Trying to use 7-factor calibration on Delta with Convex Deformation.

Hey, I have everything set up with my Kossel, save for just one last thing. I have a Delta Smart Effector, and when I go to probe the bed, the convex movement of the effector over the bed keeps me from being able to probe anywhere but the middle. I tried to make the effector lift 20mm from the bed after probing 0,0,0 so that it wouldn't scrape the nozzle against the bed, trying to scoot over there, but it didn't work.

Here's my bed.g. Is there anything I can do? Those G1 commands don't seem to work at all.

Quote

; bed.g file for RepRapFirmware, generated by Escher3D calculator
; 16 points, 7 factors, probing radius: 116, probe offset (0, 0)
G28
G30 P0 X0.00 Y116.00 Z-99999 H0
G1 S2 X0.00 Y116.0 Z20
G30 P1 X58.00 Y100.46 Z-99999 H0
G1 S2 X58.00 Y100.46 Z20
G30 P2 X100.46 Y58.00 Z-99999 H0
G1 S2 X100.46 Y58.00 Z20
G30 P3 X116.00 Y0.00 Z-99999 H0
G1 S2 X116.00 Y0.00 Z20
G30 P4 X100.46 Y-58.00 Z-99999 H0
G1 S2 X100.46 Y-58.00 Z20
G30 P5 X58.00 Y-100.46 Z-99999 H0
G1 S2 X58.00 Y-100.46 Z20
G30 P6 X0.00 Y-116.00 Z-99999 H0
G1 S2 X0.00 Y-116.00 Z20
G30 P7 X-58.00 Y-100.46 Z-99999 H0
G1 S2 X-58.00 Y-100.46 Z20
G30 P8 X-100.46 Y-58.00 Z-99999 H0
G1 S2 X-100.46 Y-58.00 Z20
G30 P9 X-116.00 Y-0.00 Z-99999 H0
G1 S2 X-116.00 Y-0.00 Z20
G30 P10 X-100.46 Y58.00 Z-99999 H0
G1 S2 X-100.46 Y58.00 Z20
G30 P11 X-58.00 Y100.46 Z-99999 H0
G1 S2 X-58.00 Y100.46 Z20
G30 P12 X0.00 Y58.00 Z-99999 H0
G1 S2 X0.00 Y58.00 Z20
G30 P13 X50.23 Y-29.00 Z-99999 H0
G1 S2 X50.23 Y-29.00 Z20
G30 P14 X-50.23 Y-29.00 Z-99999 H0
G1 S2 X-50.23 Y-29.00 Z20
G30 P15 X0 Y0 Z-99999 S7

So it brings it to just above the center point, THEN moves the probe to the Y tower, leaving no space for the probe to compensate for the higher point at the tower. None of those G1 commands actually do anything. It's as if it skips them.

Edited 2 time(s). Last edit at 03/04/2018 12:58AM by Lobotomy.

Ditch the G1 commands. In your M558 command, temporarily increase the H parameter to 30, or add H30 if it doesn't have an H parameter.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Why did you use G1 S2....? I'd expect pretty wild moves with this setting, except the S2 parameter is no longer used by RRF

@o_lampe: I was mostly just experimenting, because I couldn't find out which parameter changed the Z-probe probing height.

That ended up working fine, and it will now calibrate, but I'm running into another problem. Now the bed won't calibrate the deviation any more accurately than 3.8**. It also changes the Delta Radius, which it sets incorrectly, bringing the "raised bed at the towers" deformation back

Is there anything you recommend I do next? Next week, I'm getting aluminum milled vertices to see if it goes away.

Included below is a video of it successfully probing, and printing the result in PanelDue.
[www.youtube.com]

Edited 2 time(s). Last edit at 03/04/2018 04:11PM by Lobotomy.

DuetWiki says, you have to probe wayyy outside the virtual triangle of the carriers to get good 7-factor calibration.
The rod length will only change the scale factor ( if I got this right ), but effector tilt is encountered by 6-factor calibration already.
But there are some build rules to follow, before you can do auto calibration.

The video is too dark for me to tell what may be going on. Have you read [duet3d.dozuki.com] ?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

@dc42 The probe portion of the video really wasn't as important as the output on the panel. The beginning of the video is really just me archiving footage of what I'm doing, for my own sake. I have read the calibration tutorial, yes, although I'd never discount the possiblity that I missed something. I'll read it again after work.

The movement appears to flatten out if I change Delta Radius, but I was told that doing so leads to geometrically inaccurate/deformed prints, and isn't a viable fix? I know my rod lengths are spot on, each rod has its measurement stickered, and the difference in length between all of them is .01mm (288.12/288.13).
Now that I think about it, I have to wonder if there's a set order to adjusting Delta Radius, and probing. 7-Factor seems to change so much that my effector can now move outside the normal boundaries I had set when setting up config.g. Why would it change the M665 B parameter? Should I flatten the effector concavity with Delta Radius adjustments first, before calibrating, or after?

@o_lampe Do you mean to say that my virtual triangle boundaries need to be within the printbed, so that the effector has something to probe, or that I should set my virtual triangle to be larger, presumably through Delta Radius? Between Marlin and RRF, I've gotten a whole lot of conflicting info.

