Delta axis calibration
-
Hello everyone!
I recently updated my Delta from from a Rambo controller to Duet 3 6HC. One item which I think could gain me a bit of accuracy is if I calibrate each tower to the proper steps/mm. In my case, activating each tower and calibrating, I am a little above and a little below 320 on each tower. My results are X320.978 Y322.250 Z323.229, by calculation with the GT2 belts I should be around 320, but for whatever reason the actual measurements are a little off.
My question.
Is there any value to calibrating each tower's steps per mm?
-
@th0mpy
That is a question I've often asked myself.In theory, the answer is "yes", but then I keep coming back to a different question: Why would the steps/mm be different for different towers? If the stepper motors, pulleys, belts, idlers and distances are all identical, the steps/mm should also be identical IF the tension of the belts are all the same.
...and that usually leads me to adjusting the tension of the belts to try and get the values the same instead of adjusting the M92 values.
-
@garyd9 said in Delta axis calibration:
If the stepper motors, pulleys, belts, idlers and distances are all identical, the steps/mm should also be identical IF the tension of the belts are all the same.
Tolerance stacking?
-
That's the same thing I wondered as well.
As much as I can tell, the belts are really close as far as tension. No matter what I always get varying amounts of drift from one carriage to another. One thought I had is maybe I'm running some crappy knock-off GT2 belts/pulleys but I honestly don't know where I got them, and I believe at one point I've switched them out because I wore one out a few years ago. Of course I got them off Amazon, so they're likely not genuine. Anyway, I digress from the point.
If all things are not equal is the consensus that calibrating each tower is a good thing?
-
@th0mpy said in Delta axis calibration:
If all things are not equal is the consensus that calibrating each tower is a good thing?
If it results in more accurate movement, sure, why not?
-
Pretty much that. It's still an open loop system at it's heart so why not if it makes thing better. Under ideal conditions sure it'll be the same and it's worth a best effort attempt to make it so, but at certain point you gotta make a call how far down the rabbit hole you wanna go. Throw on closed loop servo-like stepper control boards? Characterize a bunch of stepper motors and their performance so all three axis match? There are lots of timing belt alternatives too if you want to go industrial solution (ie, priced for a business). One sanity check i would suggest though is to measure carriage movement per tower without arms attached. That would at least let you isolate whether the discrepancy is originating in the towers or the arms and effector.
-
@th0mpy in your position I would be tempted to fit some genuine Gates glass-reinforced belts and see whether they give you the same steps/mm on all towers using equal tensions. I must admit, it never occurred to me to measure the actual steps/mm individually for each tower.
-
Sorry to resurrect an old thread. My company began to require us to come back to the office now that everyone has been vaccinated. Sucks because it really cuts into my 3d printing time
ANYWAY. @dc42 I did exactly that as one of the last things I tried. I still get 1-2mm of difference between towers. I have a heavily modified Rostock Max and even gone as far as to put belt tensioners on the top to ensure proper tension.
I have found that instead of equally tensioning the towers I can tension them to get equal movement, but that really doesn't seem like the right thing to do. One is very taught while another seems to be quite loose.
I suppose you could argue that mesh leveling would sort that out, and to a point it does but I still get some discrepancy the further out I go between the XZ towers, everywhere else is great.
Seems a lot like the individual arm length is the culprit but I've moved the arms around and the same issue exists in the same place. Side note, I've even tried to rotate the effector, same thing. To me this would then maybe indicate that one tower is off with regards to steps/mm (or whatever has a direct effect on that measurement). Sure enough there is a significant discrepancy between all of them, with more on the Y tower.
So far I haven't 100% tracked this thing down, I can solve by printing a raft and the print comes out great, and dimensionally accurate.
Any other suggestions of what to try? Before you say it, I have confirmed with an engineer square that the towers are square and the build is rigid.
-
@th0mpy
Hi,
I lost ages back a delta (ex tevo little monster) and finally I get some good results changing almost everything of this printer.
I had three main problem. Tower twisted and bended and bar of different length and glass bed glued in the aluminium frame not parallel to it. I had to put some spring under it for make it parallel to the frame.
Also the end-stop are chinese optical sensor made with some plastic and they are not identical in the build. So obtain some of really accurate is impossible without changing the all frame and glass and end-stop. (I'll put instead efforts in a CoreXY with top components)
Recently I changed the belts from some chinese pure rubber to the Gate and I gained just half cent in the final calibration.(I could say also a more accurate repeatability in the 0.00x order) . Each belt of my tower are 160cm long.
I also removed the nema dampeners for make more rigid everything.
If your belt are not severely bad made can't them give such a difference in the calibration result. Must be some hardware tolerance somewhere.
I calibrate the belts tension by hands , so substantially are not calibrated.What it seems to me is that the tension of belts is putting you on the wrong way to look. Must exist some hardware issue somewhere else.
ps: my old towers at first look appeared like squared , but they where bended in the middle (and other two also a little twisted). (I used a 30cm aluminium square).
I just dismantled the towers and putted them one against the others and all the little imperfection come to be easily visible. -
@th0mpy have you tried mounting a circular ("bulls-eye") spirit level on the effector, to see if the effector tilt varies with XY position?
-
@dc42 I hadn't done more than just a simple test, the level was fine with a quick check. That said, I did not move it around and I'm thinking that would help me to uncover perhaps what @giostark had mentioned. I'll give that a whirl. One thing I was considering was getting new misumi extrusions and perhaps update the towers to use a few MGN9 rails. My thinking is that these are 1" extrusions versus 20 mm v-slot, meaning they're not meant for v-slot wheels. Using a linear rail might help this a bit.
But first, I'll check the tilt and determine if I have to tear it apart.
-
In case anyone was wondering. I was able to figure it out. After hours of teardown I determined that the geometry was off on the trucks that I was using on each tower; there was too much flex and it caused an excessive amount of bowing on the arms, resulting in effector tilt that I could not see when it wasn't moving.
I did not want to go back to the SeeMeCNC trucks so I finally found the, now defunct, trick laser trucks. I searched around the internets for a while and found the cad model for them and tweaked the mounting holes to have the proper geometry for Hayden's ball-cup arms. We're back in business.
The ripple that I was seeing before is completely gone!
-
@th0mpy I'm glad you solved it.
-
@th0mpy
Nailed , very good. Glad too !
The Rostock Max is pretty big . Did you have also the enclosure? (or are you assuming to build one?)
For my delta , 1m tall , I used the Haydn bars (40cm for 6mm diameter). Now are arriving new bars , 1cm diameter and a more heat resistant glue.
In this way I could have more rigidity and higher temp in the chamber. If you are interested , contact Haydn , he will prepare on purpose the same parts.