troubleshooting delta build with heightmapping
shadowphile last edited by shadowphile
I just finished a two year process of designing and upgrading my custom delta. It has a print volume of 350mm wide and about 400mm high at the center.
My leveling is a lot worse than before. It was usually .025mm rms with 9 points and now it's 0.1mm, ugh! Mesh bed correction only drops that down to about .05.
Interestingly though, the heightmap clearly reflects a delta configuration. That implies that the mechanics are trued up and the problem is something systemic like effector and/or carriages are too soft, or some system parameter is off! (my system was heavy and accurate but I designed lighter versions of all the moving parts. Perhaps too light )
thanks for considering my problem
Please share some additional information on your setup. What is your probe? The results of M122, the contents of config.g, homedelta.g, bed.g, config-override.g would all be helpful.
And when in doubt, going through these documents can help a lot.
thanks for the reply! I shorted the info since I figured somebody would look at the heightmap and immediately know what the problem could be.
But here are the files you requested. I was already following the second of your links about calibrating a delta printer but it doesn't help if the parts are designed wrong (or something like that).
I'm still have a lingering worry that my new design is too squishy. My system before was rock solid but the new effector is rather squishy when I push on it so I'm still studying where the actual deflections/play are occurring and whether they would influence the bed compensations.M122.txt config-override.g config.g bed.g
I'm not much of a delta guy, but I do notice that you're still running FW 2.0 which is pretty ancient at this point. The easiest update for you would be to 2.05.1 which is only a year old, but doesn't require any config changes.
You can update by uploading that zip file as is to the system tab in DWC.
More info on what's changed between 2.0 and 2.05 found here: https://github.com/Duet3D/RepRapFirmware/releases/tag/2.05.1
As for the heightmap, maybe take a look at this: http://boim.com/DeltaUtil/CalDoc/Calibration.html
I poked around with updating recently but other things were higher priority once I found out heightmaps were already in my installed version. But thanks for the link!
Thanks for the link to the calibration too. I did run across it but none of the pictures matched my 'S' curve cross-section aligned with each tower, but I definitely should read it fully.
@Phaedrux forgot to mention, my probe is a custom piezo design that detects the nozzle contact with the bed. It taps out pretty consistently (<.01mm). I even checked it in the extreme regions of the heightmap but still reliable.
Veti last edited by
M350 X16 Y16 Z16 I0
you want interpolation on and you are missing the E config
G31 X0 Y0 Z0 P500 ; probe height from actual surface (regular V6)
your offset is missing
G30 P9 X0 Y0 Z-99999 S6 ;center and auto-calibrate 6 factors
try 8 factor and 9 factor and see what the results are for that when printing.
shadowphile last edited by shadowphile
-Interpolate: 'how did that get turned off?'. On the other hand, no seemed to change when enabled it.
-offset: my offset is about .5 and varies across the bed <.01.
-G30: 9-factor miraculously got me down to like .02mm rms. However, the rods are now tuned 4mm longer than the actual rods! Not comfortable with this. I also had to adjust the scale on x and y axes 1%. However, I saved it all with M500, then changed the G30 routine back to six-factor. There seem to be hidden variables that I shoved around because now six-factor still returns .02.
So I am good!
does this just record the height at each spot and linearly interpolate between them? (after G30) Doesn't seem to improve my overall rms.
No matter, I think I got it worked out; thought I might be in some major trouble there, whew! Consider this solved and thanks for the help!
I think the lesson I learned is that the heightmap is a useful tool for troubleshooting mechanical problems, since any looseness will show up as shifts in Z. In my case the delta symmetry of the heightmap indicated a common problem with all three towers: the effector wasn't stable, loose screws.
In fact, even with a bed flat to .02rms, I still get a ghost of a delta pattern, interesting.