Edited 3 time(s). Last edit at 03/05/2018 04:45PM by Lobotomy.

Without seeing what is going on during probing, I can't make sense of the output on the panel.

Probing does not change the M665 B parameter.

I recommend that you check that your tower steps/mm are correctly configured, put the stickered diagonal rod length in the M665 L parameter, then try 6-factor calibration.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I forced the Smart Effector lights on, and tried the probing sequence again. This time, I did six-factor twice, and ended up with a comparable result of 3.9-ish mm of deviation, with no improvement upon the second pass:
[youtu.be]

This video shows the concavity:
[www.youtube.com]

Alright, prefacing the textwall with what I've done since my last post:
-Tested my current steps per mm values, and with my limited range of measuring devices, have determined that it's at least accurate to within 1mm, but I'll check with a nicer ruler come tomorrow.
-Each tower carriage screw is adjusted to within .05mm tolerance. I'd get more accurate than that, but it's clearly not my biggest problem right now.
-Changed to six-factor calibration. I reread the reprapwiki article on calibrating deltas with your firmware, and noticed that it said I'd need an oversized buildplate for it, which I do not have.
-A LOT of swearing and gnashing of teeth.

I have to buy a more precise measuring tool than what I'm using right now, but as far as I can tell, moving the printer down the Z axis by 50mm seems to equate to roughly 50mm in the real world, so I'm not sure if it's the Steps Per mm, I'm going by the original values for that parameter, included in the manual, which I recognize could've easily been wrong, thanks to the unparalleled engineering ability of the Chinese. However, I noticed that with Delta Radius, I was using an old parameter from Marlin, so I changed the radius to a far more accurate number to the real buildplate, although it didn't improve anything to a significant degree.

My day off is tomorrow, so I can devote all of that time to this. I'm just so confused, I guarantee that I've tried every damn thing under the sun about two billion times, here. The only parameters that make the bed more planar (Steps Per mm and Delta Radius), would end up making my prints either horribly out-of-scale, or deformed, from what I've been told. I'd be ready to just chuck the thing in the trash and go cartesian, if I didn't spend so much money on it. I've had this problem for 8 months. Definitely not worth giving up now, I'm just too stubborn for that.

Do you think it could be bad, or out of tune, stepper drivers? Maybe I didn't build it accurately enough? Do you think it'd be worth it at this point to get aluminum vertices, or am I just setting my money on fire here? I'll probably spend all of tomorrow rebuilding the frame, but at this point, I highly doubt that anything I do will make any impact anymore.

Edited 10 time(s). Last edit at 03/07/2018 04:32PM by Lobotomy.

Please post your config.g, homedelta.g and bed.g files.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Can do, and have done. They're all attached below. The bed diameter is 240mm, and the new magball rods are about 17mm longer (288.12) than the originals (271.5), if that helps any. I was told that it shouldn't actually affect the kinematics too much, provided that it's compromised for in software.

Thanks in advance, it would be nice to have a second pair of eyes on these files.

Edited 3 time(s). Last edit at 03/07/2018 10:00PM by Lobotomy.

Hi are you using radds or duet?

I ask as probe type for smart effector is usually p5 or p8.

Cant see much else wrong. But something is seriously amiss I'm getting 0.2mm deviation on a delta made from an old fishtank, with home-made magrod, I'm getting 0.004mm on my kossel Xl.

Is the probe consistent? If you just do G1 Z10, G30, G1 Z10......

30 times and then have a look at how consistent the probing results are.

Also try switching your M558 to p0 (manual) and use the manual calibration wizard (starts automatically on a G32 if there is no probe) and use paper and see if the results improve.

Edited 3 time(s). Last edit at 03/08/2018 05:58AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

1. In the bed probing video, the movement looks jerky, both the travel moves and the probing moves. It is really jerky, or is that an artefact of the video?

2. Are you using the carriage adapters supplied with the Smart Effector at the top ends of the rods?

3. Have you checked that your steps/mm calibration is correct? If you command the head to move up by e.g. 50mm, does it move up the correct amount, to within 1mm or so?

Also it would be helpful if you can autocalibrate once, then temporarily replace S6 on your last G30 command in bed.g by S-1 so that next time it prints the height errors. [Or copy bed.g to a macro file, edit the last G30 command in that macro, and then run it].

Edited 6 time(s). Last edit at 03/08/2018 09:46AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I spotted the problem. It's the H20 parameters on all the G30 commands in bed.g except the last one. Remove them and your problem will be fixed.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Well, wouldn't you know it, I can finally calibrate to within .039mm! Thanks! What did that H parameter do, anyway?

I guess now, all I have to do is add structural rigidity. I'll probably end up buying those aluminum vertices after all.

The h parameter in G30 denotes a height correction which can be used to compensate for effector tilt by measuring the difference between probed height and actual height if using an offset sensor. If they were all the same it would make no difference (only possibly to the computed height) but if one was missing the algorithm thinks one probing point was massively higher or lower than the others. You don't need these H corrections with smart effector or piezo probe.

Adding a H parameter to M558 sets dive height so the height the probe starts at.

Edited 1 time(s). Last edit at 03/09/2018 02:41AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions