- User Since
- Apr 3 2017, 8:53 AM (97 w, 6 d)
Thu, Feb 14
Wed, Feb 13
@matth: The fix is baked in to .87
Will roll this in with T5685
I'll make a branch for this and T5697 and submit a fix.
Tue, Feb 12
Nevermind. I am able to reproduce it.
Also, Cura will need to send "M77" after a print abort, otherwise the print timer will continue to run indefinitely. This may actually be the cause of the incorrect print statistics in T5508.
Can you send me the GCODE from a print that is exhibiting the problem?
@matth: I suspect it's the bed leveling that remains enabled after a print. Can you run "M420 S0" from the console after the first print and see if it restores your ability to calibrate? If it does, it would be an easy fix for me to do that prior to the auto-calibration.
What follows is an excerpt of "Q_CubeOff_PLA.gcode", showing just layers 20 and layers 21, with the FILL sections extruded.
I loaded "Q_CubeOff_PLA.gcode" into http://gcode.ws, and it shows extrusion speeds in different colors. There is definitely a pattern of alternating extrusion speeds in the GCODE from layer to layer and a difference in extrusion speeds between the top half and the bottom half of the part.
Fri, Feb 8
@logan: Enabling backlash compensation throughout the print will cause surface defects. This was the reason for fading out the backlash compensation. So dimensional accuracy comes at the expense of surface finish. The only way to avoid this trade-off is for the printer to have the least backlash as possible.
@logan: On Quiver, backlash compensation is on all axis. What the fading will do will decrease the backlash the further up the print it goes. So if your printer has 0.5 mm backlash, at the bottom of the part, it will be the right size, but at the top of the print, it will be off by 0.5 mm.
@logan: The auto-calibration only measures the values. It keeps the the state of those features exactly as they were before calibration began. I think Cura sends a command to turn on backlash compenasation as the start of the print.
@logan: I configured an FPGA as a step counter and here is what I was able to find out.
Thu, Feb 7
I don't have a SL toolhead, but if you get me one, I could look into this.
_detectSerialPort shoud accept both port.vid == 0x03EB and port.vid == 0x271B. Here are the possibilities:
Wed, Feb 6
@logan: It would take a bloody long time, but it may help to print a calibration cube with the micro toolhead. The reason being that it has a smaller nozzle and thus the effects due to wall thickness will be minimized. If that calibration cube comes out sized better, it indicates it might be a wall squish issue. If it shows the same size error, then it likely is a true positional error.
Tue, Feb 5
@karrad: In order to do the scaling correctly, it would actually have to analyze the printed parts for extruder 1 and the printed parts for extruder 2 independently, assuming a merged model. Because you could very well have a part where nozzle 1 could encroach on the exclusion area for extruder 2, but extruder 2 does not do so. So this is probably a tricky problem, if you want to always allow the user to print the largest possible models.
@logan: If it is not in your path, you need to do ./bossac -i -U true -e -w -b -R <firmware_file>.bin
Also, try printing a cube that is 20mm, with a 10mm hole in it. If this is a dimensional accuracy thing, the 10mm hole would exhibit less of an effect. If it was an over/under extrusion issue, then it would show up in the hole equally.
Try printing with backlash compensation off and see if it makes a difference.
Backlash compensation could also have an effect on flow, so it is possible it could throw off calibration prints by a bit, if walls are thinner than the slicer thinks they should be. But .5mm is quite a lot, so I don't suspect that walls would be an explanation.
I already explained this to @logan, but for anyone else lurking, this behavior is correct. When you increase the Z nozzle offset, you are telling Marlin that the 2nd nozzle is higher relative to the 1st nozzle, which means that during a toolhead switch from nozzle 1 to nozzle 2, Marlin will move the Z offset down lower towards the bed. So yes, increasing the nozzle offset in Z will lower the 2nd nozzle closer to the bed.