Saturday, September 30, 2006


The Borg Updates

I finally got some time for my baby, and I think we made some good progress. We assembled the X and Y stages (no motors yet, though)

I'll put a picture up here, but I've been checking out the Instructables site and I think that is a much better place for this sort of thing. I've put up two different tutorials on there. One on building the frame, and one on the linear stages. They are located here and here.

Also, I created a RepRap group on there if you guys want to join. I really think the people on that site will digg RepRap and I bet we could get other people building their own machines too. I'm starting to think that we shouldn't wait until v1.0 to get the word out. The more people working on the early version, the faster we'll have an awesome v3.0.


Tommelise sliding thrust collar under construction...

I reached the conclusion that if early generation repraps are going to be made with ordinary threaded studding we'd better get used to the fact that when you buy it from the hardware store it often isn't tremendously straight. I got the piece that I was using for Tommelise's z-axis (horizontal) straightened to a point where over 640 mm it had about +/-2 mm play. There are two ways that I could have dealt with that and kept the accuracy on spec. The first was that I could have had thicker guide rods (I used 1/4 inch) and used full sleeve bushings. That would have let the drive motor straighten the threaded drive rod. That, of course, would have eaten up torque. The other way was to let the thrust collar move freely in the x-y plane. I decided to give that one a go for this design exercise.

I designed a crude thrust collar which would let the coupling nut move in the x-y plane. This consisted of a poplar sleeve that let the coupling nut move freely in the y direction and guides, also done in poplar, that let that sleeve move in the x direction.

The thrust collar had to more or less match the width of the coupling nut (1 in) which entailed me gluing two strips of poplar together. You can see this assembly drying in the first photo.

At that point I carved seatings for the guide rods out of poplar and glued them to the 320x320 mm double strength glass work surface.

In the second photo you can see them glued in place along with the guides for the thust sleeve. Fender washers keep the coupling nut from slipping along the z-axis.

I got a bit overexcited and tried to run it this evening instead of letting the glue dry overnight. The thrust collar appears to do what I'd intended, that is, to keep lateral movements by the theaded drive rod from reaching the guide rods and requiring dissapation as friction heat.

I was able to run it for about a minute before one of the glued joints came loose. I reglued everything and resolved that video clips of the z-axis platform will wait until tomorrow morning. :-s

Thursday, September 28, 2006


A bit further...

I got a 2x4 foot piece of 3/4-in fibreboard from Orchard today and sawed it down to 2x3 to fit on my work table and serve as a base for Tommelise.

I also shortened down the span of the z-axis to 650 mm as you see here.

I bought a coping saw, which you can see in the photo, so that I can cut some small pieces to make mounts for the shaft encoders for the GM8.

Just as a test I slapped the current edition of Wired magazine (~425 grammes) onto the coupling nut to see if the GM8 would have any trouble moving it around. You can see a short video clip of the GM8 not having any trouble.

About the only obstacle to progress is straightening out the threaded rod to limit lateral movement that you can see happening in the clip.

I decide to take things a step further before turning in for the night and strapped the 6v GM3 motor onto the rig since it was already equipped with a shaft encoder and controller board.

Duct tape is a wonderful prototyping tool. :-)

The pseudostepper control regime for the GM3 gearmotor worked fine. Here is a small video clip of the running at 0.484 seconds/step with a 12 degree step size. With this thread pitch that works out to just over 0.03 mm/step.

I'm currently running it at 4 seconds/step and 22.5 degree step size. That works out to 0.06 mm/step.

Wednesday, September 27, 2006


Tommelise takes shape...

I'm beginning to get clear of the billable hours that I'd agreed to so that I have a little bit of time to think about Tommelise.

The first obstacle that I hit was the discovery that the GM8 gearmotor has mounting holes too small for any readily available bolt. Neither M3's or 3-48's are small enough to fit them so it looks like I have to get creative with ideas about how I mount the motor and encoder chip. At least I have a motor that otherwise makes that possible.

The first thing I had to do was bite the bullet and take down Godzilla.

I realised that having to do that was at least partially responsible for the fact that I hadn't done more work on Tommelise earlier.

Oddly, once I'd taken it down I completely lost any sentiment I had for it and started looking at it as a source of poplar and parts for Tommelise. Doing that should save me a bunch of trips to Orchard Supply.

After a lot of dithering I decided to go with the Reiyuki/Luberth/CNC approach.

I laid out a basic horizontal z-axis.

And then checked to see if the 12v GM8 would turn it.

It did with no effort whatsoever.

