting tables.

eD adapted this technology in 3 dimensions to allow a single stepper motor to raise and lower ARNIE's print table.

Having trained as an architect before the Great Flood, I immediately fell in love the idea and adapted it to my failed Godzilla Repstrap design.

eD encountered no end of trouble with the cabling idea and eventually abandoned it for an approach which used four pieces of studding {threaded rod}.

It tends to be forgotten but the first fully operational Reprap machine at Bath was ARNIE, not Darwin. Indeed, Bath's traditional whiskey shot glass, the second one printed after Vik Olliver's in New Zealand, was printed on ARNIE. This approach was refined in Darwin.

Rapman, a Darwin derivative, was put into serial production by Bits from Bytes and is still selling quite well, today.

This z-axis approach does have its problems, though. Studding is most definitely NOT a proper lead screw. When you undertake to use four pieces of studding to raise a 3D printer's print table, the tendency of studding to be not quite straight plus the fact that you are using four pieces of not quite straight studding can lead to some unpleasant consequences. Here is an extreme example of what can happen.

If you expand the pic, you can see a nasty juddering of layers taking place. Here is a more usual example of the effect.

This is an extreme closeup with the light accentuating the effect. The object is quite smooth to the touch. The effect is still there, though. Here is a more usual picture showing the effect.

If you expand the pic you can see a regular pulse peaking at every seventh layer. This varies depending on how you adjust your machine and how straight your studding rods are. The closer the alignment, the better your print quality.

Now Bits from Bytes set out to solve this problem in their out-of-the-box BfB 3000 printer. They used a single proper lead screw to drive a cantilevered print table.

Both the Ultimaker and Makerbot's Thingomatic use the same approach.

Recently, when I decided to kaizen the old Darwin design, I decided to see what I could do about the z-axis situation. I didn't like the cantilevered print table approach and I did not want to simply duplicate the 4 studding solution originally used. That got me to thinking about the old cabling approach that eD had used back in 2006. The problem with it seemed to be applying force to the cable to move the print table. eD tried to use a friction wheel and eventually gave it up.

It occurred to me that it might be reasonable to use a single studding lead screw to apply force to the cabling. Lead screws can apply LOTS of force. So why not just attach one to the cable at a convenient point and be off?

I am in the process of doing just that.

I have circled the lead screw's thrust collar, the cabling turnbuckle and a linear bearing. Those three elements will be connected and a NEMA 23 stepper used to drive the cabling to raise and lower the print table.

Here you can see a detail of the cabling scheme associated with a pair of linear bearings on a vertical shaft. I have got to design a connector between the cable, the linear bearings and a corner of the print table. Hopefully, this approach will let me get a smoother z-axis operation without the juddering so characteristic of the Darwin design.

Solid prints

2011-05-09T18:32:00.000-07:00

Printing solid, structural objects is quite an art. As I mentioned earlier, I am harking back to a very early effort that eD made with A.R.N.I.E, the precursor to Darwin. eD wanted to use a cabling system for several of the axes not unlike what you used to see for parallel bars on traditional drafting tables. eD finally gave up the effort when he couldn't get the cable to grip a sprocket wheel properly. I have another idea about how to do that which I will talk about in the near future.

Securing the cable and pulleys for the z-axis is tricky. I've spent quite a few days designing a lower corner mounting block for the pulleys.

At 69 cubic centimeters it's quite a large piece. One thing you learn very quickly is that when you are designing structural pieces you have to remember that you not printing an isotropic material but rather a grained material not unlike wood. As a result of this, it doesn't do to lay this block on one flat surface. That gets you a part that is strong in one plane and fatally weak on the other. Given it's size that's asking for corner curling in any case.

As a result, I decided to put the grain of the print at a 45 degree angle to both major planes.

Several tries at printing it finally got me settings that yielded a very clean, handsome print. The preparation time for this piece was about 20 seconds after I removed it from the print table.

You can see how clean the bolt holes are.

As are the recessed pockets for the hex head 5/16ths inch bolts. No warping. No corner curling. Pretty much a perfect print. The longest dimension is 80 mm.

The part mounted properly without problems.

Sampo: The return of Darwin...

2011-04-27T07:14:00.000-07:00

Actually, Darwin never really left.

Despite all of the hoopla about Mendel, a very robust Reprap machine which is virally spreading like a bad flu, Darwin derivatives have been quietly building up numbers largely through Bits From Bytes' Darwin-derived Rapman printer. It would be fairly safe to say that there are upwards of two thousand of these Darwin clones of one flavour or another in the field.

I bought a Rapman 3.0 in the Fall of 2009.

Rapman is basically a Darwin re-engineered for laser cut acrylic plastic instead of printed parts. I've never been able to decide whether Rapman is a crib of the Ponoko laser cut acrylic Darwin or vice versa. The Rapman was a brilliant choice for me. I had been working on a series of Repstrap systems. Early on in 2009 I realised that getting Tommelise 2.0 printing was going to take another half year at the rate I was going and what I really wanted to do was design things with a 3D printer in the design loop. I saw Batist Lehman's video of his Rapman in action and was sold.

Of course, Rapman has its little ways. It arrived disassembled as a flat pack. Inside were several sacks of metric nuts and bolts heavy enough, if put in a sock, to serve as a very effective cosh. Assembly was a daunting task and once finished there was a strong sense that you don't want to fool around with it for fear that it would break or some apart.

I was puzzled at first that lock washers hadn't been included with the nuts and bolts. It quickly became apparent, however, that if you put enough torque on nuts to make a lock washer compress the high impact acrylic that Bits from Bytes used would shatter. The net effect of this was that I keep a bowl beside my Rapman to collect the constant drizzle of nuts and bolts which unscrew themselves and fall on my work table.

Since then I've put well over three thousand hours on my Rapman. It became apparent some months ago that Rapman was not a heavy duty machine. I started seeing stress cracks all over the machine and was faced with a x-axis extruder mounting plate that just crumbled away from the heat of the extruder and the mechanical stress of being moved around. I replaced it with an equivalent aluminum plate.

Early on Rapman users began to design replacement parts for the bits of the Rapman laser cut acrylic that fell apart. You can see a corner block designed in white ABS by Chylld in the picture above.

You notice that I only put ONE of Chylld's excellently designed, fast printing corner blocks onto my Reprap machine. Once I got it on it has performed beautifully. Unfortunately, I had to half way disassemble my Rapman to get the damned thing installed. That wasn't an experience that I thought worth repeating.

Late in 2010 I decided that I wanted to experiment with dual extruders. Bits from Bytes had come out with one some months before and I decided that I would get one. When I called, they were so focussed on their new 3000 printer that they wanted me to switch over to that and did a hard sell. I never react positively to hard sells, so I put off the purchase. As well, the dual head Rapman had the same dimensions as the old one which meant that it had a reduced print table area as a result.

After the beginning of the new year Bits from Bytes was acquired by 3D Systems. I attended the telephone meeting announcing the takeover and was so put off by the 3D Systems CEO, Abe Reichental, that I abandoned any plans to purchase any more Bits from Bytes equipment at all.

That left me with quite a dilemma. I had never been particularly happy with the Mendel design. Tommelise had used very similar kinematics. The problem with it that it requires a larger footprint than the print table and is unstable in the x axis direction. Mendel addresses this last issue with a electronics mounting board that doubles as a reinforcing plate to cure the x-axis problem. Because of that, I decided to stay with the Darwin kinematics and run the system through another kaizen exercise.

Basically, I want to look into several possible improvements to my Rapman

a return to printed parts. My experience has shown that laser cut acrylic does not make for a robust machine.
a massive reduction in the variety, as opposed to the number, of parts needed to construct the machine.
parts design which allows for easy partial demounting of the printer.
a design robust enough to be easily shipped fully assembled or quickly broken down and reassembled for exhibitions
room for two extruders without compromising the size of the print table.
cleaner y and z axis kinematics
a more capable controller board
shifting to the successful Wade/Adrian pinch wheel extruder

My experience with trying to install the Chylld corner on my Rapman has caused me to want to rethink the whole issue of parts design for a Reprap printer. I feel that we have, in a way, gone wrong when we designed the parts for Darwin and Mendel. Because you can custom tailor individual parts with a 3D printer doesn't, to me, mean that it is good design practice. Right now, to have a proper set of spares for a Darwin or Mendel you pretty much have to print a full parts set. A lot of that stems from, in my opinion, from custom tailoring parts.

I don't think too many Reprappers are going to bite the bullet like Adrian Bowyer and Nophead {Chris Palmer} have and print whole sets of parts day after day and week after week. Most of us want Reprap machines so that we can design and print other things beside new Reprap machines. To that end, having a limited set of parts types that you can print a few at a time during down time when you are doing design, seems to me to be the way most of us are going to propagate more printers.

With respect to a new y and z axis design I want to revisit the cabling approach that eD did with the Darwin precursor ARNIE back in 2006.

But, enough talk. Here is what I have so far.

Sampo's frame is built on a 600 mm module to give it that extra space for dual extruder print frame that Rapman's 400 mm does not. So far, the entire frame has been assembled with exactly five different parts.

Here you can see the Sampo frame beyond my Rapman 3.0 for scale. I am working on the cable housings for the z-axis as this blog entry goes to press.

Buying parts

2011-03-17T07:30:00.000-07:00

This morning, I bought the linear shafting; 16x2ft 8 mm diameter segments and the linear bearings; 24x8 mm linear bearings.

The first test NEMA 23 stepper and the 24 volt power supply should be arriving today. I plan on using Bogdan's mount design for the linear bearings and an aluminum plate for mounting the extruders. Corner connectors will owe a lot to Chylld's corner designs for the Rapman, though I will not be slavishly following his approach.

I hope to be testing the firmware routines for controlling the steppers this weekend.

Eat your heart out! :-D

2011-03-11T22:05:00.000-08:00

Fired up the development board and it fell into a roundabout of pics.

Not bad for a little 320x240 display.

Development board arrived

2011-03-10T20:00:00.000-08:00

My PIC32 development board arrived today!

I also got a trial stepper motor controller with it.

The development board looks as good in my hands as it did in the sales literature.

My old PC power supply even had a power connector that fit with the board already installed. It is a relict of a battery charger for an old Sony digital camera that I finally retired last year.

I am also wanting to adopt Rapman's use of milled linear guides and linear bearings. If you look hard these are not all that expensive. For bearings, I found this little unit...

To be well within budget. Guide shafts are a little pricier. In the US it appears that 12 mm shafting is the least expensive while still being stiff enough for a printer. Most pricing seems to be running around $0.50/inch. That's not cheap, but not impossible.

Being a Scots-Irish tightwad, however, I kept looking for a better deal and ran across this...

...which comes in at a bit under $0.07/inch. I spoke with the vendor today and they are checking to make sure that it meets CNC requirements for a linear guide. The general impression was that it did, but I want to be sure.

If that price does follow, it should be possible to replace the effective, but overcomplicated z-axis arrangement of Rapman which derives from the old Darwin design...

...which has 8 mm guide rod and a plastic bushing with a 12 mm guide rod and a proper linear bearing. That extra rigidity might allow me to control the position of the print table with an single, proper lead screw attached to a cantilever beam which contacts the bottom of the print table in the center instead of three or four lead screws, or cheap studding {threaded rods}, more likely.

Certainly the leveraging the guide rods as vertical structural members should simplify the design a bit.

Going for 32 bit embedded processors

2011-03-01T10:06:00.000-08:00

In which your narrator begins to cobble a next-generation Reprap controller board together out of much bigger vitamins than has previously been possible.

The controller for the Bits from Bytes Rapman has long been the most sophisticated in the world of Reprap printers. It's use of the PIC32 MCU gives it the power to do things, like support an LCD display, SD cards and 0.1 mm gcode steps without difficulty are very difficult to do with 8 bit MCUs. There are two barriers to it's widespread adoption in Reprap printers, however.

The first is that PIC32s typically come as 100 pin surface mount chips. That means that you have to do some fairly sophisticated PCB design and get the damned chip soldered on properly. That's not impossible, but it's not easy, either.

The big barrier, however, has been the lack of inexpensive compilers for PIC32s. The one from Microchip is very good, but costs over $1K. Recently, that situation has been remedied by a Serbian firm, Mikroelektronika. I had had experience with their PIC 18F compiler and found it both very reliable and possessed of an enormous library of library functions which made it extremely valuable as a development tool.

Thus, when I heard last year that Mikroelektronika had undertaken to create an inexpensive compiler for PIC32 chips I was immediately interested. After a long development and beta cycle they did their formal release this morning. They offer inexpensive C, Basic and Pascal compilers for the PIC32.

