Monday, June 28, 2010
whoah, it works! (inexpensive dual z-stages for SLS)
As a quick follow-up to my last post, I thought I'd do a quick test of the dual z-stages without aligning anything just to see if the whole rig was in the ballpark of working.
It works! almost every nut in the thing unscrewed itself, but it works! :) and it's about 300 steps/mm with all the gearing and the screw pitch, so it's super precise!
Thursday, June 24, 2010
Dismantling the Extruder
Last night I took apart Kurzweil's extruder in order to replace the insulator.
First I liberated the whole hot end from the extruder in order to pull the insulator out. I ended up with this (apologies for my crappy cellphone photography):
Now I know when building this extruder James and I pushed the insulator all the way into the nozzle, but check out the plug of ABS that formed between the insulator and the smaller nozzle canal. My hypothesis is that at some point the nozzle got clogged and a little bit of ABS was able to squeeze between the insulator and the nozzle. Then it cooled, and the next time I operated the extruder, that little plug-seed got melted again, and some of the pressure that should have been sending the filament out of the front of the nozzle got put towards increasing that plug with new melted filament. All this is exacerbated by the fact that the insulator just above the nozzle was able to buckle and bulge out.
I was curious what that crazy bulge looked like inside, so we cut it open.
What my badly-focused cellphone picture doesn't show is that there is clear evidence of back-flow up the insulator around the filament, or in the vernacular, extruder FAIL.
The replacement we'll be using for the insulator is a thinner PFTE tube surrounded on the nozzle-side by a PEEK tube and on the other end by a similar tube of what I'm pretty sure is aluminum.
After getting the insulator out, I took the nozzle apart so I could redo it, (1)to ensure proper thermistor placement and (2) so I could use Kapton tape, which is easier to maintain than fire cement, as the rest of this post will reveal. I started by cracking off the fire cement layer by layer with some pliers. The Ni-chrome heating element survived, but the thermistor was too tightly sealed into the hole on the side of the nozzle tip, and I ended up having to just tear the leads off and dig out what was left with a pointy implement. Then I put the nozzle in a cup of acetone for a while to soften the ABS still stick in the very tip, which worked like a charm.
So now we are in the market for a thermistor. I looked all over, but nowhere seems to have all of these qualities at once:
- 200k ohm
- shaped like this =O instead of this --
- operating temperature range high enough for ABS (235-240C°)
- a response curve compatible with the preexisitng software on Kurzweil
- being sold in reasonable quantities (no, digikey, I do not want to buy 2500 thermistors!)
Hopefully the fellas at BfB will be able to help out, but in lieu of that, any tips would be greatly appreciated.
Monday, June 21, 2010
Selective Laser Sintering Part 6
I thought I'd take a second to make a quick post about some of the interesting bits of the new open selective laser sintering printer that I've been designing in between finishing up my dissertation. (I'm also looking for an interesting postdoc (or maybe the right industry job, if it's research oriented and challenging) -- if you're interested in learning more, please have a look at my graphical portfolio / academic CV. My background is pretty diverse, and I'm quite interested in branching into cognitive robotics, rapid prototyping, and potentially making these technologies a little "smarter", more accessible, inexpensive, or otherwise more capable. )
One of the overarching design goals of the project is accessibility -- it's very important to me that the design be easy to create, and very affordable. Infact, I keep a $200 high-mark in my head, with the goal of designing something that doesn't exceed that mark in terms of the parts cost. With these goal in mind, I've been designing a proof-of-concept that gets creative in order to keep things inexpensive while making them easy to create.
The typical components of an SLS system are a build chamber with a Z table, a powder feeding mechanism (that often takes the form of another Z-chamber, indexing up to feed powder in concert with the build chamber indexing down to receive the powder), and a x-y pivotable mirror or other system able to accurately aim the focus of the laser in two dimensions along the surface of the build chamber. With X-Y systems being relatively common (using anything from a few slides and lead screws or belts, like on the darwin or mendel, or using a pivotable mirror like on the first SLS powder test rig), I figured the tricky part would be in designing the feed and build chambers, and keeping this inexpensive. This could then be paired with an X-Y stage with the laser attached, and one would be set.
The design I've come up with is something that I'm really happy with. The whole idea is that, if you have access to a laser cutter, making complex custom parts can essentially be done for the price of the material. The more parts that can be cut, the fewer vitamins one would need -- and the lower the cost.
The following dual z-stage (feed and build chambers) that I've designed uses almost entirely laser cut parts -- ideally acrylic, but the engineering model I've build is made with hardboard (and even this might work out okay). The only vitamins currently are machine screws, as well as two large carrige bolts and two nuts that are used as the lead screws. (There are of course motors and electronics that also count as vitamins, but these are likely to be with any reprap system for a generation or two yet).
Two motors are used -- one for each Z-axis (build and feed). Here is one of the sample steppers for this engineering model, plus its mounting plate.
These two motor mount plates attach to a base plate. On top of this base plate is a system of laser-cut gears, where the nut for the lead screw is held captive in a large gear. The ratio of these gears is 4:1, so the motor can apply a good amount of torque to the nut and screw.
The nuts are also held in place with the top plate -- it acts as a bit of a keeper, to make sure the large central gears don't shift and can rotate in place, while also allowing access near the sides for manual turning just incase.
Here's one of the tables, turned upside down. It has a little system in it to hold the top of the carrige bold steady.
The chambers have two parts: the squeegee system, that sits on top, and the actual chambers themselves. The above is the piece that sits on top of the chambers.
The top piece has a system of groves similar to the first SLS powder test rig, except that this one is raised up a bit -- I had some concerns that since the plastic-on-plastic slide mechanism was also at the interface between powder and squeegee, that this might potentially clog fairly easily. With this new design, the slide is raised up one level of plastic, and a tiny little squeegee is glued under the slide itself -- so there's very little contact with the powder level, and much less potential for clogging.
The slide, with a little squeegee underneith.
The chamber system has two chambers, feed and build, that slot together and fit together with teeth. The intention is for these to be a close fit and glued together -- here things look a bit loose, but that's because I cut this model with 3mm material (for ease), rather than the 4.5mm material that the slots are designed for.
The top piece with the squeegee slide mechanism fits on top of the chambers.
And then the tables and dual z-stages slip right inside the chambers. There's still a bit of work to be done with alignment (in particular, the mounting plates that attach the dual z-stage to the chambers is out a little bit), but you can manually index the gears with your fingers and the tables slide up and down quite well -- and with quite a resolution! There is a little see-sawing back and forth from the size of the nut compared to the material that the gear is made out of, and also the give in the material, but I think this will likely be sorted out fairly naturally as those issues are addressed.
And so, there's the working prototype design for a dual z-stage mechanism with build chambers, for $2 in carrige bolts and nuts, $2.50 in hardboard, and $2 in machine screws! :) There will likely be some other materials required, such as a seal, but I'm thinking of using a dense foam similar to the seal material used on the first powder test rig (think pool noodle foam), that should be very very inexpensive as well.
While the design is very inexpensive, there are a few tradeoffs: There's likely to be quite a bit of backlash in the gear system. The trick is that, for this application, the tables should only (ideally) ever move in one direction throughout operation -- either up, or down -- and as such, the backlash shouldn't be much of an issue.
For the X-Y stage, I've decided not to use the pivotable mirror again, and instead make a simple one (mostly laser cut, again) using belts, that will fit over top of this system. I also harvested the stepper mechanism from a CD drive head, which I'd like to mount the laser to, in order to be able to easily adjust the focus online. During my tests just holding things in my hand, it seemed like the laser did particularly well when it was a little out of focus, and it would greatly aid testing if this didn't have to be adjusted manually! Also, it seems like a neat idea to be able to adjust focus online -- perhaps dynamic resolution at build time (heating large versus small areas), or even cutting (having a perfect focus), although I'm not yet sure where that last one might come in handy. (As an aside, I keep having ideas of building a larger x-y stage and trying to use the whole system as a dual SLS sytem and low-power laser cutter by placing a dark plastic sheet over top of the build chambers, and making a "reciprocating laser cutter" using the CD-head mechanism to 'bore' the focus into the material, but that's just unverified brainstorming right now. :) ).