At this point I'm using c-clamps to hold things together. I haven't cut the 3/8ths inch threaded rod down to 2 feet yet and won't till I decide how I'm going to integrate the x/z-axes. Mind, I know what it supposed to look like from Lubreth and Reiyuki's models. Getting the clearances sorted out and the rest integrated is what I'm not completely clear on yet, though.

Sunday, September 24, 2006


VRML Browser plugins...

I spent a little time yesterday exploring the export of VRML (.wrz) files from AoI. If you turn off the file compress option (which is a default in AoI) when you export to .wrz format you can do a 3D image manipulation on your browser (I'm using Firefox) using the Flux Player VRML browser plugin pretty nicely.

Unfortunately, the blog we use won't allow you to upload .wrz files for display. :-(

Saturday, September 23, 2006


Magnetic Mockup

Yes it's still the mockup. I'm 200km away from my workshop, but I brought the mockup with me to experiment with in the meantime.

The magnetic joint concept appears to work. The magnets hold the pieces hard enough that you can pick it up by the top, wave it around, whatever, it won't budge. Try and pull the top off and my ridiculous structure will disintegrate before the magnets let go. The points are quite hard to drag out of place as well, even when just sitting on the magnet as pictured. But they pivot without resistance when a piston is moved.

Keen-eyed viewers will note that I'm not using the rod-ends pictured in my previous post. Those turned out to be stainless steel, much to my suprise. I'm using ordinary nails.

Friday, September 22, 2006


Good news about CAPA...

From Adrian...
I've just spoken to the MD of Solvay in the UK, one Stuart Derbyshire. 
He, of course, regards this development as unalloyed good news, and I couldn't
help but be amused along with him.

He says that they have every intention of stepping into the breach long
term, and no plans to increase prices significantly short term. As he pointed out,
we will need tiny quantities over the next few years, which they can easily supply.
And if RepRap take off, then Adam Smith will ensure things are OK after that too.

I don't think we need to worry.

In addition, I do intend to put a student on making and using PLA in
RepRap after Christmas.



Thursday, September 21, 2006


Something new for the reprap builders...

Right now, communications at RepRap happens in three modes...
The SourceForge forum began when the email roundrobin began to generate too many posts in the developers' mailboxes. While the SourceForge forum has been useful I've found it limiting in a few ways. The most important of these are...
I've been participating in a tiny current affairs forum ( for years. For most of that time it used commercial forum support software. In March, however, Vlad switched over to an open source forum support system called phpBB 2.0. While I don't post at LandV too much since joining the RepRap effort I do glance over the postings from time to time. The point of all of this is that phpBB 2.0 seems to be quite a full-featured and robust system ... and it's free!

Anyhow, my son got our local server going again and I've reactivated my old boutique technology atlier domain,, again. Don't go there looking for anything just yet because I haven't got the site operational yet.

With a little luck my son will also be setting up phpBB 2.0 for me today or tomorrow. When that's operational I've thought of offering a phpBB 2.0 forum for all you RepRap builders to exchange information on as an experiment to see if it works better for them than what we currently have at SourceForge.

Adrian has agreed to link us up from the main RepRap website. Give us 4-5 days and it should be operational.


CAPA Supply disruption!

Heads up!

I just got a note from Jeff Neidinger at Solvay USA.

Dow has just announced that they are getting out of the business of making caprolactone polymer at the end of October.

It's not clear whether they are selling their production unit in West Virginia or just converting it to making some other polymer.

What we are looking at is a serious supply disruption which is definitely going to affect delivery times and might affect price.

Jeff says that Solvay is going to need 6 weeks lead time on Caprolactone from now. All of Dow's customers are keeping his phone ringing off the hook.

This is bad.


Server back up!

As you know the server I was keeping my share folders on went down a few weeks ago. It turned out that we had some serious hardware problems along with the software failure that we experienced on the old server, so yesterday my son turned the desktop that I retired last Spring into a new server. It's working fine.

With that in hand I reactivated the domain for my atelier and moved the share folders into it for the new server. You can access them at...
I haven't got everything in the folders that belongs there yet. Give me a few days. :-D

Tuesday, September 19, 2006


We have a deal on filament!

We have a deal on having CAPA filament for our Mk II's made. Image plastics...

...has offered to make short production runs if we ship them the polymer (one pound and up) for $3.75/lb. I'm making arrangements, thanks to Adrian and especially Sebastien, to have 5 kg made up into 700 metres of filament for testing with the Mk II extruder.

