Monday, November 28, 2011

Y axis mods

I came home and found that my Y pulley had snapped. I was running that belt quite tight. Luckily it seems to have broken after finishing the current print. So I’m running an 8 tooth pulley there now since it’s all I had sitting around.

Since I had to get the Y platform off to re-tention the belt for the 8 tooth pulley anyway, I took the opportunity to switch to an iteration 2 style platform, with the bushing holders glued straight to the bottom of the upper platform with no leveling provision.
It’s working and it’s a good thing. Much simpler, no need to level, much less moving mass, signficantly less hardware, the print surface is more tightly bound to the drive belt, so it probably will help with the overshoot and oscillation I’ve seen on the Y axis sometimes. And now the belt attaches to the bottom, so there’s a ton of room for putting on a belt tightener clamp. And there’s a mile of room under the printer now, enough for power supplies and electronics. I may try designing an electronics holder that fits under there.
I still haven’t figured out exactly what to do with the heated print bed, specifically where to put the thermistor. I think I might drill a hole up through the center of the bed and use a dremel or just a sharp knife to carve a small slot for it. Right now I spaced the heating board up with a couple of fender washers on each corner but that’s messy.
Also I think I’m going to recommend going slightly rectangular with the top plate and print surface. 225mm square is really pretty tight to try to get clips on the edges and still get 200mm of print surface. There’s no reason it can’t be quite a lot longer, I think I’m going to make my next one at least 250mm long.
I did offset it to the right a bit to allow for the asymmetrical extruder. I think I might have very close to 200mm of both X and Y travel now.
I also want to do something a little nicer than the Z axis adjustment I have now, which is a thing printed from Thingiverse. It’s ok but a little klunky. Also the endstop holder for the Z axis can wobble a bit, I’m going to think about a more solid mount for that, since on that axis a fraction of a millimeter matters.

Friday, November 25, 2011

GT RepRap project update

An update on the 4 GTers that I’m guiding on building 3D printers. I’m building one along with them, so if anyone else wants to just have one without building it, I’ll have one for sale when this is all done, in about 3 weeks. I’ll probably bring it to Capricon for demos, write me a check and you can take it home.
It’s a funny time to build a printer because Prusa just released “Iteration 2″ of the Prusa Mendel printer. It’s got some exciting changes, but it does mean that I’ve probably seemed horribly wishy-washy to these poor folks. The new version of the printer uses less parts, should produce better prints, and be easier to use.
Just yesterday I discovered that the official recommendation now is to use fine pitched timing belts rather than the old T5 (or XL in the US). The recommended T2.5 is hard to get in the US, but after chatting with some of the leading folks on RepRap development on IRC, They pointed me to a supplier but also informed me that it’s actually better to use “GT2″ (how appropriate) timing belt and pulleys – it’s designed to eliminate backlash in this sort of situation.
The old XL belts are still fine, but the new stuff is supposed to produce slightly better prints, and it’s not much more expensive.
I did determine that using really tiny pulleys is really bad. I put on some 8 tooth pulleys and the backlash is horrible. I went back to 10 tooth and it is much better. Looking forward to trying GT2 36 tooth.
The even more exciting development is the elimination of the upper and lower print bed. This reduces moving mass, greatly reduces complexity and also eliminates periodic levelling of the print bed, which is a pain in the rear. Apparently just leveling once during the build by adjusting the frame is good enough. If this works I’m changing my old printer as well, because the print bed complexity is irritating and leveling is a pain in the ass.
There’s an Iteration 2 build party going on in Cologne, Germany on Dec 3-4, more info here. I hope to get good tips and photos to help me put together good instructions for our folks.

Friday, November 18, 2011

Found on my printer this morning

Found this cute guy waiting for me this morning:

I’ve got a project that’s going to have me loading every color I have in the printer over the next few weeks, and I thought as long as the colors are loaded anyway, I’d build myself a little army.
TechCrap: Ultimachine 4043D PLA, 10% infill, 0 extra shells
Here’s a story idea for free: computer virus finds 3D printers left idle, uses them to build machines for some nefarious purpose. Clearly a more sophisticated 3D printer than what we have now but if we had Star Trek replicators without some kind of antivirus protection, your food replicator might make poison, or you might have killbots flying out of the thing and taking out the entire mess hall

Wednesday, November 16, 2011

New plastic source

Yay, I found a new PLA source that has a variety of colors, good plastic (4043D), reasonable prices ($65 for 2x 2.5# spools) and reasonable shipping. Now if they’ll just answer my email about how the heck I order from them – there’s no obvious way to do so on their website that shows products.
UPDATE: no website based ordering.  You want something, you email him.
I like that they have 2x 2.5# spools for the same price as most are selling 5# spools – it’ll allow me to stock up on colors without having to buy 5# at a time. Though at the rate I’ll probably be going through blue to print TARDISes, I should probably get 5# of that anyway.

Tuesday, November 15, 2011

3D Printing at Windycon

I spent the weekend basically talking about 3D printing – my throat was sore by Saturday night, but that’s not unusual at a convention. If I’d been able to buy blue PLA before the con, I would have been printing TARDISes all weekend. I did print 2 blue, 2 translucent cerulean and 2 white ones. Translucent doesn’t look as good, you can’t see detail on translucent plastic.
In future, I need to come with a box of a wide variety of printed things. I didn’t have a lot with me at first, and everyone asked “well, what can you make with it?” What CAN’T you make with it? But the possibilities don’t seem to register with people very well unless they’ve got stuff they can touch.
I printed out a TriLego piece Sunday morning and got as much excitement out of that as I did anything else.
I think I might try to make an entry for the Duckon Critter Crunch out of mostly printed parts. There are already gearboxes and tank treads up on Thingiverse. That’d be a good demo.
I found a nice 8 tooth pulley for X and Y axes and printed some out last night, so I’m switched from 10 to 8 tooth. That should increase my resolution by about 20% on those axes. I did have to reduce my jerk settings a bit, I’d cranked them up a few days ago because I didn’t know how high they could go with these motors. 20mm/s2 is apparently too high because the splines started misaligning during the print, so it’s at 10 now and a set of 4 pulleys printed out very well.

Wednesday, November 9, 2011

Today’s RepRap lessons

Use “Cool”
otherwise layers that draw quickly just blob around and don’t bond.
Use the “Orbit” method of the Cool module
not the “slow down” which just lowers printhead speed until the layer takes the requisite number of seconds.  I printed a part with “slow down” and when it got to very small layers, it was moving so slowly that the steppers started hitting resonant frequencies with the Y platform and there were horrible shudders in the printed part. Also on the last layer that was perhaps a 5mm circle, it just sat there and melted down into the already printed layers and moved around a little.
I’m printing my first part with “orbit” now – it prints the layer at normal speed then retracts the filament and just moves the head around the outside of the part for a minute while the plastic cools, then goes on to draw the next layer. This seems to be printing very nicely indeed.
Tighten down the extruder idler FULLY
Horse that thing down. I didn’t, and the filament slipped enough that the shaved off bits clogged up the hobbed bolt enough that it started not feeding filament for parts of the layer. I had to stop and clean out the bolt and restart the piece.
Power failures
I really should start printing from my netbook and put a 12V battery in circuit. The printer running PLA only draws 4 or 5 amps, so a modest 20 amp hour deep cycle gel cell would run it for several hours. I think this is a better idea than a 1000 watt UPS, and would certainly be far cheaper.

Thing #13453 – and some cleanup work on the printer

In preparation for taking the printer to Windycon, I modified my power jack holder to also hold a little coaxial connector for a smaller power supply and uploaded it as a derivative work to Thingiverse.
Since I had the printer partially apart to install this, I also did a ton of cleanup on the printer, rerouting wires, resquaring things, cutting the Y axis smooth rod down to the proper size (I used a cutoff wheel in an angle grinder this time, MUCH faster than a Dremel tool (30 seconds instead of 10 minutes) though somewhat scarier.

Tuesday, November 8, 2011

Today in RepRap Land

With all the fighting I’ve done with SFACT and Slic3r over the last few weeks, I decided that I needed to cut to the chase and try using the de-facto standard, Skeinforge. It’s got a reputation for being absolute hell to get configured correctly. It does have about eight point seven jillion settings, but I went through them and after 3 or 4 false starts (layer height too high, then I needed to add start gcode to reset the extruder at the start) it’s producing quite fine output. My first print suffered from a combo of design problems and me not using the Cool module, but I’m on my 2nd print and that should be fixed.
I got fed up and tore down the X axis and finally replaced the crappy Chinese linear bearings on the X carriage with printed PLA LM8UU look-alikes from Thingiverse. I also replaced them on the idler side Z axis that I disassembled to get to the X carriage while I had it off, and I weakened the Z axis anti-backlash spring on that side as well, it was way more stiff than it needed to be and was causing excess drag.
I did some rewiring to run the wires for the right hand Z motor to follow one of the top gantry bars under the HDPE tubing there, it looks a lot cleaner that way. I’m moving the Z axis junction point to the left hand side frame where all the other wires are. This leaves the rest of the frame very clean looking. I still have some cleaning up to do of the Z axis wiring. I really should print an electronics housing soon too.
Also the Y axis setscrew is digging into the Y motor support, and the Y axis seems to be a bit uneven in the prints (could be related).
I reduced the “jerk” settings in the Marlin firmware because drawing some of the circles at high speed was causing a vibration in the frame that transferred to the printed object.
I redesigned the power jack holder that I had trouble printing and re-skeined it with the cool module enabled. It’s printing slow because it was set to 60 seconds minimum time per layer, but I’m going to leave it all night anyway so it doesn’t matter. It also has more support than the original design so it’s all good.

Friday, November 4, 2011

Oh heck yes

Relief. stringing problem 100% solved by using Slic3r, a new gcode generation program. It’s also about 5x faster and generates gcode that prints faster, and is far easier to configure even than SFACT, which is far easier to configure than Skeinforge.
(Edit from 6 weeks later) In hindsight, the problem here was simply not using the proper retract settings, but SFACT hides retraction behind a wall and you can’t really edit it properly.  The OldDimension module might fix that, but in the end I just gave up and went to Skeinforge)
There’s still some unevenness in vertical surfaces, probably caused by slightly low flow rate (I could have overstated the filament diameter, for instance) or perhaps the hotend temperature was a bit high. I’ll work on that.

Thursday, November 3, 2011

Promising new 3D printing software

I’ve attacked every other possible source of my problem, so today I decided to try a new piece of software to generate gcode from STL files. This software is called “slic3r” and replaces Skeinforge in the tool chain. It’s written in PERL rather than Python, and claims to have a much cleaner, more maintainable code base.
(more in main post)
I do know that I really like what I see. It’s probably about 5x faster than skeinforge, and the gcode that it outputs is very fast on the printer, it doesn’t do some of the stupid stuff I always see Skeinforge doing, and though I have another test to go, it looks like it may have solved or at least radically reduced my problem with stringing.
The only major issue I see right now is that it does not lay down solid layers on “bridge” layers. This really needs to be added if only for strength of some parts, but I’m confident it’ll get there.
As a bonus, this thing is incredibly easy to configure compared to Skeinforge. It has a single config page containing the values that people actually change in SF and none of the ones that you never really have to touch anyway.
I also moved the quad core machine from my desk down to the lab and retired the old machine that was down there, since I’m using my laptop (with external keyboard and 2nd monitor) as my main machine now. It’s nice having a fast machine in the lab, it saves me a ton of time by allowing me to skein upcoming parts while printing the current one. Before I had delays of up to half an hour between parts, now I can clear the bed and immediately start another print.
I don’t know if this video shows much to anyone else but to me it has obvious good things compared to how SF gcode prints.
Here’s a pair of pulleys printed with slic3r generated gcode. I’m super pleased with these, they’re far smoother than similar parts that I’ve printed on the same machine before.  However, there’s some work to be done with void bridging as you can see from the porous surfaces here.