Monday, December 28, 2009

Printing Geneva Wheels

For some time I have been working on a magnet coil winder project. One of the things that I wanted was a tally counter to keep track of how much wire I was putting on the coil and another to keep track of how many turns that I made on my electromagnet spool.

For a long time I tried to simply tried to replicate a classic tally counter. Unfortunately, that design was optimized for the use of springs and disks and a pawl of spring steel. I wanted to do the whole thing in plastic, or as much in plastic I could, so I turned to the venerable Geneva Wheel to do the job.

Here, complements of Wikipedia, you can see how a Geneva wheel mechanism actually works.



I wrote a script in Visual Basic .NET 2008 to produce the outline of a ten step Geneva Wheel instead of the four step one shown in the Wikipedia graphic.  I read a file generated from that script using a script in Art of Illusion and created a solid object of the outline.


From there I generated the resulting gcode using Skeinforge.

























I will post the rest of the mechanism to this blog entry as I get it done.


Friday, December 18, 2009

Brute force stepper controller



Back in 2005-6 Reprap was using the venerable Pic 16F628A and was controlling steppers with the old, but reliable SN754410 chip.



The SN754410 is a very nice little chip.  It is DIP, easy to use and very reliable.  The only real problem with it was that you couldn't run more than 1.1 amps on one of them continuously.  Simon McAuliffe, a Reprap core team member from its modest beginnings in 2004 solved the capacity problem by stacking one on top of another.  While this worked, it created a rather nasty soldering problem for people like me who lack good fine motor control.

In those days we were using stripboard, known in the UK as Veroboard after it's original source.  Stripboard is a wonderful development system for DIP chips.  Unfortunately, it was hard to get in the US, a Velleman Euroboard costing anywhere from $6-15.  As time went on, Reprap went more and more over to designing boards with Eagle and sending them off to lithography shops.  Once there, it was only a matter of time before surface mounted chips began to creep onto the boards.

Surface mount is a whole other chip packaging technology which caters to automated circuit board production.  It is a bit difficult to make surface mounted chip boards, especially for clumsy oafs like me.  As a result, I stayed with DIP technology, suffering considerable derision by other Reprappers as a result.  My point was simple, however.  I felt and still feel that if a  bright 12 year old can't master the technologies and techniques needed to make a Reprap machine, we're not going to see viral diffusion of the Reprap technology.

Last year, I found a cheap supply of stripboard in the US.  Once I had a cheap supply of stripboard, I looked around for a nice, open source design tool for putting circuits on one.  Unsatisfied with what was available, I developed my own.

Last week, having got a grip on how to run my Rapman 3.0, I decided to look at developing the controls for a next generation stepper controller.  I ordered a bunch of the cheap stripboards...



...to make sure that the web advertisement wasn't a fluke.  The boards are quite high quality, made in Taiwan.


Once I had those in hand, I decided to leverage the work I'd done with I2C coms last year and design a high capacity stepper controller board.  To keep things simple I took a page from Simon's book and decided to simply keep adding SN754110 chips till I had the amperage capacity I needed.  I wanted the design to run a NEMA 17 {I own about a dozen} that I'd bought at the beginning of my involvement in Reprap.  It draws 2.4 amps, so I needed three SN754410.  Here is the design I've developed so far.






While I've configured this board for 3.3 amps capacity, it will easily seat another 2.2 amps of capacity.  The circuit design is extremely repetitive.

I got the thing built and the circuitry debugged last weekend.



It has been a while since I built up a board, so I blew up a few chips until the drill of checking every circuit trace for continuity and checking again and again for shorts caused by sloppy soldering came back to me.  I found that putting a washrag over the board when I powered up contained the fragments of exploding SN754410 chips.

In testing, I have been able to get a small, half-amp tin can stepper to run at speeds of up to 830 Hz.  I seem to be able to run the NEMA 17 at about 4 KHz at full step.  Mind, that is an estimate from my timing cycle and is probably on the overconfident side.  I still haven't found the bug in the firmware that is causing problems with directional control.

In passing, the three SN754410 controller chips stabilise at about 90 C when running the NEMA 17.  I haven't decided whether I will glue heat sinks on top of the controllers or bring the temperature down by adding another controller chip.

Wednesday, December 16, 2009

God jul!




Jul på laboratoriet



Jag tror att Gud inte tvingar oss att tänka på honom, bara att vi tänka. Ändå tror jag att vi aldrig får glömma sin son som dog för oss.

Monday, December 14, 2009

Getting overhangs right



When I bought my Rapman 3 I ordered two extruders and an extra extruder barrel drilled to 0.3 mm. When I assembled the extruder, I thought I had picked up one of the two 0.5 mm extruders. Instead, I had got the 0.3 mm one.