Monday, September 18, 2006



My latest flurry of activity ends with one dozen 2" pieces of 3/16" steel rod, ends rounded off for use in the aforementioned ball-and-socket joints. They were mounted in the chuck of the milling machine/drillpress and filed to shape, then polished with a sharpening stone.


Trimmer line progress (or lack thereof)

Fired up a hotglue gun and a 35watt soldering iron and tested the trimmer line against them;

The trimmer line got a bit sticky after a few seconds, but after 30sec, still not fully liquified. It had the consistency of marshmallow fluff.

With the soldering iron test, the line was basically pressed against the body of the iron. The line immediately turned mushy and melted, however it also started emitting plenty of wierd smoke. After removing the line there was also a bit of black residue on the metal. I'm not sure if this was because the iron was too hot or because of the trimmer line's composition (pvc fumes?).

At any rate, I am definitely liking the idea of using alternative, higher temp materials in an extruder. Soldering iron element can easily break 200degC and can be mounted without too much effort to an extruder barrel. It would give us a bigger variety of source-material, whether CAPA or recycled material.

Here's some possible sources of polymer with unknown melting points and composition. If anyone knows the composition and whether any of these will make viable sources, please respond:
Lawn Trimmer line (still testing)
Styrofoam / Packing peanuts
Plastic shopping bags (lots of them)
Plastic cups / dishware (many varities)
Plastic bottles (water, soft drink)
Melted-down junk? (computer equipment plastics

Maybe it's a pipe-dream, but the idea of being able to recycle 'junk' and turn it into usable parts/products sounds pretty inticing.

Sunday, September 17, 2006


Ugly Stewart Platform

It is built from six 5cc syringes, on platforms of cardboard, attached to both platforms with "universal joints" made of thick copper wire loops. The large platform is a 13cm equilateral triangle, the small platform is a 7cm equilateral triangle.

Needless to say I do not expect to build a functioning reprap upon this. This is a mock-up to see if I had anything close to an understanding of what I was trying to build. For something made of cardboard and held together with wire and masking tape, it is impressively stable! Precisely what I'd hoped for from a Stewart Platform.

The real version will be made with 12cc syringes, on metal or wooden platforms, held together with magnetic ball-and-socket joints. I already have all the freakishly strong magnets I'll need for that, obtained from dead hard drives.

More pictures here and here.

Saturday, September 16, 2006


The Imperial Menace

Despite being a country that converted to Metric decades ago, Canada's close trade relationship with the States prevents it from making much headway. Walk into a hardware store and everything's Imperial. Unless you feel like paying for overseas shipping, most everything you mail-order is Imperial too.

So I'm making do with Imperial parts, materials and measurements and leaving a lot of breathing room for mistakes. I go by two rules.

* Drill holes too small, enlarge them later if you need to.
* Get rod too large, lathe it down later if you need to.

From left to right, we have:

* 2" of 1/2" PTFE rod, with a 1/8" bore all the way through, and a 1/4" section of of tapped 1/4" on the end.
* 2" of 1/4" threaded brass rod, with a 1/8" bore all the way through.
* A 1/4" brass acorn nut, drilled out to serve as a nozzle. Note the solder holding it together, I machined a touch too far in and made a crack in it.

The brass rod is teflon-wrapped and fitted with nichrome wire the traditional way. I had to get special plumbers' tape meant for gas pipes, though -- ordinary stuff wasn't pure PTFE, just "teflon-impregnated".

Thursday, September 14, 2006


0.1mm Filament

I mentioned the idea of using lawn trimmer line to a coworker. He said a friend of his does a lot of jewelry crafts and they use extremely fine monofilament for some things. I poked around on the web a bit and found that it's possible to get 0.1mm (.004") nylon monofilament in large quantities relatively cheaply. I saw spools of 3000 yards going for $6. (Anyone care to work out the mass of that?) I wonder how practical it is to use nylon monofilament of that diameter. Wouldn't take a lot of heat to melt it, I would think. Not sure if it's practical to feed 0.1mm nylon, though.

Wednesday, September 13, 2006


Buying Filament

TylerM has done us a major service this morning. He linked us to a polymer welding rod firm down the coast from me at Santa Barbara that provides polymer filament for a variety of plastics.

I called down there just now to see if I could find out who was making their filament for them. Turns out that they are making their own filament .

I talked to Robin there and she indicated that they had their own filament extrusion machine. I asked her if they would be amenable to making short runs of filament with a non-standard polymers. She said no problem if we could source the filament for them or tell them where to get it.

