Hi everybody. Kossel, PLA, this is a simple test, I have a problem : X square’s side differs from Y one about 0.3 mm, the external side is 50 mm length, I have no idea how to fix that, the kossel has V-carriages with GT2 belts. The three belts are equally strained. Help me…
Did you cut the arms to length yourself?
Thanks you both. No, I bought diagonal arms already fitted, carbon fiber, to install uniball terminal we used a dima self made; for Mark, Kossel use three motors, sliding three carriages on three vertical opebeam, so I can’t set steps for X or Y axis, that’s why I don’t know how to fix it, but I’ll do as you suggest, making a higher piece, than we’ll meet here again, I hope
You should calibrate each arm - in the software, there is a translation of x and y to displacements of each arm. So there should be a way to set steps per mm for each arm’s Z movement. Run each one 10 x step/mm, measure how far off a cm it moved, and adjust.
Thanks, I’ll investigate about
Delta calibration is trickier than many tutorials suggest. When you calibrate for a flat bed on an imperfect build you introduce xy distortion. Make sure towers are parallel to each other and perpendicular to the bed. Do a calibration which computes independent radii for each tower
What’s ‘radii’? If do you mean delta_radius, how may I set an independent one for each arm? I modified that value to raise the center at the end of calibration, but I don’t know where to set it for a single arm, the three towers are good, as I can see with usually measurement’s tools, I don’t know if unreadable imperfections may cause relevant errors. Many thanks
There are calibration procedures that find a unique delta radius for each tower. it can be et in the .h file or on versions of Marlin that have settings fir independent delta radii. I’ll post a link to mine, but it is research grade .
RichieC has his popular Marlin changes. I believe it has options for limited optimizations, which might target the delta_radii without messing with rod length, etc. My version has delta radii, rod-length added to endstops in EEPROM, https://github.com/Mr-What/Marlin, my calibration routine is roughly documented at http://boim.com/DeltaUtil/DeltaCalibrationTools.html scrape to the parent folder to see all of my calibration stuff.
Many thanks, I’ll study all …
@Aaron_Birenboim dear Aaron, I read all the link you sent me, very interesting, you did a lot of beatiful work… your procedure is really util, but I haven’t a bedprobe on my kossel, I tryied to make on but with bad result, I think the tolerance of microswitch spring is more than precision we may have using the classic paper sheet method; I’d wish to know what kind of solution you did about… I’d like to have one like this http://www.pololu.com/product/136 , it is for medium/long distance, but it has a response curve useful for submillimeter distances too… what do you think about?
I mounted an endstop switch on my hot-end, http://birenboim.com/tmp/hotEndProbeMount.stl,
but every hotend is different… and for some reason, this drawing seems to lack the screw-holes to attach the endstop… but it has the idea.
I don’t seem to have any photos of the endstop switch mounted, but it is kind of like this one, with an experimental CNC probe mounted. https://lh3.googleusercontent.com/-zRA949DYo1Y/VKQ0d_V4c1I/AAAAAAAACKI/3bEGMZ5QaXg/w506-h900/20140924_085807.jpg
I found that a simple endstop switch works just about as well as the CNC needle probe I built for probing a flat print bed.
As you may have seen, my current procedure uses 255 readings, each reading 3x to check for repeatability of the sensor. Not reasonable with paper.
AFAIK, RichieC has commands to do limited auto-calibration, which can be used for a reliable independant-radii estimate. I think it is best to only optimize a few parameters at a time. It can give odd answers if you try to alter too many parameters at once.
Unfortunately, I’m not aware of a manual/heuristic procedure to estimate independent delta_radii. They tend to be numerical like mine or RichieC’s.
I do believe in the independent delta_radii adjustment. It is a very small change to the Marlin kinematic model, and seems to solve most people’s XY distortion problems.
just looked up the Polou part. seems to be IR sensor. Not accurate to sub-mm. Inductive sensors like Printrbot are popular. These types are popular, http://www.ebay.com/itm/LJ18A3-8-Z-BX-DC-6-36V-NPN-NO-8mm-Tubular-Inductive-Proximity-Sensor-Switch-/311116613389?pt=LH_DefaultDomain_0&hash=item486ffec70d they have better range than the smaller, flat one I have. For metal bed, or a thin (3mm) glass bed with metal plate or foil underneath.
@Aaron_Birenboim many thanks for your interesting about; and I wish you a happy new year… I read your page about calibration, but your script’s link point to an error page (500), may you send me that script? bye
It has not been polished for release, but the whole thing is available under http://boim.com/DeltaUtil/
If there are pieces missing, look under this folder for the file(s). Contact me if parts aren’t up there. I’ll upload them. A review of the main procedure is http://boim.com/DeltaUtil/DeltaCalibrationTools.html
as you may note, the basic/typical 4-parameter fit is from guessDeltaErr4(DP,bed). DELTA_RADIUS and endstop offsets. If you measure a test print, to estimate any residual XY distortion, you could run deltaCalXYZ() which will compute individual DELTA_RADIUS for each tower. You will need to edit a text file to enter the printed part measurements, which is a pain. So far, one calibrated, I have not had to re-visit the deltaCalXYZ() for months. However, after performing maintenance on the printer, I will re-run guessDeltaErr6(), and re-run my quadratic polynomial fit, which is a side-output from plotParabolicFit() at this time. Note that the quadratic bed-level adjustment is my own thing, only on my Marlin fork. It is kind of like the G29 bed-level probe, but more persistant, and stored in EEPROM. For me, it has remained valid until I do maintenance (like change the hot-end, or fan ducts…). Needs to be repeated after significant disassembly/re-assembly.
I could use any feedback you can provide to get this more complete for wider release. thanks.
Many thanks, just come back from a short holiday in Tuscany, now I’ll have time to do something about, I hope my tests will be useful to you. Regards
Posted a photo of my switch holder/probe, and a short video of the calibration probe being run.
Ihttps://http://plus.google.com/u/0/108203282175266795238/posts/M3xwJbg61cT I have not analysed the results yet, but data of the probe is at http://birenboim.com/tmp/bed108a.dat
I broke my glass bed yesterday, and this is a probe of the new bed