Some time ago, I resolved to build up my next printer rather than buy another BfB product. The ultra-reliable BfB extruder and controller board were two technologies that I intended to bring across. I wanted, however, to have two extruders and a somewhat bigger print table to accommodate them better than Rapman allows.

Laszlo created an inexpensive, easier to repair hot end than BfB offers which I also intend to use in the design.

As well, recent purchase of a hobby lathe ...

... will let me experiment with large diameter precision lead screws which can double as structural elements should cut down on the amount of steel in the printer rather dramatically.

The big advantage to the new Mikroelektronika compilers, however, is that they also offer an inexpensive line of prototyping boards. That will save me the development time of putting together a new board with technology that I am not familiar with. I ordered their most expensive board {$169}...

which includes everything that I need save stepper drivers. I also ordered one of their stepper driver boards. to test with it to test my firmware with ...

before I integrate the heavier amperage Pololu driver board that has been used in a number of other Reprap controller boards.

The current situation is so much better than we faced when we began the Reprap project in 2005. Then you were pretty much doomed to building up boards from scratch. These days you can buy inexpensive boards and add-on miniboards to build up an inexpensive yet very sophisticated Reprap controller board quite easily.

Should my controller design catch on, I suspect that other Reprappers would want to use this much smaller and less expensive graphics board in Mikroelektronika's stable ...

... which has pretty much everything the larger board does.

Printing small holes in small features

2011-02-14T09:45:00.000-08:00

In which your narrator discovers an out-of-the-box method of making small holes for pinned hinges in small features.

Recently, Chris Palmer blogged an exquisite article on the pitfalls of getting the diameters right in holes made in objects. It related very closely to a problem I was having in developing a printed robot hand.

Previously, I'd had no trouble in that the joints between the phalanges of the robotic fingers were very large and printed.

I loved this approach, but sadly had to abandon it simply because the large contact areas between the phalanges created excessive friction and because the number of things that I needed to have happening in a phalang made the large joints unhelpful. The need for smaller, more compact hinged joints brought to mind the work of Frank Davies with his Sarrus Linkage positioning system.

Frank used pinned hinges to great effect.

As I was designing such pinned hinges into my phalanges, I soon discovered that the substantial pins that Frank was able to use were too large for the delicate phalanges that I was working on. In fact, I finally settled on using simple paper clips (0.84 mm diameter) for the pins in my work.

In designing the joints, I followed the usual Reprap gambit and simply included the pin holes in the STLs for the parts.

The only problem with this approach was that both the hole and the hinge joint that it was seated in were very small. The joint had a radius of only 3 mm and the hole 0.42 mm.

Ordinarily Reprappers use a few print paths around the perimeter and then infill the rest. I design using thin walled parts glued together after printing, an approach that lets me create fine featured parts that are fast to print. With a feature this small, however, print roads radiating out from the pin hole very quickly clash with print roads radiating in from the joint.

It's bad design practice to let a hot print head hover for extended periods of time over or near a small feature. You want the head doing the feature then going far away quickly so that the molten plastic thread has a chance to cool a bit before the next layer is applied. If you don't get this your small feature becomes a featureless blob.

Print road clashing between the pin hole and outer perimeter of the hinge had me fiddling around with print road width for the better part of a week with indifferent results. Yesterday, however, an out-of-the-box solution to the problem finally hit me. I'd do better at printing the pin hole if I simply didn't include it in the STL, something like this.

Basically, I just plugged the hole. What that did is to limit the print road propagation to those radiating inwards from the outside perimeter of the hinge. Since I don't use infill because of the problems getting the perimeter roads to match with the infill roads, for the first few millimeters of the print had no hole due to the use of perimeter roads to completely fill the layer.

Once I had the roads calculated I pulled up the .PNG images for the completely filled layers. You can see the bit that will fill the pin hole circled in red.

It was a simple matter to pull the images into Paint and remove the inconvenient loop from the images and then continue processing the images into Gcode.

By calculating the width of the print roads appropriately, I was able to get the proper diameter for the pin correct on the first try.

The annoying part of this whole epiphany was that it took me over a week and dozens of trial prints to see the simple way of solving the problem.

Restocked

2011-01-11T06:58:00.000-08:00

Many months ago, I ordered new filament for my stocks. I order exclusively from New Image Plastics in that I know that Jim Waring there charges low rates and has consistently high quality. What with the economic troubles Jim has had trouble in recent years finding a broker willing to deal with small orders of less common polymers like ABS. He found such a broker recently, though, and was able to fill my order. :-)

Yesterday, my filament arrived; thirty lbs of 3 mm ABS at $7.95/lb and 10 lbs of polypropylene at $4.25/lb. With the ABS I already have, I think I am now set for the year.

A proper conductive polymer mix?

2010-12-24T10:28:00.000-08:00

In which your narrator begins to test a new conductive polymer mix.

Do you want to read more?

One of the long term grails of the reprap project is to be able to print circuitry. Currently, Rhys Jones, a PhD candidate working for Adrian Bowyer at the University of Bath, is devoting a lot of his time to addressing this issue.

Heretofore, we've looked at very low temperature eutectic alloys like Rose's and Wood's metal and confected methods of extruding it into preprinted channels. More recently, Rhys has demonstrated that ordinary solder can be successfully printed onto ABS.

We've looked at various concoctions of carbon, silver and other conductive materials with polymers at arm's length. Always, though, the conductivity of these mixes has been too low for them to be useful in printed circuits. Rhys, however, has looked at a two step process which may get us by that.

The main problem with printing PCBs with a reprap machine is that PCBs are not that hard or expensive to have made using conventional, proved materials. This makes coming up with a method of printing them a bit tricky because it not only has to work, but it also has to yield a product that is not all that much difficult to use than conventional boards.

Recently, I ran across a high-tech company, Integral Technologies, located in Bellingham, Washington. They've come up with a different approach to creating a conductive polymer mix which uses conductive fibers mixed with conventional polymers. Probably the most exciting is their mix of very fine stainless steel fibres mixed with a blend of polycarbonate and ABS.

The volume resistivity for this material on the data sheet runs to 0.06 Ohm-meter. This is far too high for a useful conductive plastic. I rather shrugged the material off until I got a look at a few non-technical demonstrations of the material. This one shows two small bars of the material used to conduct 110 v current running an electric drill.

Interestingly, Bill quoted me a price range of between $9-45/lb depending on the composition. More interestingly, the PC/ABS mixes which resemble what we are already used to using in Repraps are at the cheap end of the price spectrum.

As well, they demonstrate measured resistance across an extruded plate of their material using an analog multimeter.

There was obviously a mismatch between what their data sheets said and what the non-tech demos indicated. I explored this question with Bill Robinson at Integral. It boils down to the "skinning effect" mentioned in the first clip. Simply explained, it means that the metal fiber doesn't tend to show up at the surface of an extrusion. This means that the material is partially insulated at the surface of an extruded surface.

That left the issue of how, in the second demo clip, ordinary multimeter probes registered effectively zero resistance when measuring resistance across a extruded plate of their plastic.

Bill sent me some demo plates of the material a few days ago. I have a digital multimeter rather than the analog Radio Shack multimeter.

Multimeters are not particularly good at accurately measuring low resistances. Anything reading below about 5 ohms for most multimeters is not something you want to bet your life on.

I tried duplicating the measurements in the clip and kept running into skinning effects. In places I could get readings less than 1 ohm, but others I would be in the mega ohm range. I made a cut across one end of one of the samples. You could barely see the fiber ends. Integral is using VERY fine fibres.

I then drilled 1.5 mm holes through the plaques and put the probes through those. I was able the get more consistent and lower resistance measurements. Mechanical contact between the probes and the fibre ends is how good conductivity is achieved.

The best I was able to get consistently was 0.6-0.8 ohms over about 100 mm of the material.

Bill at Integral suggested that if we used a 0.5 - 1.0 mm extrusion nozzle we out to be able to avoid jamming of our extrusion nozzle with the stainless steel fibres. Making strip extrusion as we can with a Reprap printer should tend to align the fibres much better than a simple injection moulding would do. The problem as I see it will be making a conductive connection between two traces and making a connection between a trace and a component.

The question now is whether what we know about this material now justifies having a sample turned into fibre for further testing.

Getting into threading...

2010-12-19T14:16:00.000-08:00

When I bought my Sherline lathe, I went for the loaded option. As I am finding out, it was a very good decision. I don't think I'd have been able to use at all it without the digital readout box. I understand how people in times pass could have used a micrometer to do measurements between cuts. I doubt that I could have had the patience, though.

After my last attempt at doing a deep drilling and cutting down a piece of aluminum bar stock, the next thing I wanted to do was use the threading attachment. Between Joe Martin's Sherline book, Tabletop machining and the Sherline Accessories Shop Guide I was able to puzzle out how to re-rig the lathe for threading. My motive in getting a threading attachment was to have the possibility to cut my own lead screws and thrust collars. The studding used in the Rapman and Mendel 3D printers, not to put too fine a point on it, suck the middle out of canned Vienna sausages. When you buy studding it is rarely, if ever, straight. When you try using it as a z-axis lead screw, it does not very pleasant things to your print quality.

As well, I am quite fond of the polymer pump design on the Rapman. My primary reason for wanting to do my own threads was to be able to make the threaded drive shafts for those, too.

The biggest shock for me in this adventure happened when the instructions book told me to demount the lathe motor. It turns out that you cut threads on the Sherline lathe with a hand-powered handle. Using it reminds me of my grandmother's old foot pedal powered Singer sewing machine that I learned to use when I was about four. It has a very similar, 19th century feel to it.

By and large the threading system design is quite good. The real pain in the tail, however, is that little black lever that you see below the spindle in the above picture. Getting that to work right had me taking the whole lathe apart about three times. Sherline needs to rethink that part of the design. It really sucks.

Other than that, the threading attachment was quite easy to use. It is set up to do standard and metric 60 degree threads. The carbide cutting tool did that quite easily.

For a first try I did a really coarse metric thread, 2 mm.

I figured that that deep a thread would require the most muscle. It was a bit tiring but not at all impossible.

For lead screws, the Sherline will handle about 15 inches of threading. If the diameter of the stock is less that about 9 mm, a longer unthreaded piece of stock can be snaked through the spindle. Given that manufactured lead screws about cost $1/cm and thrust collars typically cost $30-35, this capability will save me some money over the mid-term.

First lathe experiments

2010-12-13T12:11:00.000-08:00

I finally got a few days off, so this morning I began to do things with my new lathe.

My initial motive in getting the lathe was to have the capacity to make my own extruder hot ends. Given my total lack of experience with lathes, I decided to see if I could try sinking an 1/8 bore in a piece of aluminum round stock using a fixed drill bit in a chuck inserted into the tailstock. An idea of what I did can be seen here.

After a considerable time, I was able to sink a 50 mm hole into the centre of the round stock. The steady rest that you see supporting the end of the aluminum round stock nearest the drill bit reduces vibration and misalignment.

When I worked with Tommelise 1.0, I used a 5/32 inch braised copper tube (1/8 inch ID) hot end several inches long with its top only heated using nichrome wire. I was able to get well past the whole issue of having filament melt in the PTFE thermal break with that approach. These days, I hope to do something similar using aluminum.

Here you see the drilled round stock after I've trimmed it down to 12 mm outside diameter and trued up the end. It takes a nice finish.

Arrived

2010-11-20T17:06:00.000-08:00

The Sherline lathe arrived on Thursday. I was working that night but stayed up extra late and more or less got it assembled. It went together without a lot of drama. The digital readout box makes whether you bought the lathe with American or SI verniers irrelevant since you can select metric or "Imperial" on the digital readout box.

The readout box also includes an optical tachometer to keep track of the rpm rate of the lathe. The DC controller knob will give you stable rotational rates down to about 200-250 rpm.

This morning, I mounted the lathe on a piece of chipboard for stability.

I bought a couple of pieces of aluminum round bar, a piece of stainless round bar and 3 feet of hot rolled steel round bar to practice on. You can see one of the aluminum round bars mounted in the lathe.

Taking a different direction

2010-11-16T20:54:00.000-08:00

A few months ago, I had planned on acquiring another BfB Rapman with two print heads this time so that I could explore the use of PLA as a support material for ABS prints.

With the sale of BfB to 3D Systems, continuing troubles with BfB firmware and the catastrophic failure of my expensive Rapman hot end and the discovery that BfB isn't warranting such failures, I've decided to build my own second printer.

To that end, I've just purchased a Sherline 4410 metric lathe.