She is supposed to call me this afternoon with some ideas about what sizes of production runs they'd be willing to do and how much of a surcharge on the basic polymer cost they'd be charging.

Santa Barbara is just under 300 km south of me. It is the university and technology town where my son attends university. I'll be moving him down there for his second year in two weeks and have already made arrangements to visit Abbedon at that time.

This is super!!!!



I saw this thing in a local craft shop yesterday. It's a device to cut patterns in pieces of paper for scrapbooking. My first thought was, "I wonder if it could cut copper traces from adhesive foil." The website makes a big deal about not having to hook it up to a computer. In fact, it looks like you can't attach it to a computer, which means you're at the mercy of the company to provide new patterns. Doesn't seem like much of a selling point to me. I wonder if it could be hacked.

Tuesday, September 12, 2006


Javascript for stepper motor wiring

A Javascript routine to determine stepper motor wiring. (via Make:)

Monday, September 11, 2006


Making filament the easy way?

I've been pondering on a way to make filament hands-free and here's what I came up with:

The left is 2-3" PVC pipe capped at both ends. Inside is a heating element, probably soldering iron or nichrome. The top has a bike pump nozzle for pressurizing, and a brass nozzle attaches out the bottom at a 45deg angle. The filament drops into the coldwater bath, and once cooled, gets bent upward to a cable spool, which is PIC controlled for a preset level of tension.

I'll bet it still needs tweaking, but it might be a way to easily extrude large amounts of 1-2mm filament.
Any thoughts on how to improve on this design?

Sunday, September 10, 2006


GM8 Coupling

Once I'd modelled the drive shaft developing a coupling for the GM8 and GM3 was easy. This one is pretapped to be threaded for a 3/8-24 threaded rod.


Weekend update Mothra

Looks like Mothra and Zach's Borg are almost neck-and-neck. Well here's the final update for work done this weekend.

X and Y motors are mounted (NEMA 17's) and appear to be good. I used Plaasjaapie's polymer coupling idea which worked quite well. Current accuracy is pretty bad, but the guide-rods are still unglued and swaying, so accuracy sways ~2mm sideways. A little work should clean that up and bring it back under .5 at least.

I am a bit concerned at the torque needed to pull the Z axis up and down. This might have to be gearmotored using Plaasjaapie's ongoing experiments. A nema 23 is out of the question.

Dremel tool is mounted for the time being, and I'm just barely starting work on the electronics.


The Borg Becomes Stronger

Today I got some more work in on my model. The main focus was strengthening it and making it more stable. First, I added diagonal cross bracing to add strength. Then, I fixed the error that I had made last time. I had made the boards only go long enough to attach to the vertical pieces. I didnt think it out well enough so the horizontal boards didnt come together to form a square. When I replaced the side boards with longer ones, the strength of the thing increased quite a bit. It is really sturdy now and I think will be able to stand up to quite a bit of action.

Secondly, I attached the mid level supports where the Y/Z stages will eventually be mounted. Since the unit was designed with the dimensions of the rods that I have, they didnt actually have anything to rest on, so I attached some 1x4's to the inside. In the picture, that is what the rods are resting on.

This Wednesday, my friends and I are going down to the surplus sale to see what useful things we can find. Theres a good chance we'll be able to get a copy machine to gut for parts, along with all the other strange assortments of things that are always for sale at surplus. Hopefully we'll make a good find!

Saturday, September 09, 2006


Just some hot air...

Here is an idea I'm working on for very small RepRapping. To give a sense of scale, the green cube is 1 mm cubed, and the little blue smudge is 0.1 mm cubed. That would be 100 micrometers (previously know as microns) wide, I think. Getting kinda small...

I'd been thinking of shrinking the MkII extrusion head and it seems to me that smaller nozzles would be more and more prone to clogging. The CAPA has to be manipulated in a home workshop setting where dust contamination could occure easily. Also from what I can understand high viscosity extrusion at smaller diameters demands high pressure. Wouldn't a doubling of extrusion rate at a given diameter imply a quadrupling of pressure? Combine both and we need too switch to alternatives from the screw drive filament feed.

The rendering shows a very fine filament of 0.1 mm in diameter, (thin vertical white line) formed beforehand and being fed through the nozzle (brass pencil thing) by two pinch rollers. Since 0.1 mm diameter filement is so small, there may be no need to use anything more than a hot air source (the strange silver cone thing) to melt it at the last possible moment when it comes into contact with the workpiece. Maybe the trick is to leave a good gap between the nozzle tip and the work surface. I'm thinking at this scale the filament should be acting like a stiff bristle, so there may be no need to guide it right down to the surface. This would give room for the heat source to melt it just at the point of contact, no sooner. (If hot air seems too low tech, how about a DVD-R laser?)

