Thursday, September 22, 2011

Better, Faster, Stronger...

Over time my Y-axis in particular has gotten more and more troublesome. I was using PLA bushing holders and bronze bushings. When aligned well it worked great but the tolerance was very tight and the slightest change in the rods caused the bushings to bind. I made the mistake of trying to lubricate the setup with some white grease I had for my bicycle, and it worked initially, but eventually it seemed to thicken up and made it worse overall. So I cleaned up the rods and used some light machine oil that at least got me back to where I was before the bike grease blunder.

My copy of Digifab's LM8UU holders
I had ordered some LM8UU linear bearings a while back and finally, about three weeks after the order, the boat from China my new bearings. This was definitely the time for an upgrade. There are several options for bearing holders available on thingiverse, but I chose Digifab's bearing holder because of the provision for cable ties, and the ability to access the attachment screws with the bearing in place. I crossed my fingers through the whole print as the Y-axis had started skipping steps recently. The holders printed out fine and I started making up a new bed using a scrap piece of smoked plexiglass.

Old bronze bushing plate and new LM8UU plate
I decided to go with just the three bearings and I have to say it worked out very well. I was initially concerned because there seemed to be quite a bit of play with just one bearing on the rod, but once all three were mounted to the plate and in place on the machine, it was fantastic just to feel how smooth the movement was and with no noticeable slop.

Fully installed bottom plate with LM8UU bearings
I also decided to try using just three leveling screws instead of the usual four partly because the new holders didn't leave much room near two of the corners of the plate, and partly because I found trying to level four points to be annoying as one always seemed to be loose. I still haven't decided if I like it better or not. Need some more experience to really know whether I want to stick with it.

I used the cable tie trick to tension the Y-axis belt and it worked great. Thanks to whoever posted that on thingiverse (nophead I think). Once I got the belt on, the amount of resistance to movement with the plate increased dramatically. Way more than could be attributed to just the stepper. The only potential source of friction I could see now was where the belt rubs the fender washers next to the 608 bearings. I loosened the nuts holding the fender washers and the resistance dropped significantly. I had no idea the side of the belt could create that much friction from slight misalignment with the bearing. So I moved my bearings a bit and tightened it all back down again and the plate movement was much better.

Printed off a quick rocket shot glass and I have to say I'm completely happy with the new Y-carriage. The walls are much smoother and more consistent. Time to do something about the X-carriage.


Thursday, September 15, 2011

HBP Testing and New Filament

First the good news. I finally did some testing with my prototype glass heated build plate and was able to successfully get to a usable temperature for ABS. I first tried to test it by laying it on a piece of G10 which my current build platform is made from and hooking it up to a suitable ATX power supply. After 15 to 20 minutes, I wasn't able to crack 90°C. My first thought was that this was an issue with the heating conductor since we kind of missed the mark a little on the conductor trace and ended up with too high resistance. But while cleaning up from the test, I quickly realized another possible issue when I felt how warm the G10 had gotten.

For the next test I decided to put the plate on a sandwich of cardboard and foil tape. This worked much better and I was able to reach a decent temperature fairly quickly. I also took some shots of the plate with the IR camera at work during the test to try to get an idea of the heat distribution. I don't think that worked out like I had hoped. I'm fairly sure the camera was seeing through the glass to the heating element and cardboard below. But even though the pictures might not be an accurate representation of the heat on the surface, measurement with a thermocouple showed the top surface to be right where I needed it. And if my calculations are correct this is only using about 6.6 amps from a cold start, and somewhat less once it's warmed up. This is just on a 130x150mm plate, so a larger plate will require proportionally higher current.

Unfortunately, I'm not sure our idea for the thermistor will work out. I didn't know how the arduino was measuring resistance until I did a little research, and with my limited experience in electrical engineering (nearly none), it didn't occur to me that the amount of resistance we were getting out of the thermistor trace on the plate would be difficult to work with. The arduino with RAMPS is doing a voltage comparison by creating a divider on a 5V circuit with a 4.7k resistor on one side and the thermistor on the other. The typical thermistor is 100k and has a really huge swing in resistance which works great with the 10 bit resolution of the analog inputs. But my thermistor only vaies from 230 to 290 Ohms over the temperature range I'm interested in which isn't enough. I could swap out the 4.7k resistor for something close to the value of the thermistor, but I'd rather not mess with the board, and I'm not sure what affect the increased current (~50mW by my calculations) would have on the arduino board. So I think I'll just tape a 100k unit to the top of the board for now.

I ordered some new filament from Ultimachine. So far I had only used some samples of black, green, and silver PLA from Ultimachine (it's not chocolate, but you can get a few Octopuses/i/odes out of it) and a pound of natural PLA from MakerGear. All of those had worked really well and I had no qualms about ordering more variety of plastic from Ultimachine. I got a pound each of the orange, blue, and silver PLA, and a pound of natural ABS in anticipation of having a usable HBP. I managed to print the small gears and pins of the screwless cube gears thing in natural PLA before I ran out. I thought I'd give the orange a try for the big gears. I took a caliper to the filament and got measurements all over the place, from 2.8 to 3.1mm. I also noticed that the texture of the plastic was significantly different from any other PLA I had tried. It seems much softer. This did not bode well, but I decided to give it a try anyway. It did not work. I did manage to get one successful 20mm calibration cube, but had issues with everything else. At first I though there was a problem with the extruder because the filament was stalling. I tried cranking up the heat thinking maybe it just wasn't melting as quick as the other stuff but that didn't help. I took apart the cold end to see if the hobbed bolt had gotten gummed up by chewing into the plastic, but it really wasn't bad. I tried really cranking in the idler bearing on the filament, and it helped a little, but not enough to get a decent print. My final conclusion is that the filament is just too big in spots to fit into the hot end smoothly. I changed over to the silver and it worked fine. I hope that I can trade in the orange, because it's not going to work for my machine.

I clearly need to blog more often so I don't feel like I'm writing a novella every time I decide to post. Next up, dealing with mechanical issues and some ideas for vertical X-axis design.