I hope you've enjoyed this post! back to thesising!
Keeping with being open-source, I've placed the design files for the first powder test rig, as well as this current full design project, up on thingiverse:
Powder test rig: http://www.thingiverse.com/thing:3389
This project: http://www.thingiverse.com/thing:3390
thanks for reading! :)
[part 1] [part 2] [part 3] [part 4] [part 5] [cogsci.mcmaster.ca/~peter]
Monday, June 14, 2010
Hackerspace Brisbane gets its first Mendel!
lots more pics here:
After a number of days of furious tweaking it prints!
* pulling the extruder apart at least 10 times
* a leak
* a hot-zone conection failure
* a missing bearing
* mis-aligned cogs
* PTFE failure ( saved by a hoseclamp)
* two thermistor problems
* inverted end-stops to normal
* stepper driver pinout weirdness and ground loop issues - 5 hours here!
* X & Y axis needing reduced stepper power
* E & Z needing increased stepper power
* cog on X stepper came off
* X stepper mounted 4mm too low
* X belt too loose, then too tight, then too loose!
* a couple of random nuts falling off from someplace
* smooth-rails on X were insufficiently tightened, see belt issue.
first printed object:
a bit of video of a bit of the first print:
* at 1:40 I show overhead view at 2:00 you see electronics on the back at 2:25 you see host software.
Scott and I had our first RepRap session this evening. First off, we made sure the extruder was connected up correctly. When James and I were building Kurzweil a year ago, we lost the "12 Way Connector" that came with our BfB kit. We fudged a solution to get the extruder connected, but it's fragile, and got messed up during transportation across campus for the end-of-year poster session a month ago. Printing a new 12-way connector is an excellent candidate for a future project.
Once we were sure the extruder was connected properly, we performed the tedious but necessary rituals of bed-leveling and z-axis-setting (thus further fueling my desire for one of those well-constrained Mendels...).
We started noticing problems while printing the test rafts. The extruder was clicking, and I could see the filament jumping back, so I released the pressure bearings a bit. I may have gone too far, because the flow from the extruder became pretty feeble, so I tightened them back up half as much as I loosened them. That seemed to make the extruder run more smoothly but the flow was still pretty weak resulting in this test raft:
I think the problem is either in the tension of the pressure bearings, or some kind of clogging in the nozzle or the insulator, which, if you were wondering, looks like this:
I know that the bulge is a problem others have run into. The Bfb kit called for us to cut off a ring of the nozzle to restrain the insulator, but we didn't have the tools to do that and Ian at BfB assured us we'd be fine without it... I'll look around for solutions but if anyone reading this has a take on the situation I'd really appreciate the input.
...And We're Back
While today is not the first time I've worked on the RepRap since I last posted 14 months ago, it is the first time since then I've documented my struggles. Let me bring you up to speed (I'll try to keep it short)...
In the Fall semester of 2009 I took a Computer Architecture course taught by Michael Siff at Sarah Lawrence College. For my independent project I attempted to design a system of modular mechanical logic gates. I seriously underestimated the scope of this endeavor. I was trying to simultaneously:
- learn Blender
- learn Skeinforge
- design complex pieces using Blender
- maintain and troubleshoot Kurzweil
All while also trying to keep up with an intense Psych class and a very full Music program. Needless to say, I did not succeed, though I learned the hard way that making an idea into reality never goes as smoothly as one first imagines.
Slightly discouraged, and reluctant to set myself up for disappointment again, in Spring 2010 I took a break from RepRap. And then I graduated.
Now, in what may be my last month and a half in the Westchester area, I've set another possibly preposterous goal: wherever I go next, I want to take pieces for a Mendel with me. But this time I have help!
Scott Calvin, an SLC Physics professor, and my excellent Don, wants to learn the RepRap so future SLC students may experience its agony and glory. And what better way to learn how a machine works than fixing one that don't work so well? Hopefully with our powers combined, we will be able to get this puppy running.