This approach may be more linear when it comes to increasing deposition speed. As long as the software and hardware matches the head's speed of movement with the feed rate of the filament the concept could work. The hot air flow would have to vary along with it, the faster filament feed, faster air flow, and it could all stop on a dime.

Hot air also happens to be used to melt solder, and this also suggests it might be possible to feed a filament of solder instead of CAPA. (Thanks to Rieyuki for that idea) Same head, just a different filament and suddenly you're reprapping electronics.

Finally, if this idea could be pulled off at a given size, it seems it could be shrunken down again without really having to change the fundamental concept. Once it's working, how small could it get?


Volumetric Model of GM8 Gearbox...

I spent the evening with calipers trying to get the better of the Solarbotics GM8 gearbox.

The green dots at the right edge of the gearbox are bolt holes.

If anybody needs a copy of the .aoi file of this let me know.

The point of this exercise is to design a mounting for the AS5035 shaft encoder chip and its magnet. Here you can see the chip in grey atop the yellow magnet. There's not a lot of elbow room for this sort of thing with the GM8 geometry. As I see it at the moment the problem is largely going to lie with the securing of the chip itself to a mounting block.

Here is a mounting block being designed. The green rods are positioning holes for bolts and the red ones feed holes for filaments leading from the AS5035 chip to the top of the mounting block. You can see the AS5035 chip embedded in the block.

Here is the mounting block seen from the top. The red rod with the rounded tip is to create a slot for the comms cable connecting the shaft encoder chip and the control board.

Finally, you can see the lid in yellow that locks down the filaments and comms cable.

Here are two exploded views of the mounting block and the lid.


Motor spinup

Well, the motor spins.

I wrote a simple program that spins the motor one direction for a second, then spins the motor in the other direction for a second, and repeats infinidum.

Unfortunately, the motor only wants to spin in one direction.

If I reverse the voltage manually by connecting the motor directly to the voltage inputs, it works.

If I remove the motor from the H-Bridge, and measure the voltage there, it's reversing the voltage.

But, if I connect everything together, the motor pulses every second, but continue to spin the same direction. If I measure the voltage coming out of the H-Bridge, it's magically no longer is reversing every second.

Something from the motor must be getting back into the input electronics. I'm not sure what -- the power on the motor is pretty isolated (H-Bridge runs on 5V, the circuit driving the H-Bridge runs on 3V, and the motor runs on 12V). The capacitor on the motor should be soaking any back EMF, so I'm just not sure what's going on at this point.

I'm going to need to solve this problem before I start in figuring out how to get the encoder hooked up.

It also appears I've fried the NAND gate that triggers the enable ports on the H-Bridge. I must have shorted something trying to debug the voltages. After this, I might just decide to avoid the NAND gate altogether, and just drive the EN pin with it's own output from the CPU. This means I'll only have enough ports to drive 3 motors. Seeing that there may be better solutions anyway for synchronization, I think I'm going to go back to the one CPU per motor, more inline with standard RepRap.


Mothra rises to face Godzilla

Full-days work today, but I made a lot of progress, as you can see:

For linking each platform to it's threaded bolt, I carefully smashed a 1" coupling nut into its guide-hole

At first, the X platform was very tight. Wasn't sure exactly what was wrong, but I wiped it all down with WD-40 and it worked perfectly. FYI, a power drill is perfect for doing end-to-end friction tests :-)

It's definitely a pain to get those rods perfectly in-tune with the sliders. I actually drilled slightly larger holes on one side so I could have some room for play. Once I decide it's perfectly aligned, I'll squirt some hot glue or polymer into the hole to tighten it. Hot-glue actually has some bounciness and might even help lower vibrations.

While I fully intend on making this a polymer extruding repstrap, it will be a lot easier for the time being to just strap a dremel tool to the end for testing. I'm still pondering different extruder designs, whether peristaltic or psi pumped or standard filament driven.

Motors still have to be mounted/tested and I still have to solder all the control boards, including the JDM programmer. So while it's still quite a ways from being finished, it is a good example of building a repstrap frame on a budget.

Any questions on what tools/methods were used in making this, or requests for more pictures, or anything like that; feel free to ask.


Making gearmotor couplings from scratch for Tommelise...