Oddly, I hadn't noticed the difference. The Rapman 3 extruder's filament pump was more than powerful enough to force 0.5 mm flow rates through my 0.3 mm extruder orifice without complaint. Thus, I was able to print what appeared to be 0.5 mm objects ... until we got to overhangs. When the extruder got to the overhangs it literally squirted the hot plastic into the voids festooning the print with what looked like plastic tassels.



I'd have probably never figured it out what was wrong save that my extruder orifice jammed.  When I cleared it with a 0.25 mm strand off of a multistrand wire, I noticed for the first time that the ABS being extruded looked a bit thin.  In fact, when I measured it, what was coming out was 0.38 mm in diameter.  There was no way a hot filament that thin was coming out of a 0.5 mm orifice.  That was when I realised that I had put in the 0.3 mm extruder barrel in error.

Knowing that, I reduced the flow rates and layer height to 150 and 0.25 mm and the overhang problem magically went away.



The new print is, of course on the right.

A problem with overhangs



I don't know what is going on. When my Rapman/Skeinforge system prints horizontal holes in objects they typically come out looking quite good. When I try to do an overhang, however, things start to go wrong.

Right now I am printing out the parts for jdavis' Sarrus linkage. There is a spine segment amongst the parts that looks like this.


The center span has what looks like a relatively easy span.  When I go to print it, however, it comes out looking like this.




I am pretty sure that there are some settings in Skeinforge that cure this up. Does anybody know what they are? Thanks!

Wednesday, December 09, 2009

Segmented aprons work with ABS



I was lazy. I didn't even bother regenerating the gcode for x-carriage-lower_1off using ABS assumptions instead of HDPE assumptions. The first time I ran it just after resetting the Rapman 3 for ABS I didn't get the extruder orifice close enough to the print surface and the raft peeled away after a while.





I reset the z-axis and got it right the second time.  Hint:  The first level of raft extrusions have to be 2 mm wide or it will peel.

I'm too tired to do anything more than wait for this print to finish.  I'll finish the blog entry in the morning.



Okay, so I wasn't too tired.  HDPE on the left, ABS on the right.

The point is, however, that segmented aprons pretty much completely suppress warping in the Mendel x-carriage-lower_1off part.  Using ABS gives you much finer detail than HDPE does for a 0.5 mm orifice.

z-leadscrew-base-bar-clamp_2off



I treated Mendel part z-leadscrew-base-bar-clamp_2off in the same way, viz, a segmented apron. You can see what this looks like in Netfabb.





As printed, the part looked like this.





Cleaned, it looked like this.





There was a very mild bit of curl on the right-hand edge {~0.5 mm}.



There was a small problem in that the near side of the part tried to delaminate.  The part has two horizontal holes that run the length of the part.






I think, perhaps, that the settings that I am using with Skeinforge didn't get me a proper fill there.  More to the point, as I learned eons ago as an architect, it's much easier to design a part than it is to design a part that you can build.  For HDPE, it seems to me that if this part was maybe 4-5 mm wider leaving the longitudinal holes in the same place, this delamination wouldn't have occurred.  Mind, it may not be a problem with PLA and ABS.






Other than that, no problems.

If you want to play with this part associated with segmented aprons, you can get the STL file here.

Several people have complained that they can not get the x-carriage-lower_1off part to print in ABS.  I've rerigged my Rapman 3.0 for ABS and am attempting to print it now.  I hate the stink of ABS.

Tuesday, December 08, 2009

Got it!



I'd heard that people printing Mendel were having trouble printing the largest pieces of Mendel in anything but PLA. Since PLA filament is hard to acquire in the US at this point I thought it would be nice if those hard-to-print parts could be printed in some other plastic. Since I am working with HDPE right now, I decided to have a go with HDPE.

The trial part I've begun with is x-carriage-lower_1off. I've discovered that a segmented apron positioned 0.5 mm away from the perimeter of the part does the job of successfully suppressing warping. Here you can see the spacing between apron and part.





Here is the completed print.





I've cleaned off a bit of the strings.  I understand that the newest Rapman firmware release pretty much gets rid of strings.  I haven't taken time to install it yet, though.  There was only one instance of corner curling.





I've circled the corner in red.  The curl amounted to about 1 mm.  Belt sanding leaves you with a flat bottom to your part.  The fill was raft material.  I got the part out of the the apron and off the raft with my belt sander.










If you want to print this on your on machine you can download the STL for the part combined with the segmented apron here.  I made no changes to the part save adding the segmented apron.

I had to rotate the original part to work on my Rapman 3.0.  You many have to rotate it again to get it to work on other Reprap machines.  I would appreciate hearing if this approach works for ABS.  I have ABS here and could try it, but I want to see if I can apply the same technique to do what I understand to be the other big Mendel part, viz, z-leadscrew-base-bar-clamp_2off.