It will be arriving Thursday.

Annoyances and changes

2010-10-28T21:25:00.000-07:00

UPDATE: I let my anger run away with me in this blog post. I've just spent time with Ulf Lindhe at Netfabb. Ulf has sorted out the problem I was having with the boolean operations and got me upgraded to version 4.6 at no charge.

It's taken a few days, but I can now print at about the same quality with a 0.5 mm extruder as I was able to print with my now ruined 0.3 mm extruder. I've gone back to work on the active telepresence project and the present task is to create linear servos to as the muscles to operate the hand and fingers. I'm building my own from gearmotors rather than buying after some bad experiences trying to adapt servos designed to operate the flaps and rudders on model airplanes and sailboats to operate hands. Andreas Maryanto was good at this, but I am definitely not.

Doing this, I am leveraging the herringbone rack and pinion scripts that I wrote. While Nophead's advice to simply not fiddle with my 0.3 mm hot end settings got me 95% of the way to where I wanted to be, that last 5% took a little tweaking. I'm quite happy with the final results.

The pinions peeled off of the rafts with a minimum of trouble and only took a touch on the belt sander before they were usable.

That was relatively easy. When I started working on the rack, however, I quickly ran up against the limits of Art of Illusion. While it is easy to assemble a set of solids to represent what I want to print in AoI...

Doing the Boolean opeations to make them into one solid is quite beyond AoI's boolean ops capabilities. Bogdan told me repeatedly that I could do this sort of thing in Netfabb Professional, so I read up on the method and set about to use my now copy and discovered that the Netfabb people had somehow disabled this feature in my copy. ~~I had a email from them about version 4.6 {I have 4.4} but no clear idea if I was going to get a free upgrade or have to pay them even more to do something that I am already supposed to be able to do.~~

The more I think about that the more pissed off I get. If I don't hear from them in the morning, I am going to build boolean ops into Slice and Dice and give it away for free. The way I feel just now I might just do that anyway. I am getting sick and tired of getting ripped off.

Getting to the bottom of the Rapman hot end failure

2010-10-20T20:31:00.000-07:00

In which your narrator gets to the bottom of why it was that his 0.3 mm Rapman hot end failed and perhaps even why it started causing resets prior to the failure.

The hot end failure that I experienced on 14 October is a relatively rare event. The only other person with a Rapman who reports the sort of ABS leakage that I experienced was Klazlo.

Once I had my Rapman repaired and usefully printing again, I turned my attention to doing forensics on the failed hot end. Once I took a hard look at the hot end and the pictures that I took immediately after the failure hints of the means and mode of the failure began to recommend themselves.

Taking a look at one of the forensic pics that I took on the 14th two items jump out at you. I've circled them in red.

If you look closely at the top red ellipse you will notice that one of the three bolts which secures the spacer on the hot end ensemble was either not tightened adequately at the factory or, more likely, worked its way loose via hundreds of heating and cooling cycles while in operation. This threw the alignment of the hot end out opening a route for hot ABS to escape between the PEEK sleeve and the aluminum extruder barrel sleeve that fits around it. You can clearly see the billowing flow of ABS in the picture. ABS was only leaking on the side of the PEEK barrel where the spacer bolt was loose. That part is straightforward.

The vertical ellipse on the right hand side of the hot end is much more ominous. If you look closely, you can see that one of the thermistor leads has exposed above the sleeve leading down to the extruder tip where the extruder is housed. A bit over 4 mm of exposed thermistor lead is open to the air. You can see that happening a bit more clearly with this picture.

The larger red circle encloses the exposed lead. The smaller one shows the recession of the purple plastic insulation on the lead that occurred once the ABS began to boil out of the gap between the PEEK and the aluminum and contacted the lead.

At first I thought that the hot ABS had actually caused the gap in the insulation. Closer inspection, however, revealed that the lead had most likely been stripped too liberally and originally was only just barely covered by the yellow silicone sleeve. The heat of the billowing ABS almost certainly caused the shrinking of the insulation from just being covered by the silicone sleeve to its present situation. It's easy to see the shrinkage in the second circle.

I then removed the PEEK cylinder from the aluminum extruder head. Originally, I had thought that the PEEK sleeve in contact with the aluminum might have shrunk. It measured 10.09 mm outside of the aluminum sleeve and 9.9 mm inside. That could have been either because of heat shrinkage of the PEEK over time or some sanding of the PEEK tube at the factor to get a clean fit. Unless the parts had been measured before the hot end was put into service, it would be impossible to tell for certain.

I then removed the silicone tube from the inside of the PEEK sleeve as you can see here.

This is the most revealing forensic evidence in the whole exercise. At first, it appears that the silicone tube has shrunk on the hot end. While that is true to an extent, the reason for the shrinkage is actually because the inside diameter of the PEEK sleeve has shrunk because the PEEK has expanded within the confines of the aluminum sleeve.

It was fortunate that I bought several unassembled hot end kits before BfB began making premade ones because I had some disassembled PEEK and silicone tubes. The unused PEEK sleeves had an outside diameter of 10.0-10.1 mm and a consistent inside diameter of 6.36-6.40 mm.

When I measured the failed PEEK sleeve the cold end inside diameter was 6.40 mm but the hot end had been reduced to 5.88-5.92 mm. The silicone tube outside diameter on the hot end was more or less the same as the inside diameter of the PEEK sleeve.

What was revealing, though, was that the hot end of the silicone tube was virtually entirely closed. It was impossible to slide a 2.9 mm ABS filament through it. I was able to blow tiny bubbles through the silicone tube into a glass of water. What that says is that the ABS filament was being melted in the silicone tube and squirted into the aluminum extruder barrel rather than being melted in the extruder barrel.

I don't think that that was how the hot end design was intended to work.

Getting back to that exposed thermistor lead, it is well known that blowing hot air past plastic surfaces will create a static charge. Not only was that exposed lead within millimeters of billowing hot ABS it was also touching the bottom acrylic plate of the x-axis carriage {part 10034}. Periodic static discharges off of this heated acrylic surface were most likely at least deranging the temperature measurements off of the thermistor and also causing the MCU board to reset. During last winter, I walked in socks across the floor and picked up sufficient charge to where when I touched the MCU board the system reset.

Conclusions:

The failure of the hot end had two aspects.

The loosening of the spacer bolt, most likely from repeated heating and cooling cycles of the hot end allowed for a hot end misalignment that created a path for leakage of hot ABS out one side of the hot end between the aluminum extruder sleeve and the PEEK sleeve for the silicone tube.
The swelling of the PEEK sleeve over time reduced it's inside diameter almost entirely closing the silicone tube and causing the ABS to have to melt in the silicone tube rather than in the aluminum hot end. This was certainly not the intent of the hot end designers.

It seems quite obvious that the combination of a PEEK sleeve enclosing a silicone sleeve for the plastic filament need to be rethought.

The placing of an MDF plate and an acrylic plate at the top of the hot end spacers through which the securing bolt has to pass also needs some careful. Replacing the MDF with a PEEK plate with the bolts secured by lock washers to the PEEK plate and recessed through the acrylic with the acrylic and PEEK secured by a separate trio of short bolts might be more successful. An aluminum plate as a replacement for the MDF might also be considered.

Chylld is a parts design genius

2010-10-19T15:28:00.000-07:00

As I mentioned earlier I replaced to destroyed 0.3 mm hot end with one of my spare 0.5 mm hot ends and began to get used to using the larger nozzle size. After some frustrating tries at adjusting flow rates Nophead suggested that I stop fiddling with things and just use the settings that I'd used before. He was right, of course.

A quick print of some of the bones of the hand yielded good results, so I decided to have a try at one of Chylld's corner replacements for Rapman cut acrylic.

So far there's been no sign of the wall-to-wall resets I was getting with my Rapman in the hours before the 0.3 mm hot end finally failed.

The corner block printed in 2:15 and mounted with a minimum of drama.

There was a minimum of post print part processing required. the x-axis bar was a little tight, but a brushing out of the hole cured that. All bars had a tight, friction fit and the tightening screws were largely cosmetic except when the printer wanted moving.

Chylld is a parts design genius. I've never seen anything work quite this easily.

Operational again at 0.5 mm.

2010-10-17T21:21:00.001-07:00

Okay, I got one of the spare 0.5 mm extruder heads mounted. I measured the output and discovered that I was getting swell to about 0.66 mm. That works out to 0.34 mm^2 cross sectional area.

When I was working with the 0.3 mm extruder head I was getting a swell to about 0.42 mm which works out to about 0.14 mm^2 or about 40% as much.

I tried printing a test raft using the old extruder set points and noticed that the base print roads for the raft, which had been well separated before, are overlapping now. In the morning I'll lop 60% off of the extruder's stepper speed and see what happens.

Repairing a catastrophic failure of the Rapman 3.0

2010-10-17T14:34:00.000-07:00

Little did I know when I was upgrading Slice and Dice to do prints of laser-cut Rapman parts that I would soon be confronting the problem of replacing such parts without having a 3D printer to print the parts. That is an ENTIRELY different problem.

I had upgraded Slice and Dice to do a better job of printing my sample laser-cut part from last time and was doing trial prints when I noticed that I was getting an enormous number of resets. This was odd because ordinarily I only get one or two resets per month since I uploaded firmware version 4.0.2. I checked the humidity and it was a bit dry, so I fired up the hot mist humidifier and drove the relative humidity up to 52%. No joy.

The next morning I undertook a new print and noticed a mushroom of ABS had formed under the extruder.

When I attempted to remove the extruder, the mushroom of ABS proved to be too big to easily get out of the mounting hole. I then began to disassemble the x-axis carriage that holds the extruder only to discover that the top plate was being held together by the grace of God and nothing else. It crumbled into two major pieces and half a hundred small fragments when I began to remove the bolts.

As well, one of the tiny little bars that hold the x-axis belt had also broken in half as is readily visible in the previous picture.

As you might imagine, this all made me very cranky. A quick calculation revealed that I had historically been spending considerably more on preassembled BfB hot ends that I was on the filament that goes through them. I run a lot of filament, too.

I spent several hours chatting with both Iain and Andy at BfB. They were attempting to be as helpful as they could, but somehow I came away with the same feeling that I get when I take my car into the garage and they say recursively, "It MIGHT be X. We could work on that and see what happens." I began to steam up a bit when they started talking about my not using "BfB gcode". They'd latched on to that fact that I'd written my own STL processing software that produces gcode right out of their manual and tagging this as a possible problem. They then started talking about the fact that I was using very short line segments {0.1-0.1414 mm} and that was possibly touching off heretofore unsuspected bugs in the firmware. I pointed out that the Rapman is supposed to have 0.1 mm resolution and you can't print objects at that level of resolution unless the firmware can handle that short a line segment.

I finally decided that I had to stop talking before I said things I'd regret later {I do that a lot}, and have a good think.

Here's what it came down to...

I needed a new top plate for the x-carriage and a new hot end.
I was going to have to pay for these and it was going to take a week to get them
It was entirely possible that the hot end had been destroyed by a firmware bug
There was nothing to say that the new top plate would last any longer than the last one, viz, ten months.

No matter how I looked at that equation it just didn't seem to balance. A big bone in my throat was the fact that BfB wanted me to pay to replace a hot end that it was very possible that their firmware had broken. A bigger bone was that there was no guarantee that if I put the new hot end that I bought into the system that the firmware wouldn't ruin it, too, in short order.

It seemed to me that the most reasonable course was to see if the problem was with the firmware. I typically print with a 0.3 mm hot end. I like what that kind of resolution does for my prints. Before I settled on 0.3 mm, however, I bought two, preassembled 0.5 mm hot ends, so I had those in stock.

I also had Bogdan's experience that replacing the top plate on the x-carriage and grounding the hot end to it would stop the resets. This from the observation that resets were most often caused by a static charge building up on the plastic of the x-axis carriage and extruder and then discharging, causing a reset. BfB which is apparently located in a damp environment had never encountered this issue. I noted that resets tended to get quite common when the relative humidity in the room holding the printer dropped below about 42%. I'd sorted out resets, except for the ones I encountered most recently which have led to the hot end failure, I think, by using a hot vapour humidifier. After seeing how the top plate had crumbled, however, the notion of replacing the acrylic one with an aluminum one began to sound very attractive.

I decided to acquire the means to cut both acrylic and aluminum. Harbour Freight in Salinas had a very nice scroll saw on sale for $69 which would reputedly do the trick.

I bought that, a sheet of 0.22 inch (5 mm) acrylic and a billet of 3.35 mm aluminum plate. I decided to cut an aluminum top plate first. I began the process by simply tracing the bottom plate, which was identical to the top plate onto the aluminum with a fine tip marker.

