Tuesday night, after most of the crowd had left, I got the 3D printer to move and extrude reliably, and printed two pieces with it. The first was the same part we had printed previously, before we started having nozzle problems: it is a printer part which acts as a spring mount for the home-position optosensors. The other is a Kwartzlab logo rendered in 3D by Alex with some modifications (scaling and a base to hold all the parts as one) by myself.
The two smaller L-shaped bits are artifacts of the GCode generation. The print head returns to the XY home at the end of each layer, and once the Z axis has moved up, the first thing done is to print another layer on this piece. I think its purpose is to snag drool that occurs during the home and Z move operations.
Last week’s conjecture that the feed motor shutdown was a thermal shutdown of the driver chips appears to have been correct, since the drivers worked fine again today. My other conjecture about the liner channel constricting because of differential thermal expansion appears to be wrong, or at least there is enough clearance between the liner and the nozzle body to allow normal outward expansion. I was able to insert the shank of the same drill size with the nozzle either hot or cold. In the process, I did however notice that the lower end of the liner tube and the the nozzle did not line up perfectly; I cleaned this up by inserting the same drill point first and twisting it with pliers to ream out the liner/nozzle joint.
Doug’s conversations at the Detroit Maker Faire would suggest that the nozzle problems are due to dirt blockage, insufficient lubrication (nothing said lubrication was recommended), and running the nozzle too hot. The latter at first seems counter-intuitive, but if the nozzle is too hot, the filament will start to soften and become sticky within the liner tube, causing excessive drag.
To get things properly tuned, I removed the motor and nozzle from the extruder. I adjusted the pinch roller pressure screws to the minimum pressure required for the drive shaft to grip the filament. Any additional pressure just increases the drag within the feeder.
To adjust the nozzle temperature I heated up the nozzle and manually pushed filament into it. At first I was getting bubbles in the extruded plastic, just like last week. Turning the temperature down to 230°C eliminated most of the bubbles except when the filament was pushed slowly (giving more time to the filament to heat all the way up to the nozzle temperature). A setting of 220°C seemed to give good extrusion.
I also turned the motor current down until the unloaded stepper just turned reliably.
After putting everything back together again, the print head was moved to the dump position and I had it extrude. The stepper current had to be turned up a bit, but the motor was not heating up like previous times. After extruding about a foot of filament it seemed to be time to try printing something again.
The opto-sensor spring was the first thing to try, given that we had printed a couple with limited success before. It turned out better than any of the previous attempts, with almost no spider webs. I was manually applying a drop of light oil to the filament every few minutes. It looks like it would be productive to add an oil-soaked felt pad where the filament enters the top of the feeder. This would serve both to lubricate the filament and also to wipe off any dust which might otherwise contribute to clogging.
The projecting tab on this spring still suffered from not enough cooling time between layers, giving it a somewhat distorted shape. But it is still an improvement over previous attempts.
After that success with the spring, we moved on to print a 3D version of the Kwartzlab logo. The original 3D model was done by Alex W, and I modified it by scaling it a bit and placing it on a round base so it would print as a single part rather than four loose pieces. For a first-time attempt, it turned out pretty good:
There are a few spider webs there, and some artifacts show up especially in the base disk. The horizontal lines are from the edges of the painter’s tape we use to cover the printer table. The general diagonal lines are the infill, which is laid at 45° angles, but seems to be missing a couple of rows. I don’t know if this was a printer problem or a flaw in the GCode generation. The infill certainly does not just start at one edge and work across as one might expect, but instead does apparently random stripes.
Also unexplained was why some of the parts of the logo were solid (with infill) and others, notably the small circle and two squares, were hollow. Each of these has a roof which was just formed by stretching the molten bead across the gap one way, then crossways to that, so each roof has a bit of sag in it and is not solid. Unfortunately I haven’t had time to take a close look at the original Sketchup file for clues as to why this happened.
There is also evidence that the table is not quite aligned with the plane defined by the X and Y axes because some parts of the edge of the disk are cleanly formed while others have a squished appearance. Setting the Z zero is tricky to do since there is quite a bit of give in many of the parts and trying to insert a feeler gauge between the nozzle and table is largely a matter of feel. I think the proper way to set the table level and Z zero is with a special test print pattern which consists of a stepped part one, two, and three layers thick, located at each corner of the table. If the Z zero and leveling are correct, measuring the parts with a micrometer should give readings that are one, two, and three layer-thicknesses.
Now, although the basic printing is working now, it is still far from a turnkey setup. There are far too many extra steps involved, including editing the .STL file to uppercase all the ‘e’s in the exponential-notation numbers, and editing the generated GCode to replace the M109 (set temperature and wait to stabilize, which unfortunately often waits forever) with M104 (set temperature). There are also too many ways to get things into some sort of snit…