The Solarbotics GM gearmotors have a sturdy but very short (5.4 mm) plastic drive shaft. Obviously, Adrian's coupling design for the Mk II would work beautifully. Unfortunately, I'd have to redesign it for the GM8 then ship the specs over to Adrian and then wait for a few weeks for them to be made on the Stratasys and shipped. I'm just not patient enough for that way of doing things, so I invented another coupling.

First, I took a length of 1/4 inch round rod and used my old Demel tool with a grinding wheel on it to carve two flats onto the end of it leaving the same width between the two, 0.196 inch as the drive shaft on the GM8 has.

Then, I took a coupling nut for the threaded rod I'm using. In my case using 3/8ths inch threaded rod is wonderful in that it is bigger than the 7 mm drive shaft. I used a standard 3/8-24 bolt to plug the bottom, filled the top with CAPA and popped in in my toaster oven at 300 degrees F.

After about ten minutes the CAPA had gone transparent and slumped, so I added more and popped it back into the oven. I had to do that once more before I had the molten CAPA level where I wanted it.

At that point I inserted the makeshift mandril that I'd made into the molten CAPA and then ran cold water over it to solidify the CAPA.

I then checked for a fit with the GM8 and as I expected the slot was too narrow. I heated the mandril with my paint stripper (which is basically a hair drier {which would have done as well} on steroids) and widened the slot slightly till the GM8 drive shaft fitted properly.

The process seems to make a good coupling in about 20 minutes for the first one for about US$0.79. :-)

Thursday, September 07, 2006


More progress:

Exact dimensions and measurements of what's finished will probably be released this weekend, since this is starting to come together pretty close to my expectations. Now I just hope it dosen't need a whole lot of torque, so I can fit the relatively small stepper motors to it.


Construction Begins

I've begun work on my first Reprap, although I'm not sure what to name it yet. I went with a design that I feel is very basic, yet very workable. The basis for the design was that we (my friends and I) were able to scavenge some REALLY hefty smooth rods from a broken copy machine we got nearly free from our local college surplus. The design size was then based around the length of those rods. the basic layout is to have the extruder on the top of the frame on the x axis. mounted below that, about halfway up the frame we will have the y axis with the z carriage mounted on top of that. althought this halves the area i can print with the y axis, i still have a very large print area, at least 1 cubic foot, which should be plenty.

we constructed the frame out of more scavenged goods, this time spare lumber we got from old dorm lofts that people discard when they move out at the end of the semester. its a pretty basic design, and we havent mounted any of the cross bracing or the mounts for the YZ stage, but this should give you an idea of how its supposed to work.

Also, I just got a package from Sebastian in the mail that contains a whole bunch of CAPA. I hereby publicly accept your challenge of the race to first extrusion!


Beginning Tommelise

Now that I have an operational pseudo-stepper motor and working control board philosophy to power a small-scale RepStrap, I am beginning the Tommelise project. For those of you who don't read your Hans Christian Andersen in the original Danish, the tale of Tommelise is better known in English as Thumbelina. I've always preferred the original name over the English transliteration. Mind, the transliteration wasn't bad. Tommelise would have come out Thumblisa. The translator did a good job.

I have a lot of both material and experience from building Godzilla that I think will be useful in

I'm going to be staying with milled poplar and making more use of 1-1/2 x 3/4 inch cross section modules.

I've got 3/8-24 (0.95 mm metric pitch) and 1/4-20 threaded steel rods
(0.79 mm metric pitch) which should be adequate for the dimensions that I am looking at. Both diameters mate fairly well with the GM8 gear motors that I will be employing. I will probably use 3/8-24 for the z and x axes (horizontal) and the lighter 1/4-20 for the y axis (vertical).

Going with Reiyuki's CNC gantry philosophy should allow me to use the same axis design for at least the z and x axis and probably also for the y.

My first task this weekend will be to knock together a z-axis platform to see if the GM8 is going to be happy turning that 3/8ths inch threaded rod. I'm not too worried about that since the GM8 has a considerable amount of torque, but I'd like to be sure all the same.

Wednesday, September 06, 2006


Simple Shaft Connecting Cheats - "SSCCs"

This is a 3D render I made of some real world test parts I've fabricated.

The brass item at the top is a small brass pipe into which I inserted a stiff rubber hose. In the rendering I've shown it in cut away to reveal the two shafts embeded in it. It took a fair amount of force to shove the roughly ground down end of the 1/4 inch threaded rod on the left, and the ground down cordless screwdriver bit on the right side.