I did the rough cut with the scroll saw and dressed it with a grinding wheel and a half round ring file after having set the plate in a small vise.

I then remarked two holes at diagonal corners of the plate, drilled 1/16th inch guide holes and widened them to 13/64th inch {as close as I could get to the 5.2 mm holes as I measured them on the original piece. I did this with a hand drill after securing the plate in a vise.

This done I then bolted the acrylic bottom plate to the developing aluminum one and drilled the guide holes for the rest of the holes using my Dremel drill press.

I then removed the acrylic bottom plate, secured the aluminum plate in the vise again and drilled out the rest of the holes.

Afterwards I cleaned up the finished plate with a wire wheel and checked it for fit on the x-axis.

At this point my dyslexia set in. The plate is not symmetrical and I'd got it flipped and completely reassembled and tested the carriage this way. I didn't notice the problem till I tried to fit the extruder into the top plate assembly and discovered the symmetry problem. The next pic is from the original, incorrect assembly.

The new top plate works smoothly. Now I've got to swap out my ruined, 0.3 mm extruder with one of my spare 0.5 mm ones. I will be able to see if I have a serious firmware problem or whether I just have a design fault with the hot end. It could be either, or both.

I'm entering a big of a crisis with respect to 3D printing. I bought into BfB's Rapman because I wanted to do some printing instead of screwing around with printer design and problems all the time. At the time, a year ago, it was a good move. Rapman was a bit pricy, but it was solid and the components of it were affordable. The 32bit MCU board was a delight after all of the Linux/Arduino/Sanguino/Bullshitino nonsense. Some months ago there was talk of extending the Rapman MCU to where you could parameterise the firmware setpoints to deal with different machines and extruders, like the Mendel, for example, or even machines you'd designed yourself. As it stands, it's not clear that BfB can design reliable firmware for their own machine, much less a parametric firmware app that would make it applicable to a wider range of machines.

On top of that, they've recently jumped the price rather dramatically.

As it stands, BfB's Rapman has two Achille's heels; their firmware and their hot end. Neither are reliable and the hot end is very difficult to repair. I'm told that BfB is working on successors to the hot end, but that does me no good at all. I'd like to shift over to something like Nophead's power resistor driven hot end. The problem with that, however, is that I'll have to design a MCU to drive it and the printer both. By the time I've done that, BfB is out of the picture, since the those two components are what is defensible as corporate worth in the BfB.

I don't know quite what to do.

Replicating laser-cut parts

2010-10-10T18:34:00.000-07:00

In which your narrator confects his own idiosyncratic method of replicating laser cut parts.

The recent acquisition of the UK manufacturer of the reliable Rapman printer, BitsfromBytes {BfB}, by the American firm 3D Systems has exacerbated a long standing problem that the Rapman's users' community has had, vis, getting drawings of the laser cut acrylic parts that make up the system. While BfB uploaded drawing files of one of their early Rapman designs into the Reprap website, they have neglected to do so with version 3.0 and later models to the best of my knowledge.

That wouldn't be such a problem save for the fact that the acrylic parts in higher stress parts of the Rapman tend to develop shear cracks and spalls after hundreds of hours of operation. Rapman owners are then left to either request replacement parts from BfB or print their own spares. Interestingly, there is a healthy spares design effort underway for the Rapman, part of which is hosted on the BfB website itself.

Here the printed white ABS grips for the x-axis linear bearings replace the lower plate for the extruder carriage in a much stronger configuration that the original. Indeed, at this moment, there is a very exciting redesign of its printable parts underway by an Australian newcomer that greatly reduces the print time required for what were finicky corner blocks.

All this aside, there are many pieces in a laser-cut printer that simply need to be printed as-is in ABS rather than redesigned. Heretofore, I found myself carefully tracing the parts out on paper and using calipers to get the dimensions. While that approach works well with many parts some, like the extruder z-depth stop plate are much more problematical.

This particular plate is the antithesis of rectilinear and locating the holes and pockets is tedious to put it mildly. I'd thought that it ought to be possible to use an ordinary 2D scanner to capture this sort of information. Today, I gave it a try and discovered that it was not as straightforward an operation as I'd imagined.

I had an extra copy of this part that I'd bought still in the protective blue film, so I threw it in my Epson Perfection V500 Photo scanner, a very cheap and extremely high resolution machine. Even with the blue film the image captured didn't have a lot of colour information that would let you think that you could separate the part from its background.

I pulled that one into Slice and Dice and tried to separate out the image with RGB manipulations to no avail. I then tried putting a intense red background behind the piece figuring that the blue would override the red and tried playing with the colour mixes in both Slice and Dice and Photoshop, again to no avail.

Two things were killing me; the transparency of the part and it's depth, which you can see very clearly in this scan. When I tried to just capture the edges by going high contrast, the depth of the object spoiled everything.

It occurred to me that I could paint the part on one side with tempra paint, which can be wiped off of slick surfaces. Rather than drive into town and given that the part was already protected with a blue film I simply spray painted it with red enamel. I then put it back in the scanner with a piece of dead black HDPE sheet behind it and instantly got the colour separation I needed.

Popping the scan into Slice and Dice and filling the black with green, I had an image I could work with. As you can see, it was a little nasty, largely due to the messy laser cutting on some features and a bit of flare here and there.

A few moments in Paint cleared that up.

Now that I had crisp colour separation, defining the part boundary in Slice and Dice was trivial.

While I was in paint I took the pixel counts across the diameter of the outer circular feature boundary and measured the same feature on the part with calipers. I then adjusted the size of the image in Paint.

With a bit of pushing and shoving in Slice and Dice, I processed the part and created a print file. Here you can see the print roads for the part.

With a print file, I did a trial print to check the dimensions. The acrylic part lay precisely over the printed part.

I photographed it again with the original part slightly ajar so that you can see the holes in the print.

If you look closely at the several layers of part print that I ran before aborting it you can see that I need to increase the print flow a touch. More importantly, with a part of this size and complexity, however, I am going to have to write a routine that makes a better job of reducing transition distance between print roads. Transitioning was taking far too much of the machine time. That's on my "to do" list now.

What I've demonstrated here is not a smooth operation on Slice and Dice just yet. I was mostly trying to prove the concept. That was a huge success. What this means is that owners of laser cut reprap machines can readily exchange parts information with nothing more complicated than an ordinary 2D scanner and a set of calipers. This should give us considerably more flexibility than we have at the moment.

I am certain as I finish this blog entry that someone is going to show me a simpler, faster way to do this in a few hours. That's certainly happened before. If not, though, we have this approach. :-D

When ego takes charge

2010-09-25T14:47:00.000-07:00

Nordom recently printed a brilliantly executed M30x70 bolt, nut and washer ensemble.

I was, of course, quite jealous of his accomplishment and still am. It is just a beautiful thing.

That got me to wondering, though, just how small a bolt one could print? As usual, I was too impatient to wait till Nordom got permission to distribute the STLs for his bolt and finally found something similar in Thingiverse.

I hate recessed head bolts so I replaced it with a regular hex head and scaled it down to a M10x20. After several tries I had something useful.

I've put it beside a similar metric bolt on the Rapman for scale.

Although the metric scale threads work in ABS, it seems obvious to me that something cut a bit deeper would be more useful.

Now that I've got that out of my system, I'll be going back to working on my carpal assembly.

An interesting stringing behaviour

2010-09-19T08:08:00.000-07:00

The ability to reverse the extruder on my Rapman printer during transitions between print roads has reduced stringing to an enormous extent. What stringing I do get tends to be very thin and feathery.

I've noticed an interesting stringing behaviour since the release of firmware version 4.0.2, though. You can see it here.

What you will get is a thin string between two objects being printed being propagated and then the string acting as a brush on th extruder orifice at intervals between roads. These brushed accretions build up into quite beautiful forms resembling frost.

When you get just a tiny bit of string hanging off of your print you will see the brushed accretion building up at 45 degree angles into something resembling fractal patterns.

This phenomena isn't a big issue for me. The accretions are thin and fragile and brush off easily without sandpaper.

They are pretty, though, in my opinion at least. :-)

Dealing with detaching rafts

2010-09-16T19:50:00.000-07:00

In which your narrator seems to have come up with a way to prevent raft peel for ABS on an acrylic print table.

About two weeks ago, the rafts for my ABS prints started detaching from the right hand side of my print table. I was losing one out of two to one out of three prints that way. At first I decided that my acrylic print table had simply got too warped and I had too much variation in level on the acrylic. I removed the acrylic print table and checked its flatness with a milled straight edge.

Indeed, it was a little warped so I used a belt sander on it till if was as flat as the milled straightedge. That seemed to work for about an hour but I was soon back to where I began. I then decided that the table was now adjusted properly and went to a great deal of trouble getting it so on my Rapman. Same result.

In desperation I began to print on the left hand side of the print table and the problem went away. It was still troubling, though.

On Tuesday, I finally caught on. In the last weeks the weather had cooled to where the outside temperature was in the teens more often than not and dropped into the single digits {Celsius} in the early mornings when I began to work. I looked at the printer table and noticed that the window I used to ventilate the work area was on the right hand side of the printer. The next time that I had a raft detach I measured the acrylic work surface temperature with my IR thermometer and discovered that whereas it was about 25-26 C on the left hand side of the table the draft from the window dropped that to 22-23 C on the right hand side. I was rather shocked that I got that wide a variation in surface temperature over a few centimeters distance, but I certainly did.

I closed the window and the peeling instantly stopped.

That remedy wasn't workable because of the ABS fumes, so I rigged a portable heat lamp onto my camera tripod to shine on the print table.

The radiant energy keeps the acrylic print table at 34-40 C with the window open.

I haven't had a raft detach since then. I've done a few dozen prints, mind.

Second go at the carpal ensemble

2010-09-10T19:35:00.000-07:00

Acid test

2010-09-09T11:43:00.000-07:00

I decided to give the new non-loop road finder routine a tough workout to see how robust it is. For that I designed a 20 mm diameter herringbone pinion gear.

I did a solid print for strength. As you can see, there were no problems. The stl's processed without drama and the gcode is good.

Changes to Slice and Dice

2010-09-08T21:04:00.000-07:00

Slice and Dice was fine as long as your parts were more or less continuous along the z-axis and not so complicated that you got a lot of clashing with print loops. When I got to working on the thumb joint on my telepresence hand, however, I started having a lot of trouble with both of those limitations.

A major strength of Slice and Dice is that if you have a dodgy STL file you can clean up little imperfections on the slice images. That's wonderful until you have a part that you want to print that has little in the way of commonality between slices. I found myself fixing faults on 30-40 slices in Paint. That was seriously not fun. The main problem, as it developed, lay in the implicit dependence of Slice and Dice on looped road descriptions. Once you get into complicated parts it becomes very difficult to meet the app's expectation of clean looped print roads.

This is the part that started causing me trouble.

Virtually every slice is unique.

I rewrote the road making routine to deal with non-looped roads this evening. This is the resulting test print of the part shown above.

A problem that the old Rapman firmware was that it expected minimum print road line segments to be about 0.6 mm long. When I finished the rewrite, I was not looking forward to fitting line segments to the non-loop print roads. I had the output, but it was all 0.1 mm print roads.

Andrew at BitsfromBytes has said that the updated firmware would handle 0.1 mm roads with no problem. I frankly didn't believe him because the older firmware would slow the print down to 6 mm/sec with 0.1 mm roads. I decided to give it a try with the 4.0.2 code, however, and it turned out that Andrew is absolutely correct. The Rapman firmware has no trouble print sequential 0.1 mm line segments at my chosen print speed of 16 mm/sec.

That made the print file for the two halves of the thumb joint some 8 meg long. Since I have a 1 gig SD card, however, that's no trouble at all.

Wondering what the fuss is about

2010-09-06T09:04:00.000-07:00

In which your narrator wonders what all the fuss is all about?

I guess I just don't get it. When you work with a plastic, after a while you get a feel for what it can do and what it can't. I started getting along along quite happily with no heated bed and ABS.

I print with a 0.3 orifice at an axis speed of 16 mm/sec. I suspect that I could kick it up to 22 mm/sec without a lot of drama, but I don't want to take the time out to play "who can print fastest" games at this point.

Once I started doing thin walled pieces and no infill my warping problems virtually disappeared. Most of the pieces I design have a largest dimension <= 90 mm though I've printed a herringbone rack that was 250 mm long. My acrylic print table temperature stays at about 25-30 degrees. You also don't find my prints warping after a few days from the internal stresses that Bogdan has talked about. I've seen that with HDPE. I was also printing with cross-hatched infill in those days, too.