Monday, December 07, 2009

Printing the big bits of Mendel



I've been hearing that people are struggling to print the largest parts of Mendel in ABS, so I've decided to see if it can be managed in HDPE using variations on my "no-sweat" method.

I tried Sunday with a full apron around the "x-carriage-lower_1off " part.  The print failed for several reasons, the most important of which was that the apron in shrinking mechanically connected with the print in the y direction  As well, having a full apron around such a big object added to the contracting force put on the raft by the printed object.  That force caused the raft to break free from the polypropylene print surface.

This morning I broke the apron up into four parts, not unlike what I was doing when I was doing guying and flanges a few weeks ago.  You can see what I've done with a screen grab of Skeinfold's Viewer.





Just some interim pictures.





Breaking up the apron so that I didn't have long stretches of apron adding to the stress that the print was putting on the raft/polypropylene interface seems to be helping.





I've got way past where I was when the print failed last night and so far I haven't heard a single one of those little crackly noises that signal that the raft is beginning to pull away from the print surface.

Okay, I stopped the print.  We have a qualified success here.  I could see a few things going wrong so I stopped the print.  Here is what it looked like in the printer.





The major problem, as you can clearly see, is that the final little flange on the left was melting down.




I didn't set up Skeinforge for a cooling cycle.  I am still getting used to how to do that, so I am going to just print the flange and test settings till I get something satisfactory.

The major problem I had was that when I moved the apron segments away from one end of the print I got them a little too far away.




The lift on this corner was 1-2 mm on the near side {circled}and ~1 mm on the other.  Looking at the other end.





You can see that the left side {circled} lifted about 0.5 mm while the right side was perfectly flat {arrow}.  I got a little over-ambitious about moving the apron segments away from the print.  I can fix that with zero trouble.

This is going to work.  It's down to a matter of just tweaking the apron position a bit and sorting out some Skeinforge settings.

I expect that the combined STL for this can be used with good effect in ABS, since ABS is reputed to warp much less than HDPE.

Saturday, December 05, 2009

Killing flocks of birds with one stone



I've got the curling on HDPE prints down to something quite manageable. Last night I decided that I could dress down the remainder if I only had a belt sander. A little while ago, I went into town and bought a nice little Makita belt sander. It wasn't the cheapest, but it would lay, belt-side-up, quite firmly, which is what I needed.

It took about 15 seconds to grind the raft off of Bogdan's corner block for Rapman and dress off a few other rough places.



I bought a new little drum sander for my Dremel {~$4.50} and cleaned up the seating for the z-axis bearing.



It also did a nice job cleaning up the hole for the z-axis threaded rod.




One added benefit which might be the Rapman or might be the HDPE or might be the Skeinforge settings is the roundness of the horizontal holes in the print.






None of this teardrop nonsense.

I'm beginning to think that HDPE is a very serious contender for printing Mendel parts.  It's dirt cheap, readily available, strong, doesn't make nasty fumes AND now we know how to work with it.  I expect that polypropylene, which is cheaper still is going to be just about as good.

Now here is the serendipitous Christmas present.  That Makita belt sander grinds HDPE a treat and puts it in a little bag, or a big one if you want to sew one.





Guess what?  The problem of grinding plastic so that it can be recycled into filament has just gone away.  Virtually any kind of extruder can eat plastic powder.  It takes a heftier one to use 3 mm pellets or shreds of a similar size.

Build yourself a slope-sided hopper on top of that Makita and your grinding problems are solved.  When you wear out your belt, go down to your hardware store and buy another.  The Makita uses a 3" x 18" belt and it costs a bit over a dollar.  No difficult-to-sharpen, never mind dangerous, macerating blades.

God Jul, everybody!  :-D


Can we print belts?



Yes, if we step back and look at what a belt does instead of simply trying to replicate existing ones.

Belt driven Reprap machines have always bothered me. They're fast and now that the price of NEMA steppers is getting reasonable, they're even becoming economical. It bothers me, however, that the belts and sprockets can't be printed very easily.

Last night I was cleaning up a corner block that I'd printed and was cleaning out a bearing seat with my Dremel tool fitted with a cylindrical abrasive head. These are a nice little fitting that lets you slip what is a cylindrical piece of sandpaper onto your Dremel. They are cheap and very effective.



I was looking at it when I realised that the toothed belts are toothed to prevent slippage and that there are other ways to prevent slippage than teeth.  An abrasive cylinder, for instance.  With an abrasive sprocket you wouldn't need teeth and the belt wouldn't need teeth either.

It was a matter of a few moments to design a 600 belt in Art of Illusion that would easily fit on my Rapman print stage.




A few minutes more and I had it run through Skeinforge and printed it in HDPE.





Cleaning the belt took a few minutes, not no big deal.




The resulting belt is very flexible and quite strong.




I've got a pile of NEMA 17s.  I might just print myself out a Mendel now.  Mind, I will be using my OWN electronics.