Note that neither is of the same size, the rubber hose acts as a very forgiving adapter. Also, note that the hose is shorter than the pipe so it can't catch on the edges. It will then be pushed deeper into the pipe from both ends, increasing compression.

Next we see a spring wrapped around the same rubber hose, not sure how good this is, as rotation direction may affect this one. Also the spring doesn't transfer torque quite as well as a solit tube, but it sure works better than plain hose. Maybe more cheating could help, like wrapping tightly with PCV tape.

That BTW is another quick trick, just tape the hose before inserting shafts, if nothing else is available. Alternatively, tape can be used to pick up the slack between the outer diameter of the hose, and the inner diameter of the pipe if they are not a snug fit. In a real pinch, rolled up paper works well!

Lastly we see a brass pipe split down one side with a length of hose inserted. Two small offset clamps and it can now handle a massive amount of torque compared to the original rubber hose, yet still be complient enought to forgive missalignments.

Although rubber hose works, neoprene and silicon hose may work better, as they seem to grip more. Cheaper clear vinyl hose will also do in a pinch, but it tends to slack off with time, so it's not the best option. You can tell it appart from silicon hose, as silicon is usually cloudy and kinda bouncy, like neoprene. Might be good for blocking the transfer of vibration.


Z8 Compiler - poor optimizations

So, I'm trying out the Z8 compiler to see how well it creates lean, mean machine code.

I'm not very impressed. While the feature set available is quite impressive, when it comes code optimization, the compiler appears to be at the baby level. It's going to be a lot of work if the code isn't quite fast enough, or a bit too bloated for these small microcontrollers. Or maybe I should just work on that forth compiler/interpreter ;-) Or.. switch to the AVR and use GCC. That would be very nice.

I've sent off a request to the support at Zilog. It will be interesting to see how they respond to the issue; -- is the customer service good or bad? I guess I'll find out.

Some of these things I've noticed:

1. The lack of any apparent ability to create normal inline functions. The white-paper indicates they left the 'inline' keyword out of the compiler to prevent code bloat. Unfortunately, leaving it out creates code bloat if you like to create lots of small self-documenting C functions. I may be forced to use ugly #define 'functions' just to get lean and mean assembled code; or perhaps, I'll have to switch to assembly. Yuck.

2. Including assembly instructions that have no effect, for example:

ADD R0, #0 ;; useless
ADC R1, #0 ;; useless
ADC R2, #0 ;; useless
ADC R3, #1 ;; ah. Finally. Something important.
;; Ok, why wouldn't you just use ADD R3, #1?


Progress on Mothra

Back from travel (business) with an itch to build

As you can see, the X axis is almost finished. I still need to figure out how to properly mount some bearings between the frame and the threaded rod, and then I'll probably try to mount the first stepper after that.

The Y and Z axes will need critical alignment with some careful cutting, so I'm gonna hold off until X is up and working.

Oh, and a reminder to any other builders, if you don't want to waste parts (like I did before making this version), follow the 7 p's:

Poor Proper Planning Precedes Piss-Poor Performance

More updates to follow, and hopefully after X is finished; exact cut, drill, and parts lists.


New Character Enter

The day I stumbled across my mind was seized as if by an army of robotic minions. I am so going to give this a shot. It is doubtful I can compete with the heavy-hitters here in quality or ingenuity but there's work getting done already.

Observe the rough-and-ready converted-drilling-press milling machine that didn't exist last week. And I can contribute... um... totally nonstandard imperial-pitch parts that nobody in their right mind would use. Oh, and one-piece acorn nut nozzles turned on a 3" lathe that's older than I am.

I really wish they'd kept the name Polymorph in Canada. One feels like an idiot wandering from craft store to craft store asking for 'friendly plastic'. Anyway. The Teflon, motors, nichrome, pics, and polymorph are in the mail. If I can build the printing head, I will know the rest is possible.


Reiyuki's design done with HDPE cutting boards...

I ran across this little proof-of-concept CNC machine design several months ago. Looking at it again this morning, I realised that it's functionally equivalent to Reiyuki's design.

It's fun to look at and gives you an idea of what you can do with materials that you can buy right out of your local hardware store. :-)

Monday, September 04, 2006


Plaas wins! Bumping works! :-D

I've made a GM3 gearmotor behave like a stepping motor.

Here's how you do it.
So far I've got it down to 4 rpm with a considerable amount of torque available. You can see the individual steps occurring at that rate. I suspect that I can make it go slower still.

There is enough torque at that speed that I can not stall the gearmotor shaft by squeezing it with my fingers. I have been able to stall the gearmotor operating at low voltages using PWM with little problem.