I got into thin walled, no infill after I realised that if I went that way I got pretty fast prints that way at lower print head velocities.

Watching you guys reminds me of the first and second year architectural studio students that I used to lecture to back in my university professor days. They'd make models of buildings out of either "shipboard" {2 mm solid cardboard} or carve them out of expanded polystyrene foam treating it like it was so much cool butter or cheese. We used to say "form follows chipboard". If you followed the careers of those students they usually spent the first five years of their careers designing actual building that got built that looked as if they'd been carved out of cool butter. That's an extremely expensive way to design and build buildings. Those guys either got out of the habit fast or wound up doing architectural detail drawings for other designers who'd developed a feel and respect of the potential and limitations of the materials that went into their buildings.

Around here I see parts designed like you were more used to using an expensive CNC milling machine to carve parts out of a block of steel or aluminum.

That looks really cool but begs the fact that you're not taking advantage of the strengths and avoiding the weaknesses of the material you're working with, viz, plastic and extruded plastic at that. Somewhere along the line with Reprap we got the idea that we ought to be able to design parts in any shape we wished and whatever the material wanted to be be damned. You can see the trouble it's caused us in the chase after things like heated beds and support materials. We're spending a lot of time on that chase when we could be designing killer apps that make having a much simpler reprap machine very desirable.

That's just an old architectural technologist talking, I suppose.

I've included a link of the telepresence hand that I'm currently designing. That is human sized, btw. The largest part dimension is a touch over 80 mm. No warping whatsoever on any part.

Thin walled, no infill, snapped or glued together. There'll be a few screws in it to secure some elastic bands that return fingers to their rest positions. I haven't figured out how to secure elastics bands with snap on parts or glue yet. I'm thinking about how that might be possible all the time, though. :-)

Making a lid for a potentiometer mounting box

2010-08-28T00:06:00.000-07:00

In which your narrator shows you a few tricks of the trade with thin-walled prints.

In my telepresence robot hand project I'm not using off-the-shelf servomotors but rather making them from scratch. A servomotor basically consists of a gearmotor, a potentiometer and some electronics, in my case an MCU.

You saw in my previous posting how I'd developed the hand segment and sorted out the gearmotor positioning therein. That done, I had to turn to see to mounting a potentiometer. I decided to simply add an extension to the back of the hand segment to contain the potentiometer.

I'll not go into detail about the design of the box itself but rather cover the trick one uses to design the lid for the box. Here is the box.

The mounting box is nothing special. It's a simple square box with mounting rack groves in the sides, seating for the potentiometer and screw posts to secure the lid.

The box is simple to design. It took me about seven design iterations to make everything fit properly and to get the proportions worked out. Using the Reprap 3D printer at every step meant that it was a very low risk exercise.

The box itself in Art of Illusion 2.8 was a simple box in which I'd removed the mounting slots on the sides with two boolean ops.

Once that was done I simply removed the voids where the potentiometer would be seated and where the mounting screws would be fitted. I've switched Art of Illusion over to wireframe mode so that you can see the voids since they don't penetrate the surface of the mounting box.

Designing thin walled parts is a bit like designing an old photographic negative. You design the skin of the object and the holes inside of it rather than trying to design the object itself. When you print the object you simply don't use any infill. The top of the box is just another kind of infill, so it is omitted, too.

Here you can see how the printed box seats the potentiomenter.

Now, designing the lid you simply take the Art of Illusion file for the box and leave out the void that seats the potentiometer.

You want the lid to have holes that match the screw posts in the box so you leave those voids in.

Next you slice off the top of the box leaving the screw hole voids exposed. You do this in Art of Illusion by simply creating a block and merging it with the top of the box down to where the screw voids begin.

Then you remove the top of the box using Art of Illusion's boolean ops function.

While you don't want to have a hole big enough to slip the whole potentiometer through, you do need to accommodate its shaft. A simple way to do that is to simply take a copy of the seating void cylinder and put it back into the object as a solid rather than a void.

Since we've sliced off the top of the box you can see the top of the cylinder now. Now it's just a simple matter of reducing the radius of the cylinder to a bit more than the radius of the protruding shaft.

Now do a boolean op to remove that cylinder so that you there will be a hole in your lid.

When I looked more closely at the potentiometer I noticed that there was a little metal tab on one side to act as a stop to keep it from rotating around its shaft. I had to design a little slot into the lid to accommodate that tab. I did that by simply locating a little block where I wanted the slot to be.

Then I did a boolean op to remove the space occupied by the block.

At that point I brought back the cube I used to slice off the top of the box and moved it down so that exactly the thickness of the lid was exposed.

I then did a boolean op that chopped off the bottom of the box so that only the lid remained.

At this point I discovered that I'd made a mistake. I'd designed the hole in the lid so that it fit the shaft. Actually, the shaft fitted into a seating hub in the potentiometer that stuck out about a millimeter. I brought back in the shaft cylinder that I'd used a moment ago and widened the radius to a bit over the radius of the hub.

I then removed that cylinder via a boolean op which widened the hole to accommodate the seating hub.

Now comes the tricky part. The lid as designed will sit on top of the box. I want it to recess into the box. I could try to do this using Art of Illusion, but there is a much simpler way much less likely to fail.

Remember that the box will be printed without infill. In this particular case we will print it with two 0.8 mm print roads describing the perimeter with a thickness of 1.6 mm. Print roads tend to be rounded on the inside of the box, a fact that makes working them with Art of Illusion a bit messy. It is easier to just process the lid in Slice and Dice.

Once you've done that you simply go into the Filled folder and pull out the image of the print roads for the lid.

In the present state of development of Slice and Dice with complex prints roads like this it is necessary to go in with Windows Paint and clean up the print road image a bit. Here you can see a few flaws in the roads image.

I've circled two of the most obvious. In the upper right hand circle we have a feature that is less than 1 mm in diameter. The lower circle encloses a loose print segment that is too small to print. It's best just to remove those in Paint. You will also, when inspecting the print, note that several of the print road loops are broken by a missing pixel or have tags of pixels hanging on. Those should also be fixed or removed.

You usually only encounter this kind of problem on slices with very complicated print roads. That's not a big deal and it is a quick thing to fix in Paint. Since Slice and Dice processes images rather than arrays of numbers at each step you can go in and make changes with Paint if there is some flaw that you want to fix. You can even alter the print roads in Paint if that suits you.

Altering print roads is exactly the trick we will be using to make the lid recessed. The box has a perimeter of two print roads. We simply go into Paint with our lid and erase the outer three print roads.

That lets the print box lid fit into the box and sit atop the screw posts inside. We could have simply removed two. In practice, however, that requires a bit of touching up with fine grit sandpaper on the edges to get the lid to fit. Removing three roads leaves a 0.8 mm gap between the box and the edges of the lid. That's good enough for this job.

Here you can see the potentiometer box with the newly printed lid lying beside it.

Now you can see that the lid fits nicely over the potentiometer.

After that it's a simple matter of securing the lid with metal screws and putting on the potentiometer's washer and nut to complete the job.

Now you check the fit between the potentiometer box and the hand segment.

You are now ready to connect your potentiometer box onto the back of the hand segment with boolean union ops in Art of Illusion and then print out the resulting large part.

This is how you create a large complicated part by designing it as a set of smaller, simpler parts.

Designing with a Reprap machine in the loop

2010-08-25T23:16:00.000-07:00

In which your narrator celebrates a return to craftsmanship made possible with the addition of a Reprap printer into the design process.

In the five years I've been on Reprap team I think that possibly the most terrifying and depressing transition of the many I've made during that time was actually having a 3D printer on my worktable and having to face up to turning some of my pipe dreams into physical realities.

Let me tell you, it's a lot easier to dream up something than to actually cobble it together and get it working.

To begin with I had discovered that thin-walled prints tend to be quite strong, properly designed, and don't have nearly the trouble with warping that we've had with large, filled objects.

For a project I wanted to create an animatronic hand for a telepresence robot project I've wanted to undertake for several years now. If you go out to and buy one you're looking at anywhere from $5-50K for one hand. I haven't got that kind of money so I dug around on the internet till I found a guy in Indonesia who'd made one for $300 in parts.

Andreas Maryanto's approach is both brilliant and cheap. I used his project as a starting point and was able to use his blueprints for getting proportions and sizes right.

I was also impressed with the Meka Robotics hand design.

Meka's hand was expensive, but had several interesting points. It was self contained and used elastic bands to return the hand to an opened position. I liked that.

I started small, with a fingertip. Actually, half of a fingertip. I did the design with Art of Illusion 2.8. AoI has always been about the easiest to learn, free 3D CAD app around. With the introduction of version 2.8 the problems with the boolean operators have been largely solved.

While AoI 2.8 will, occasionally, create a flawed STL file, these can be easily repaired in the free Netfabb Basic app. The combination of Art of Illusion and the free Netfabb app, now available on the cloud as well, makes for an extremely powerful combination.

I wanted the joint between phalanges in the finger to be integral with the print, that is, not requiring any additional hardware. I printed and worked with the finger tip for quite some time.

Eventually, after a few dozen prints and redesigns, I got it to do what I wanted.

After that, the rest of the finger came together rather easily.

So, I had the joints right, but how to integrate the actuating tendons was a puzzle. Within a week or so I had a tentative solution. I could incorporate slots in the phalanges of the finger to guide and protect the tendons. This meant that I had to import the STL files for the parts back into AoI and carve out the slots to seat the tendons. Here you can see the STL for the second phalange slotted.

After slotting the second, third and fourth phalange in AoI and reprinting them I had a tentative solution for flexing the finger.

At that time I used a single tendon running through guides along the top of the finger. That was later to change.

I quickly explored using servos very much like the ones Andreas had used.

...and just as quickly rejected them as overspecialised and too bulky for use in my hand. From there I went on to creating my own servos from scratch using gearmotors and potentiomenters driven by a micro controller.

Once I'd settled on a gearmotor that I had a supply of...

...it became a matter of designing a section of the actual hand and accommodating the gearmotor inside.

Notice that the design of this part is only aiming at holding the outline of the gearmotor and providing a recess for the finger. The goal is to work out the seating of critical parts and getting the general shape of the part right before trying to accomplish more.

At that point, I needed two finger assemblies to get a feel for the clearances and volumetrics between finger ensembles.

Attention to the reel that would wind up the tendon that contracted the finger was the next step. I'd decided by this time to use a Meka-style elastic band on the top of the finger to return it to the open position.

The reel was a tiny, finicky part, given the clearances in the hand section. I printed two of them at a time to avoid having to do a lot of pausing for the reels to cool between printed layers.

That done, I needed to do a bit more work with the hand section. First, I needed to seat the gear motor. The hand segment was a big, rather complicated part, so rather than trying to redesign that I simply designed a plate that could be glued onto it that would achieve the effect.

By this time I had slotted the reel to make attaching it to the nylon tendon easier.

While I glued that gear motor support plate to the hand section, I could just as easily do a boolean union in Art of Illusion between the two parts and print them as a single part, or not, as convenience dictates.

From there, I designed a housing for the potentiometer that butted onto the base of the hand section. You can see it on the left-hand side of this picture.

Notice how I haven't bothered gluing this part to the bottom of the hand segment but rather just clamped it. I've not settled on all the dimensions and placement of this part yet, so gluing isn't justified.

Similarly, you can see here that I just took the hand section into the free Netfabb app and flipped it into a mirror image as a start towards completing the hand segment.

The point of all this is that you can take on some really challenging design exercises if you will just take them a little bit at a time. Having a Reprap 3D printer {in my case a Rapman 3.0} makes it possible to print out a partially designed part and try to match it up with the bits that go in it and the bits with which it must integrate.

While you can do all that in a 3D CAD system you have to have a very good sense of volume and space to do so. With the Reprap printer in the mix, you don't have to be that good.

Don't be afraid to use filament during the design process. You're not wasting it.

Cloud CAD

2010-08-22T08:12:00.000-07:00

Ordinarily, I publish only my own work on this blog. In this particular case, however, I'd like to pass along a nascent project to make OpenSCAD available in a computing cloud environment.

Tony Buser is the man on this undertaking. It's a very worthy one, imo.

Introducing Snakl!

2010-08-19T18:54:00.000-07:00

He used to live on Tommelise 2.0 but recently took up residence on my Rapman 3.0...

