51 Reply by adrian (edited by adrian 2013-04-30 05:30:30)

  • adrian
  • Member
  • Offline
  • From: Canberra, Australia
  • Registered: 2013-01-15
  • Posts: 1,623

Re: i need some advice for circles and sticking

wilheldp wrote:

How do you determine the amount of backlash in the x and y axes?  I did the z-wobble calibration that required printing 3 cuboids, but I haven't seen any x and y backlash calibration cubes.

Print a 50mm object such as a long and thin rectangle then measure it... the difference +/- the 50mm's assuming calibrated wall thickness ("Extruder Calibration") is roughly your backlash.

The more precise method, is to print the Dial Gauge holder and use a Dial Gauge braced against the case or a block. Move the head in 1mm increments and ensure that you see 1mm changes on the dial gauge. Any deltas is once again the amount of backlash.

Whilst backlash is present in the X/Y axis due to how the belt teeth interface to the pulleys, in reality its far more minimal than you get out of the Z-Axis. Unless you need extreme precision in the X/Y axis and very tight tolerances, I would consider it a nicety and not a neccesity in comparison to dealing with Z-Axis backlash.

52 Reply by mconsidine

  • mconsidine
  • Member
  • Offline
  • From: Randolph, VT
  • Registered: 2013-03-07
  • Posts: 122

Re: i need some advice for circles and sticking

wilheldp wrote:

How do you determine the amount of backlash in the x and y axes?  I did the z-wobble calibration that required printing 3 cuboids, but I haven't seen any x and y backlash calibration cubes.

I mounted a digital dial indicator onto the bed, such that I could move the extruder in the Y-direction 1mm at a time.  Then I wrote down all the readings as I used the "manual" mode in RH to move the extruder.  Did this going in forward and reverse for a few millimeters.  At the switch in direction, where there would have been a 1mm move if there were no backlash, there was instead a move of about .6mm.

I also recorded the indicator readings for all the nominally-1mm moves.

Did this routine 5 times and averaged the results to judge both backlash (thats where the .382mm came from) as well as what the true stepping rate was for the Y-axis stepper.

Because of the size of the dial indicator and my off-the-cuff clamping, I couldn't measure the x-axis in the same way (I need to find a way to bolt on a non-flexing flat surface to the extruder).  What I did instead was print 3 cylinders with 0, .1mm and .2mm settings for the X-axis backlash.  Then I eyeballed which one looked best.  That's the one you see in the picture.  FWIW, I've so far left the X-axis stepper at 88 steps/mm.  The Y-axis came in very slightly less than that - around 87.92 if I recall.

This was on an out-of-the-box SD3 (with only a firmware update).  So I'm sure things could be improved with the whole belt-tightening routine.  At the end of the day, though, I don't particularly care if the improvement is made in hardware or software as long as the end result is the same.

Hope this helps,
Matt

53 Reply by mconsidine

  • mconsidine
  • Member
  • Offline
  • From: Randolph, VT
  • Registered: 2013-03-07
  • Posts: 122

Re: i need some advice for circles and sticking

adrian wrote:

Whilst backlash is present in the X/Y axis due to how the belt teeth interface to the pulleys, in reality its far more minimal than you get out of the Z-Axis. Unless you need extreme precision in the X/Y axis and very tight tolerances, I would consider it a nicety and not a neccesity in comparison to dealing with Z-Axis backlash.

If I can get the code ported over, I'd like to see if the Z-axis can be adjusted in the same way.  What I'm thinking, though, is that the Z-axis might be affected by periodic error (again, evident in telescope mount worms/gears).  In which case it might be possible to create the equivalent of a "pointing" model (where software adjustments are made to improve the pointing accuracy of a telescope) that is simply a function of the Z-axis coordinate.  Haven't really thought this one through yet though ...

Matt

54 Reply by adrian (edited by adrian 2013-04-30 13:05:40)

  • adrian
  • Member
  • Offline
  • From: Canberra, Australia
  • Registered: 2013-01-15
  • Posts: 1,623

Re: i need some advice for circles and sticking

Is that any different to the existing z wobble calibration software fix? There is also the "hysteresis" fix which kinda sorta worked, but really was about backlash (dunno where the hysteresis entered into it.. smile )
http://www.soliwiki.com/Calibration_of_the_Z_wobble ?

55 Reply by mconsidine

  • mconsidine
  • Member
  • Offline
  • From: Randolph, VT
  • Registered: 2013-03-07
  • Posts: 122

Re: i need some advice for circles and sticking

adrian wrote:

Is that any different to the existing z wobble calibration software fix? There is also the "hysteresis" fix which kinda sorta worked, but really was about backlash (dunno where the hysteresis entered into it.. smile )
http://www.soliwiki.com/Calibration_of_the_Z_wobble ?

I had initially thought the hysteresis fix was what I wanted to deal with the backlash but found it had been pulled from the firmware.  After that, I tried looking into Skeinforge's lash settings.  But for whatever reason found that my Skeinforge install wasn't extruding (traced the path, but the E codes had very small values in them.)

When I came across the NoLash thread it suggested that Skeinforge's lash setting maybe wasn't implemented quite correctly.  So I didn't go any further down that rabbit hole and instead chose the more familiar burrow of converting code to C.  Which would work on Linux anyway ...

Yes, the link you posted to the calibration of z-wobble does in fact look to be what I had in mind.  Thanks for making me look at it smile - I was aware of it, but hadn't gone there yet...

Matt