Improving auto-calibration.
-
I too likely need to add H values to my probe points. My main issue is with the variability of the test for when the nozzle is gripping a sheet of paper. There's some subjectivity there.
What increments are you moving when you're calibrating those H values? 0.1mm, or lower?
I'm also a bit confused about how you determine the H value. Do you do a paper test like is outlined for configuring Z0 in the IR Probe setup? (Move to that point, lower the head until it grips a sheet of paper, set that as Z0, move up 5mm, then probe and then subtract the height at the bed center from that new value?
I use 0.025mm increments on my delta with 160 steps/mm. If you have 100 or 200 steps/mm then you should use 0.020mm instead. The procedure I use is:
1. Move the head to over the probe point.
2. Lower the head until it just grips a sheet of paper (this is where I use 0.025mm increments near the end).
3. Send G92 Z0.
4. Raise the head 5mm.
5. Send G30 S-1 to lower the probe until it triggers.
6. Read off the Z height in the web interface.
I usually repeat steps 4-6 a few times to make sure the trigger height is consistent. The H parameter for each point is (trigger height at that point) - (trigger height at centre of bed).
HTH David
-
David, that's exactly what I was looking for - thank you!
(You might consider adding those steps to the wiki for clear instructions on setting an H value for each point)
Is there a way to set the web interface to jog at 0.020? I have the "half move rate" checked, which took it down to 0.05mm
-
You could make a macro with a relative move. Something like this for going down:
G91 ; relative moves
G1 Z-0.020 ; go down a bit
G90 ; absolute moves -
You could make a macro with a relative move. Something like this for going down:
G91 ; relative moves
G1 Z-0.020 ; go down a bit
G90 ; absolute movesThat's exactly what I do, except I use 0.025mm because that is an exact number of microsteps on my printer.
-
Mine's at 80 steps/mm
As an aside, I'm only calibrating down to 0.06 now, so hoping this makes a big improvement
-
When you say, "just gripping a piece of paper, is that until the paper can't move, or until we feel some resistance?
-
One thing i found out while doing this was that my endstops were loosely goosey… i tightened them up and now at least i can get repeatable results
-
When you say, "just gripping a piece of paper, is that until the paper can't move, or until we feel some resistance?
I lower the nozzle until i can feel a significant resistance. The important thing is to be consistent.
-
What do you guys think about my H-values… most of them are negative... i'd expect for them to be centered around zero, not mainly negative?
-
Qdeathstar, in what way does the auto calibration got worse?
Your H values seems ok to me, I am probing 32 points and my H values add up to 1.6
I don't think the effect of the end effector tilt would simply make them center around 0 -
Sorry, i did not find the real trigger height, i subtracted z-values (which made the h-values the inverse of what they needed to be)… I think i have the issue fixed, im testing now. The z-height overall was a little low on the first go.
-
Glad to hear you got it solved
-
well, i got that part of it solved but now it seems like the area across from the y tower is a little low and the area close to the y tower heading around to the z-tower is low…
i'm wondering if i can make manual adjustments to the H values to dial it in?
-
You should only use H corrections to compensate for actual variations in trigger height, which on a delta printer are usually caused by effector tilt. If you are still getting significant errors close to points that are included in the auto calibration bed probing, either fix the geometrical errors that are causing them, or use the new grid bed compensation in the 1.17dev6 build after auto calibration to compensate for them.
I have it on my list to add a wiki page on delta calibration.
-
Thank you. Geometrical errors in this case are most likely to be a warped bed, correct? Or is there something else i should check?
Also, should the nozzle hit the bed at z=0 for x=0 & y = 0? When i check there i got .25mm off the bed?
-
actually, it turns out when i found out my endstops were loose, i never readjusted z=0….
I redid the h-value test and now it's almost perfect!
-
Also, should the nozzle hit the bed at z=0 for x=0 & y = 0? When i check there i got .25mm off the bed?
Yes it should be much closer to zero height than that. Perhaps your trigger height in the G31 command is not set correctly?
-
Thats not the only issue i had, after reading your comment DC42, (and haivng some other issues with prints) I discovered my steps/mm setting was off. I had it as 82.5 steps/mm instead of 80 steps/mm…
Originally, i had to fudge some of the H-values to get the circle to print correctly, now it looks great w/o the fudging! (even better). Ill edit in a picture once its completed.
Just posting this for posterity. Thanks for your help.