Snakl does all the hissing! :-D

Chasing bugs in Slice and Dice

2010-08-19T11:57:00.000-07:00

I had completed my narrow profile telepresence finger and then set about designing a mount for the gearmotor and potentiometer to drive it. The mount came out looking a bit weird.

Trust me, though, there were good reasons why it took the shape that it did ... in my twisted mind, at least.

The first thing I discovered was that my print roads routine is not extremely happy with sharp ended print roads like you see here.

Fortunately, I was able to pull the few unique slice images into Windows Paint and clean them up rather handily in just a few moment. Slice and Dice is very good about giving you ways around problems you encounter with difficult parts. The messy print roads were only the beginning of my troubles, however. When I tried to turn the image into an XML description of the roads all hell broke out. My pathfinder routine, which worked well enough for most slices, really hated this gearmotor mount.

I finally gritted my teeth and spent the time chasing up the many bugs in the pathfinder routine. Once that process, which took about two man-days, was complete, I was able to get a pretty good XML conversion.

Red overlays indicates well formed print roads. The black residue shows where the routine failed.

You can see the failures circled in blue. Looking a bit closer at a few of the failures you can see that the routine breaks down when the distance between two parts of a print road drops below 0.2 mm. That's a ridiculous case, but slice and dice routines regularly encounter ridiculous cases. Here is a typical one.

Here you can see that the pathfinder routine jumped the 0.1 mm gap between two sections of the path. Another thing that became obvious is that the routine can't handle paths which are less than 1 mm in their greatest dimension.

None of these faults are serious enough to cause me to continue working on the code at the moment. I'm back to printing.

Peeling parts off of the raft

2010-08-14T21:52:00.001-07:00

I am rather far along in the finger design for the telepresence hand. Now that I have flexible sensors and a need to modify the design to accommodate them, I decided to take the time to reduce the width of the finger design to something approximating my own finger. This entailed reducing its width by 25%.

Doing that was a simple matter of rescaling the STLs in Art of Illusion for the narrower width. Once done, however, the phalanges were so small that using the belt sander to take off the raft was something that was no longer practical.

Heretofore, I have simply printed the raft and part all at 240 C and used the belt sander to take off the raft after printing. With the more delicate narrow finger phalanges, however, I could not hold the parts in my fingers safely and had to use pliers to hold them for the sanding. Unfortunately, getting even removal of the raft by that method was a sometime thing.

After a few goes with the delicate phalanges, I decided that it was time to upgrade the Slice and Dice code to vary temperature by layers in order to make the part separable from the raft.

I use a rather traditional raft with heavy, widely spaced print roads for the first layer to even out any misalignment of the print table and flaws in the print table. I print this at a print speed of F240 and an extrusion rate of S360. For the next and last layer of the raft I quadruple the print speed to F960 whilst keeping the extrusion rate the same. All of this is done at a print temperature of 240 C.

What I decided to do for a removable raft was to keep the first layer as it was and then add a third layer to the raft. The second and third raft layers were to be printed at a lower temperature and then the print itself done at 240 C as before.

I did a series of prints lowering the temperature of raft layers 2 and 3. The sweet spot seems to be at about 215 C. At temperatures higher than that adhesion between raft layer 1 with 2 is too high and adhesion between layers 2 and 3 and the print object is too high as well. I could not get reliable extrusion of ABS at temperatures lower than 215.

The new print protocol produced a raft and print which were reasonably easy to separate with one's fingers as you can see here.

the bits of layers 2 and 3 sticking to the printed part are easily removable with either the belt sander or a piece of conventional sandpaper in just a few moments.

Finishing the reel

2010-08-08T20:43:00.000-07:00

I discovered that while the mismatch of inner and outer loops can be sorted out on a single slice fairly easily, things get tougher when you have successive slices getting smaller as I do with the 45 degree angled slopes on the servo reel.

What I did was to make two halves of the reel with boolean operations like this...

Slice and Dice lets me control the print roads on each slice, so I simply arranged it so that inner and outer loops did not meet on one side...

...and left off loops on the other side to make the reel into two halves that fitted together.

So, it was effectively back to glued together plastic model airplane dodges again.

Improving print roads calculations

2010-08-07T22:43:00.000-07:00

In which your narrator rewrites the Slice and Dice routine that calculates print roads on a slice.

Having got a working finger design for my telepresence hand project, last weekend I set about to rig it to a servo motor to see what issues would come up. One of the items needed was a reel which screws onto the servo which takes up the tendons on the finger and moves them in concert.

The 40 mm reel looks like this...

I wanted to make the reel solid since it will be under considerable torque. To achieve that I used my concentric print roads option rather than cross hatching. I am not fond of cross-hatching mostly because the join between the perimeter of a print and the cross-hatching is so often mechanically poor. Using concentric print roads yields a much stronger print.

I calculated concentric roads very simply by painting a strip of pixels on the boundary of the slice to create an interior print road, deleting the strip to define the print road and then painting another strip inside of the new, smaller boundary. I repeated that simple process until I ran out of slice.

Heretofore, I'd used concentric prints for relatively shallow prints like this phalange...

This looks pretty good. I had been quite happy with the concentric roads routine till I applied it to the spool. The spool was a deep object, 40 mm in diameter with an interior and exterior boundary. You can see what happened...

The roads track the original boundary quite nicely till you get 6-8 roads away from the boundary. Then, geometric features of the original boundary are sufficiently magnified that the roads bear little resemblance to the shape of original boundary. The reel provides a really nasty example of this kind of distortion.

The reel that resulted was actually very strong, but looked nasty. Last night, I undertook to see if there was another way to do the job.

After several false starts, I discovered that if I began with the original slice boundary in one picture box and then mapped a filled circle of the proper radius onto each pixel in the boundary I got extremely smooth print roads. The reel slice you were first shown looked like this with the new routine...

You will notice that the interior and exterior boundary roads match each other very well but show a gap between the interior and exterior roads. The hole in this slice of the reel is a screwdriver access port for securing the reel to the servo drive shaft. The dimension in this slice is just big enough to pass the screwdriver tip. I widened that access port by two millimeters and the problem with the roads mismatch disappears as you can see here...

The reel consists of two parts; the reel itself and an insert which is glued into a recess in the reel which seats over the servo drive shaft and allows for a screw to secure it. You can see how this works in this exploded view of the ensemble...

As designed the print road pattern for the reel slice defining the recess for the insert looks like this.

If you look closely you will notice that the meet between inner and outer roads is a bit tight. The outer dimension of the insert is not critical and can be adjusted a bit to even that out. Care must be taken, however, to not compromise the roads in the insert in the adjustment.

The more experience I get printing dimension critical parts the more I understand what a balancing act getting the proportions is. When you are printing a critical part you can't just slap it together in CAD, pump out an STL and print it. The print road cross section is an integral part of your design and you have to exercise considerable cunning in making everything work together. This experience is much like what I learned after I graduated from architectural design studios eons ago. It's easy to design something that looks nice if you have any sense of aesthetics at all. Designing something that looks {or works} nice that can actually be built is a far harder task.

In any case, I've got the new print road mapping routine running rather well and am going back to work on my telepresence hand project.

Applying VTEC to Slice and Dice

2010-08-04T21:35:00.001-07:00

In which your narrator lowers the execution time of Slice and Dice by upwards 90%.

Now that I am well into designing the hand for my telepresence project, I am running into the limits on the processing speed of Slice and Dice in the design cycle. In the past, I could run the code through a good wash and brush up and reduce the execution time by 90%. That would take about a man-week that I don't want to spend. Instead, I chose to take advantage of a characteristic of the parts I design to achieve faster execution speeds.

Typically, I have been designing parts that look somewhat like this.

When you look at it closely, you notice that there are really only two different cross sections through the vertical axis.

Now Slice and Dice as I wrote it is not particularly bright code. I wrote it to be simple, not efficient. When you slice this object into 0.25 mm wafers you get 39 slices that look something like this.

Slice and Dice tediously processes each slice as if it were unique. I took a day off to see if I could take advantage of the geometric repetitiveness of my parts. To do that I initially thought to simply match each image with succeeding image and throw out the duplicates.

Life wasn't quite that simple. While the slices looked identical, miniscule differences in the way the slicing routine interacted with triangles showed up in the flooded pictures seen above. I did find, however, that if I allowed flaws in the pixels of up to 0.2% of the total in the flooded images the 39 images were reduced to exactly 2...

Which is what you would have expected if the slicing had been geometrically perfect. Thus I had to do cpu intensive processing on two slices instead of 39. That saved about 94% of the processing time that I had previously been using and got my processing speed into the range I was used to seeing with Skeinforge.

As an added advantage, the reduction routine picks up broken slices very nicely.

This approach would not, I think, work with a programme like Skeinforge or, unless I misunderstand how it works, Netfabb. It could, however, easily be applied to Adrian's host code for Reprap.

Printing in trays

2010-08-01T22:01:00.000-07:00

Ordinarily, I like to print parts one at a time while I am design them. The telepresence finger, however, has reached a rather advanced state, so I've written a small script to shuffle two print files together so that they can be simultaneously printed.

The travel time between prints gives them a chance to cool and lets me print much smaller isolated features like the post circled in red successfully than would otherwise be possible.

Some thoughts and observations about having a Reprap machine in the design cycle

2010-07-28T09:58:00.000-07:00

In which your narrator reflects on the rather radical difference between what he perceived the design process would be pre and post the advent of practical Reprap printers.

Do you want to read more?

From 2005 till November of last year, I spent most of my free time trying to build a Reprap printer. It was good fun and I learned a lot of things. Mostly, I learned how not to be intimidated by technology that I was not familiar with. For me, this was an extremely valuable lesson.

When my day job began to require more and more of my attention towards the middle of last year, my development of Tommelise 2.0, my own repstrap project began to suffer. Finally, at the end of October I sat down and did an unpleasant assessment of my situation. I could continue to work on Tommelise, or I could simply buy one of the Reprap-derived kits that were beginning to emerge from small enterprises started up by other core team members. I estimated that it would take something like another 6-8 months to get Tommelise 2.0 printing properly. I subsequently ordered one of Ian Adkins' Rapman 3 printers.

Rapman 3 is a re-engineered clone of the first generation Darwin design. I bought it rather than the somewhat cheaper Makerbot primarily because it had a substantial print volume but also because an associate, Batist Leman, had been blogging his experience with the model. Batist was very impressed with the Rapman and communicated that very well through his video clips of it in operation.

Prior to November, I was designing and making parts with the object of making a Reprap machine. Afterwards, I was designing and making parts WITH a Reprap machine. The distinction can be easily lost on those who have not used a Reprap machine.

What I have discovered in the eight months since is that the availability of a Reprap machine makes a massive change in one's design practices.

Prior to November I lusted after a professional 3D CAD package. Features like dimensioning and reliable boolean operations cluttered my thinking. The problem was that I was thinking like an engineer, viz, I wanted to make careful designs in 3D and then have print them out on an accurate 3D printer. I no longer feel that way.

With the advent of Art of Illusion (AoI) 2.8.1, I stopped lusting after "professional" CAD packages even though by that time I already owned several. My design cycle now looks like this.

(re)design (a) part(s)
print the part(s)
manually and visually see how the parts work together
repeat the process until satisfied

Precise dimensions became not very important while things fitting together properly did. The linkage between these two factors was not as strong as one might first suppose.

With a Reprap printer it became a simple matter to run through a dozen design cycles to get a parts ensemble to work in a matter that suited me. I currently design with the Reprap printer as part of the design cycle rather than using it as a final step after design cycles are finished.

At the beginning of this year Skeinforge, which I had been using previously, went through a rough spot. This encouraged me to invest in the upcoming Netfabb software prior to its release. As the Skeinforge rough patch continued and the folks at Netfabb's release dates began to slip, I finally became frustrated and then angry. I wasn't able to do the design work I wanted.

In frustration, I unearthed my old Slice and Dice code from the Tommelise project and began to bring it up to date. Having used both Skeinforge and the nascent Netfabb offering to process STL files into print instructions I evolved a very different approach to processing solids files than what either of these two products offered.

Both Skeinforge and Netfabb basically take your STL and give you print instructions. They do it very quickly and efficiently. Unfortunately, if you have a dodgy design file they can produce some very crazy print files. Ones ability to respond to crazy print files is, perforce, limited. With both Skeinforge and Netfabb, the majority of craziness is generated when ones STL files are flawed. The need for perfect STL files had fueled my previous lusting after "professional" 3D CAD packages.