The step that I've achieved so far appears to be about 15 degrees. For a threaded rod with a 1 mm pitch that gives us a resolution of 0.067 mm.

I suspect that if I were to use the L298N chip and apply braking between steps I could get even more torque.

I've continued raising the wait period and got the rate down to just at 1 rpm in about 48 steps (7.5 degrees).


Using RS232

So, after researching a bit, I've discovered it will be easiest to send the RS232 signals thru a line buffer to condition it to voltage levels more suitable for the Z8. I've ordered some MAX232 chips to handle this. I'm hoping to get them around the end of the week.

I've also raided the electronic junkyard at my father in-law's, and was able to scavange 4 opto-mechanical mice. These will have the cords removed to be used for serial line comm (hopefully, this works), and I am going to experiment using the optical devices as shaft encoders; I'm going to wait until I get serial comm up first, before I do anything serious on that front.

In the meantime, I've started to explore some of the capabilities of the little 8K EZ8. In some ways, I'm pretty impressed:

~(lots) INT16 additions per second.
~125K INT32 additions per second.
~60K INT16 multiplications per second.
~20K INT32 multiplications per second.
~15K FLOAT32 additions per second.
~11K FLOAT32 multiplications per second.
~9K INT16 divisions per second.
~4K FLOAT32 divisions per second.
~3.5K INT32 divisions per second.
~500 ATAN() calculations per second.

It was actually a bit faster than I thought for the floating point, and a little slower than I expected for the INT32 divisions. All tests were performed on a 5MHz processor. I could add a 20MHz crystal and get effectively 4x the performance, but I'm going to avoid doing that right now, since I think I should have enough power to do what I want (I'll only be doing INT32 addition initially.) I couldn't get a good benchmark for the INT16 additions. My timings were based on a loop of 1000000, 100000, or 10000 operations, but even with the loop overhead, the 1000000 INT16 additions didn't seem to be adding anything over an empty loop that my rough measurements could detect (I.E, start the program, watch for the LED to light up to indicate it is started, then watch a second hand on a clock as the LED goes off to indicate it was done. Yes, very, very low tech performance benchmark timings here.)

The reason I decided to try this benchmark is I wanted to see how much of a hit I might take if I decide to try using ADC channels on an analog shaft encoder, rather than a digital shaft encoder. It appears I'm not going to be doing exact calculations if I go this route, but an atan lookup table with 255 entries should work and be accurate enough for my purposes.

Another purpose to my madness is that I've been thinking more about splitting the control out to separate Z8's. I should have plenty of horsepower to do some fairly good heuristics, even using floating point if i need to. My plan would be to use a master controller chip to synchronize everything. It will wait for an output from each motor driver to toggle to an "I'M DONE AND READY TO STEP TO THE NEXT POSITION", and when all motors are done, it will toggle the "OKAY, STEP TO THE NEXT POSITION" that will be sent to an input on each of the slave drivers. The nice aspect of this kind of system is that it will be as fast as the slowest component. No timing is necessary. I think I'm going to think a little more on this; If I decide to move forward, I'll probably blog some additional schematics (I'll include the power supply and serial line schematics with this..)


Quick Briar Patch design charrette...

I've found the Art of Illusion an invaluable solids modelling tool for seeing how pieces fit together when you are trying to make a concept take on a solid form. Here are a couple of screen grabs of my first stab at it.

I've made the working surface gear much larger than it would likely be in reality.

Adrian offers these two critiques of the concept...

You can't easily use it for outlining at the fast speed, as that ideally needs
to be continuous round the shape.

When you're infilling you don't just run back and forth. You have to plot quite a complicated joined-up zig-zag (see my blog of August 13) to avoid turning the extrude head on and off all the time.

For the first, I can't see how for outlining that it will be much slower than an ordinary cartesian machine.

The second, however, is much more serious in my mind. The meat of the critique is that you'd have to turn the extruder head on and off a lot. That's absolutely true. Also, given Vik's experience, the Mk II has a tendency to dribble which can make a bit of a mess of the end product.

Adrian has developed quite an efficient ploughshare method of doing infill. Taking this approach can't take advantage of it.

I had thought that it might be better to just do the infil for a layer first and then trace the outline afterwards.

I've been considering reversing the polymer pump just a touch as a way of turning the Mk II off in a way that doesn't dribble. You can't do that with the existing Mk II controller board. That's all untried at this point, though.

Anyhow, that's where I'm at with the idea at the moment.

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]