I responded with Slice and Dice by internalising Mandelbrot's maxim that noise is always going to be there and it will be unpredictable. Instead of striving for perfection, Slice and Dice concentrated on giving me options to deal with imperfect design files. I began by keeping each slice of a design file as an image that I could bring into simple image handling tools like Windows Paint and manipulate. Paint let me patch and alter flawed slices to suit myself. I no longer needed a "professional" 3D CAD system. AoI was just fine for my purposes.

From there, I discovered that if you use 2-3 print roads to define a print's perimeter you wind up with a strong part even without infill. I began working to improve my control of perimeter print roads and soon found that I could make parts not unlike those in old fashioned plastic model airplane kits which used injection-moulded parts that you glued together. Infill became unnecessary in all but the most extreme cases. Hollow parts were tremendously quicker to print and not nearly so prone to warping. as "solid" parts. I don't use a heated bed and doubt that I will anytime soon.

Conventional Reprap thinking views a 3D printer as being able to print, more or less, any 3D object. Development has concentrated on infill and support materials to achieve these ends. Both are wasteful of printer time and materials, in my opinion. I try to design parts which keep in mind the strength and weaknesses of my printer.

As you can see with this telepresence 'bot finger, I think I am getting pretty good results.

Memories of plastic model airplanes

2010-07-26T11:22:00.000-07:00

In which your narrator harks back to his youth and fingerprints ruined by trimming plastic flash off of model airplane parts with double edged razor blades.

I have an aversion to infills. Recently, I realised where the aversion came from. During the 1950s you could buy plastic model airplane kits for $1-2. These were made with injection moulded parts that came on little flat plastic Christmas trees. Exacto knives in that era were both expensive, for a child at least, and the blades dulled quickly and were largely beyond the ability of a child to resharpen. As a result, this child nicked paper-thin double edged razors from the medicine cabinet to trim the parts off of the tree and trim the flash off of the parts.

That worked fine, except for the cuts which the blades would make on ones fingertips which eventually grew into permanent scars. That was no big deal in those days.

Having trimmed your parts you glued them together with Testors or Duco cement {ethyl or butyl acetate} and, with a bit of paint, you had a lovely airplane to hang off of your ceiling by a thread and dream about flying.

Early on in the Reprap project, we used Solarbotics gearmotors. When you opened one of the gear boxes of, say, a GM-3 you found that the housing had been made using the same injection moulding process that model airplace manufacturers used. With a working Reprap printer I quickly began to loathe the clumsy, infilled parts that we were designing. They wasted both time {to print} and filament as well. Worse still, one tended to stick the parts together with expensive nuts and bolts.

Why not make parts that fit together like the old model airplanes?

The trick there is that the old model airplanes inevitably had a left side and a right side or a top a bottom that were often simply mirror images of each other. This weekend I was design a finger tip with left and right sides. I had designed the left side and got it working acceptably and was going back to do the right side in Art of Illusion {AoI} when it occurred to me that it would be much simpler to just write a script to swap the sign of the coordinates of the axis that I wanted to mirror around. Writing that script took about five minutes and saved me a lot of time in redesign and reprocessing of the resultant STL since my script operated directly on the Gcode.

Printing the two halves was trivial.

It took a few moments to scrape off the raft and a few drops of cement to finish the part.

The result was an elegant looking part that would have been a right bastard to design and print conventionally.

While I've incorporated this simple mirroring script into my own Slice and Dice STL processing software it would be no big deal to incorporate it into the more widely used STL processing routines available to Reprappers either open source or commercial.

First prints

2010-07-18T14:01:00.000-07:00

Slice and Dice is pretty much working now. I'm not completely happy with how the infill and the perimeter prints hook up, but that's mostly parameter tuning. I'm debugging Slice and Dice while I am designing a robotic hand. Here is the distal phalange {finger tip} that I've done for a first try.

I've left out the infills in that they are not necessary. Here you can see the first print that I am happy with.

I've really got to break down and find a camera that can do better closeups for small objects. My periodontist uses one for photographing teeth, an Olympus, which might be the ticket. I may well buy one after my quarterly tax payment is done towards the end of this month.

As you can see in this second pic, there is a separation between the outer and inner print road for the flexural axis. I've got to reduce some the outer radius there or lower the extruder flowrate and go for three print roads instead of two on that part of the print.

In production

2010-07-17T08:21:00.000-07:00

Got the whole Slice and Dice ensemble running together last night and did some test prints. Vectorization cured up the extruder head slowdown completely for curved print roads. Sweet! :-D

Operational again

2010-07-10T06:05:00.001-07:00

I was able to put the final touches on the Slice and Dice upgrade last night. The app is operational again. Now I can get to work on that telepresence hand project. :-)

Vectorization of pixel defined print roads actually working properly

2010-07-06T19:17:00.000-07:00

In which a very kind Adrian Bowyer takes pity on your narrator.

In my last posting on 21 June, I laid out a method for vectorizing a pixel-defined perimeter. Having read the posting, Adrian Bowyer, a published expert on this sort of thing, took me aside and showed me how to do the vectorization efficiently instead of just at all, the solution I came up with.

Using Adrian's approach, I was able to achieve my goal of getting exact vectorization of the print road and push out the average segment length to well over the 0.3+ mm that was required to assure a print head speed over the major axis of at least 16 mm/sec.

You can see an example of such a vectorization here. All values, save the last, are in tenths of a millimeter.

The last value in the listbox, "rho/axis" should allow me to adjust the print speed on the fly in G1 statements to assure a constant head velocity for print road segments greater than 0.3 mm in length.

I should be printing ABS parts again by this weekend, which is good since I want to redesign and print one of these.

Vectorization of pixel defined print roads

2010-06-21T13:36:00.000-07:00

In which your narrator describes the end, hopefully, of a search for a reasonable way of converting pixel-defined print roads calculated by Slice and Dice.

For the past two months I've struggled with what has been, for me, a very nasty problem, viz, converting pixel boundaries defining print roads for my Rapman printer into equivalent vector descriptions. As I mentioned previously, in order to get a major axis velocity of 16 mm/sec on the SD card Rapman 3.x you have to have GCode vectors of no less than 0.3 mm. If you are printing detail finer than that the lag caused by reading off the card and processing the GCode slows down the print head dramatically.

From my literature search, vectorization is a fraught process, especially if you are not willing to accept considerable degradation of your pixel-defined road. With 3D printing, perforce, we really can't accept degradation.

Over the last few weeks, I've finally confected a method which seems to do the job. As with most things I do, it is relatively simple and straightforward.

Suppose we have a perimeter path for an involute profile gear...

The method grabs a 2.1 mm patch {note the red circle} at the extreme left of the path which you can see here...

Here you can see the individual pixels making up the print road. The first thing we do is define the starting point and direction of the road from the centre of the patch.

At that point I pivot a scan from the center of the patch and identify the closest fits at ranges varying from 1 pixel to 11 pixels.

In this case, the longest perfect fit to the involute profile was 4 pixels.

Red pixels represent the fitted vector while blue indicate the remaining pixels. Notice that this vector is four pixels long, just enough for the Rapman to operate at 16 mm/sec.

The patch is then re-centred over the most distant red pixel and the process repeated until the full loop is vectorized.

Looking at a simpler problem, consider a hexagonal print road...

Isolating the beginning of the print road yields...

Scanning this straight line gets a perfect match at 11 pixels. Thus the way I have the code set up now creates vectors of up to 11 pixels in length, a maximum length of about 1.4 mm. It's reasonable to ask why the vectors are kept so short. In fact, it takes much longer to process the print roads with longer vectors and there is no reason, SD cards having huge storage capacity, not to have large print files.

As I get time in the coming weeks I will be fully embedding this method into Slice and Dice.

New blog...

2010-06-13T12:27:00.001-07:00

I've finally done enough work on my active telepresence project that I need a blog to document what I am doing. For anybody interested, you can access it here.

Processing gcode instructions with the Rapman firmware

2010-05-09T09:37:00.000-07:00

After running a variety of speed tests with the Rapman, I gained enough data to estimate the processing time for two kinds of gcode statement. The G1 statement processes in the Rapman 3.1 firmware at a rate of 17 msec/instruction. M108 instructions process at a rate of 19 msec/instruction.

Interestingly, the rate for the M108 instruction is actually significantly longer than for the G1. Given the floating point instructions necessary to interpret a G1 instruction this seems very odd. It would seem that much more time is spent reading off of the SD card than processing the information on a line of code. Either that or carrying out a M108 instruction in the Rapman firmware is more complicated that it would appear at first glance.

As a practical matter what this means is that if one wants to print a layer with 0.1 mm print road lengths you are looking at a maximum print speed of around 5.8 mm/sec. A print speed along a long road component axis of 16 mm/sec would imply that your average road segment would be no less than 0.3 mm.

Further tests

2010-05-06T22:45:00.001-07:00

I did a very careful, new series of speed tests using F960 (16 mm/sec) over a much longer diagonal baseline, viz, 1.7 meters, with the extruder running.

Here's what I found...

Step (mm) Seconds (s) Velocity (mm/sec)

   0.1 214 5.77
   0.2 110 15.40
   0.3 121 14.14
   0.4 80 21.21
   0.5 80 21.21

Note that...

21.21 mm/sec ~ 16 mm/sec x 2^0.5

...which is about what you would expect for firmware that didn't correct for the angle of the print road.

As you can see, the break comes between gcode distances 0.3 and 0.4 mm and is quite abrupt. This means that the vectorization routine for Slice and Dice can be quite crude and work well with Rapman.

What does F960 really mean?

2010-05-05T14:51:00.000-07:00

The vectorizer in Slice and Dice is pretty inefficient. With certain slopes and curved surfaces I'd noticed that the print head velocity could drop by as much as 25% down from 16 mm/sec to 12 mm/sec. I'd spent the last weeks working on improving my vectorizer when it occurred to me that I really didn't know how much improvement I needed to make on it.

To sort that out, I did a series of test runs where I moved the inactive print head from x = -75 mm to x = 75 mm using different steps with the nominal head speed set to 16 mm/sec (F960). Here are the very strange results that I got.

   Step (mm) Seconds (s) Velocity (mm/sec)

   0.1 25 6
   0.5 5 30
   1.0 3 50
   10.0 2 75
   150.0 2 75

As an example of the gcode I used...

G90
G21
M103
M105
M104 S0
G1 X-75.00 Y0.0 Z20.0 F960
G1 X-74.90 Y0.0 Z20.0 F960
G1 X-74.80 Y0.0 Z20.0 F960
....
G1 X74.70 Y0.0 Z20.0 F960
G1 X74.80 Y0.0 Z20.0 F960
G1 X74.90 Y0.0 Z20.0 F960
G1 X75.00 Y0.0 Z20.0 F960
M103
M104 S0
M103

I did further tests and discovered that unless the extruder is running the extruder head speed defaults to 75 mm/sec. When it is running F960 seems to be the speed of traverse of the longest dimension {xy} of the print road.

Specifically, I ran a 150 mm diagonal and discovered that F960 {16 mm/sec} clocks in at about 23 mm/sec. when you divide 23/2^.5 you get right at 16 mm/sec which means that the traverse speed for the x and y axes, which with a diagonal are equal, is 16 mm/sec.

BTW, I've started getting an SD Error2 report at the end of one of these test runs. I'm not at all sure what that is all about.

Bookkeeping and janitorial duties

2010-04-12T08:03:00.000-07:00

There isn't much that is flashy to report at this time. A lot of my time in the past few days has been focussed on a telepresence 'bot project I have intended to design and build for years that the Rapman finally lets me design and print parts for. Oddly, though, getting the basic software modules that will control the 'bot all working together looks to be far harder than printing and building the 'bot itself. I guess that having worked on Reprap for five years now makes me a lot less intimidated by hardware tasks than previously.

Last week, I discovered that Slice and Dice's vectorization routine, which converts pixel defined boundaries to print areas into vectors usable in gcode, wasn't particularly nice. It was not reducing the boundary definitions nearly as much as it ought to have been. What that means as a practical matter is that it generates a lot more gcode than it ought describing very sort little print roads of a few tenths of a mm. Rapman is not designed to deal very efficiently with that kind of gcode and what happens is that the print speed drops well below the nominal rate set with the "F" specification. That not only wastes time but also puts more plastic on the print than it needs to have in areas where the perimeter curves. THAT makes for poor print quality.

I got to the bottom of that problem over the weekend.

As I set up for testing some new approaches to vectorization, I remembered that shifting from BMP to PNG reduced the disk requirements for a print project dramatically. I have plenty of disk space, but the reduction in disk requirements led me to redesign the files handling for Slice and Dice so that one can readily process and work with more than one print project at a time. That job holds no challenges as such, but is merely a bit of tedious restructuring files handling routines and testing to make sure that they all work. I've got that about two-thirds done.

Thus, slowly but surely Slice and Dice begins to look like very preliminary beta code.

Collapsing the hard disk requirements of Slice and Dice

2010-04-08T05:35:00.000-07:00

I managed this by the simple expedient, suggested by my son Adriaan, of replacing the BMP format slice image files with the Portable Network Graphics (PNG) format which uses a lossless data compression scheme. This reduced my slice image files in size from 16-22 Mbytes per slice to 80-100 Kbytes.

Delta 'bot Axis

2010-04-03T01:24:00.000-07:00

It's one thing to design and print plastic parts that have to mate with steel rods. As I am finding it is quite another to design plastic parts that have to mate with other printed plastic parts. During the last week in my spare time I proved out a design for a hexagonal column which I think might accommodate a herringbone rack and a sleeve seating a pinion gear, stepper motor and two guide plates.

Heretofore, I have been butt seating the column sections and then post-tensioning them. While that was okay for demonstrating the design approach, worries about creep said that I needed to have a male/female attachment technology so that I could fit sections of the column together and avoid having them do a lateral creep out of alignment. That proved to be considerably more difficult than I imagined.

This last weekend, I decided to take a different approach and instead of doing a conventional slice and dice of an STL, I decided instead to work with the print roads of my thin-walled column instead. What I did is a radial 45 degree overhang inside the column onto which I printed the male insert. I think it is worth mentioning how I achieved that.

The column was achieved for most of it's length with three concentric print roads of 0.6 mm width. What I needed to do was to build a ledge inside the column. When I began the ledge I widened the innermost print road to 0.8 mm. On the next layer I replaced the 0.8 print road with two 0.6 print roads. This process left me a 0.2 mm miniledge onto which the new innermost print road could adhere. I repeated this process until I had a ledge wide enough to print the male insert onto.

You can see the insert successfully executed here.

I would like to be able to say that that was all an easy exercise. It wasn't. I did exactly thirty text prints like the one you see before I was satisfied with the seating of the guide strips, herringbone rack and the male/female column connector system. I also had to chase several heretofore unsuspected bugs in Slice and Dice AND extend the coding somewhat. All the same, it's done and works so far.

At that point, I decided to see how tall a monolithic column section I could print. It appears that Rapman three can handle about 225 mm in the z-axis. I designed a 200 mm column with a 4 mm male connector and am printing it at present.

It is 2120 and I am up to about 50 mm. I'm doing a 0.25 mm layer every 45 seconds, so I expect to be finished at about 0500 tomorrow morning.

With such a large print I began to encounter xy and z faults. The xy faults were annoying, but the z faults tended to plunge the hot extruder head into the print. Upgrading the Rapman firmware to 2.0.8 got rid of the z faults but the xy faults remained.

I reduced the column test print to 55 mm, relocated the Rapman further away from my PC on a table by itself away from the cold window and switched ABS to HDPE because the acrid fumes were getting to be too much when I wasn't able to open the windows.

A thorough cleaning of the extruder with compressed air was also undertaken. After a few test prints I got the flow rate sorted out for HDPE and was able to successfully print a 55 mm column section in HDPE.

Design study for axis columns and stepper motor sleeve

2010-03-24T22:10:00.000-07:00

This is a very preliminary design study for a paired column section and stepper motor sleeve of one of the three vertical axes required for a Delta Robot.

This is a segment of the herringtone rack that slots into one of the three slots on the column.

This is all very tentative. I am trying to learn what fits together with a minimum of fuss. I've found that if I have shapes in my hands I can see their possibilities and what wants doing with them a lot easier than if I just look at them on a screen. Hell of a situation for someone who trained as an architect, don'tchaknow. :-D

Stacking slices to make simple boolean unions

2010-03-21T23:36:00.000-07:00

Art of Illusion is a pain in the neck to work with. On the other hand it has the shortest learning curve of any 3D modeling package I've ever encountered and is simple enough for an eight year old to operate.

When I say pain, the primary problem is the boolean ops module. AoI does boolean ops with triangular mesh defined solids. Doing boolean ops on mesh defined solids is a brass bound bastard of a problem. There is a tendency to condemn AoI because its boolean ops module isn't very good. When you go out and start looking at other 3D modeling packages, commercial and open source, you begin to notice that virtually none of them, or all of them for all I know, can import mesh defined solids formats like STL. When you look at how these apps do boolean ops you soon find that the objects you can define have to be very carefully defined out of shape primitives and that these primitives, suitably adjusted by size, orientation and position parameters are what go into the boolean ops module.

That's cheating.

Before I get off into a pointless rant about that, let's get back to the problem I was addressing. AoI is rather simple to write special purpose scripts for. Recently, I wrote one for generating herringbone racks. It was a pain in the neck to design that script, but once it was done it worked perfectly. The rack has a 5 mm flange on either side.

When I set about to design a beam into which that rack could slot, however, I discovered that it would be a very good idea to have some space between where the rack stopped and the beam started. To do that I needed to add a wider, somewhat thicker slab under the rack. I had two choices. I could either go back and rewrite the script {shudder} or I could glue that rack onto the heavier slab using a boolean union op, sort of like this.

While AoI would let me merge the slab and the rack, when I went to do a boolean union op, it simply wouldn't let me. Not nice. It was infuriating to be able to look at what I wanted on the AoI display and not be able to turn it into an STL.

That got me to thinking. Basically, all I wanted to do was put the rack on top of the slab. If I put the slab through Slice and Dice to create a set of slices and then repeated the game with the rack, keeping the same alignment, it would be a simple matter to just stack the rack slices on top of the slab slices. Of course, when I set out to do that I unearthed several previously unsuspected bugs in the Slice and Dice code which took me most of Sunday to chase down and fix.

It works now, though. Slice and Dice can do a simple boolean union between two objects by stacking slices. I've tested the code and ran it through the whole exercise in the morning.

And there it is.

The slab is 80 x 25 x 2 mm. The slab and flange are 100% fill. I printed with 0.1 mm layers using a 0.3 mm extruder at 16 mm/sec. There was no corner curling or warping at all. This print is extremely strong.

Time isn't free

2010-03-20T15:30:00.000-07:00

Slice and Dice takes up a lot of disk space because it creates and operates on bitmap images, each one of which takes up 22 megabytes, and keeps half a dozen for each slice of a print. That eats up the gigabytes in a hurry.

The problem is that if you have a nasty STL that you've pulled out of Thingiverse, for example, you have to repair the faulted slices. On the recent torso that I printed about 6-7% of the slices were faulted. This usually meant dotting in one or at most two bits on an image to put things right. The problem is that after you've done that on 30-40 slices you're rather hesitant to throw all that work away because you need the space.

Some weeks ago, I discovered that if you zipped bitmaps you could reduce their size by about 95%. This morning I looked into converting over the storage modules to do this using a compression/decompression api. I quickly realised that this kind of job was going to require 3-4 man-days to do. Instead, I went down and bought a two terabyte external disk.

No peel, no warp, no backlash

2010-03-19T19:17:00.000-07:00

I got Slice and Dice working well enough to let me do some serious R&D. After a side excursion with partial STLs of hands and naked ladies, I got back to work on the herringbone rack and pinion technology.

My first exercise was to see if I could print single herringbone pinions without having a lot of meltdown problems. Skeinforge had a lot of problems in that regard. In fact, Slice and Dice let me do that.

I printed it pretty much solid for strength and paused it for a couple of minutes midway through the print and let it cool down for a few minutes. No problems. This one was printed with 0.25 layers.

Next came the rack. I decided to use the exercise last night to press the limits of Slice and Dice. I figured that I ought to be able to print a 250 mm diagonal layout rack. That let me find a half dozen limits bugs in the S&D code and get them cleared away. I also decided to see if I could print a useful 0.1 mm layer.

In fact, I could. Here we are about halfway through the print.

It took about an hour, by the way.

Here it is, complete.

And a closeup of the teeth.

For some reason the extruder is not shutting off between layers. While that is easy enough to clean up with side cutters, I'll be diving into my code to see if I can sort that out in the next few days.

The pinion gear mates with the rack with no backlash.

I didn't make much of an effort to clean up the pinion which you can see in this outdoor pic.

I've found that the colour rendering is always better outdoors.

Finally, for those of you who haven't a sense of scale looking at the Rapman print table.

My next task will be to figure out how to put the pinion on a printed, extended shaft so that it can be secured from both sides and mate with the short axled NEMA 17 that will drive it. Then it will be a matter of integrating my Pololu Allegro 4983 microstepping driver board into my I2C bus and driving the thing from my Microchip 18F4550 uC board.

Printing the naked lady with Slice and Dice

2010-03-17T21:11:00.000-07:00

After my experience trying to print the naked lady with Netfabb I decided to have a go at it with Slice and Dice. I had considerable trouble getting the polymer flow right to bridge the base of the pundendum. Once I got that right, I set up a shell print with zero infill.

The print went well with the exception of one delamination flaw in the left leg as you can see in this pic. The lack of support that a light infill would have provided became apparent as I printed up to the top of the breasts and the area around the collar bones where considerable bridging was required.

Aside from that, however, Slice and Dice produced a superior print quality as you can see.

Very little of the z-axis juddering was apparent and the surface quality was superior. Indeed you could even see the modeling of the body with triangular segments in places.

That was fun. Now back to work.

Back to Plan B

2010-03-16T19:43:00.001-07:00

After spending some time with Netfabb for Rapman Beta, I've come to the conclusion that it isn't ready for prime time. That's not to say that it won't be and that, when debugged, it won't be a brilliant piece of software, attractively priced. The problem for me is that it isn't helping me solve the sorts of printing problems that I have ... NOW.

I reached a similar conclusion about Skeinforge over a month ago. At that time I began upgrading and rewriting the Visual Basic .NET Express 2008 code in my old Slice and Dice app that I originally developed for Tommelise 1.0.

The deal breaker for Netfabb was the grim performance with respect to z-axis juddering. I did a comparison not, I'll readily admit, a completely fair one. I created a 50 mm diameter hollow cylinder and ran it first using the "very fine" setting on Netfabb. This is what I got.

This test object gives you a clear idea of the effect of the z-axis juddering. You can also see delamination taking place on a rather large scale and you can see the attempt by Netfabb for Rapman to style some sort of cover over the top of the hollow cylinder.

Now Bogdan has rightly pointed out that a lot of my problems here stem from the fact that Netfabb prints at 260 C and wants a fan that I haven't bothered rigging. That is not, however, the only problem that the beta has at this point, however. For purposes of comparison, I ran the same test object through my own Slice and dice with a 0.25 mm layer setting {Netfabb at "very fine" uses a 0.125 mm layer}. Here is what I got with Slice and Dice.

No delaminations and a much, much less pronounced z-axis juddering. I generated the test object with Art of Illusion and used a 0.05 mm conversion of the cylinder to a triangular mesh. That left me with a 64-sided approximation of a cylinder, if I remember correctly. The facets are clearly visible on the Slice and Dice print and nowhere to be seen on the Netfabb print.

There was simply no comparison between the two insofar as print quality was concerned.

After the disaster that I had printing herringbone pinion gears Bogdan asked for the STL files and did quite a nice job printing them with his 0.5 mm extruder {I use a 0.3 mm extruder}. I decided to run those through Slice and Dice, too. While I was at it I got in and cured, not nicely mind but cured all the same, the flaw in the AoI herringbone pinion gear script that I wrote some time ago. I will be publishing that with a download link in a few days for anybody who needs it. Actually, you can cure the flaw with the Netfabb repair feature, which is a brilliant little facility that has saved me no end of pain and suffering dealing with STLs coming out of AoI.

That done, I did a solid ABS print of the pinion with Slice and Dice.

.
The quality is VERY good.

I printed it in about fifteen minutes including a 5 minute cooling period that I did via the simple expedient of pausing the print midway. I've found that if you let your print get much hotter than 90 degrees you get detail blurring.

Right now I am processing the torso that I earlier attempted to print with Netfabb with Slice and Dice. It will be interesting to see what kind of print quality Slice and Dice gives me with that art object.

Netfabb for Rapman Basic ... FAIL

2010-03-14T21:08:00.000-07:00

Okay, now I'm not happy any more. I designed and used Netfabb to clean up a set of 4, 12-toothed herringbone pinion gears that I have successfully printed many times on the August 2009 release of Skeinforge. Here is what they look like in Netfabb for Rapman Basic.

Here's what happened when I ran the gcode and "printed" them.

I think it's pretty clear that the raft technology that Netfabb has designed does not work for small, detailed objects.