mesh calibration issues

[klaubauf]

hello.

i have a modded rostock max v2 with some bed leveling issues that i was hoping for some insight on:

first, the mods:

duet wifi
tricklaser ball end arms
tricklaser aluminum effector plate
tricklaser trucks
tricklaser fsr build plate
e3d v6
bondtech extruder
johnsl + fsr x3
mahackney inspired fsr mounts
pei build surface (sanded)


despite my auto calibration being something like 0.090 i get these terrible results on bed mesh leveling.

does anyone have any idea what might be up? i've checked for belt slippage, binding in the fsr housing, fsr sensitivity, johnsl reponse, bearings and so on, yet i always seem to get some permutation of this sinewave readout with ~1mm spread

thanks in advance for your help!

[dc42]

The ridges and valleys running along the X direction mean that the height of the effector depends on whether that XY point was approached from the +X or -X direction. This could be caused by loose or sticky joints.

The big drop at the right hand side is odd. It could indicate that the FSRs don't trigger as well over on that side. If you are using steel-cored belts, it could mean that the steel core of a belt has fractured (this is common if you run steel-cored belts over small pulleys).

The curve along the rest of the X direction is probably caused by auto calibration trying to reduce how bad the errors at the extreme right are.

[rootboy]

Observe your hotend as it probes each point. Do you see the hotend sliding horizontally any? Or does it make a quick tap and moves on to the next one? If you see it sliding, then it's your FSR mount causing the problem.

713Maker makes aluminum mounts for the JohnSL FSRs. This is what I use on mine. Even so, they can hang up as well.

If I had it to do over again, I would attach only two of the plungers to the bottom of the bed plate and let the third one float. I may end up doing just that eventually since you only need two of them fixed to keep the bed from shifting.

https://713maker.com/rostock/rostock-ma ... -mount-kit

There is also the Duet Smart Effector that would take cup binding out of the equation. If you have the stock plastic ball arm mounts, then 713Maker has an adapter for these as well.

https://713maker.com/rostock/duet-smart ... or-adaptor

I'm debating about going with the Delta Smart Effector but I haven't heard anything on their pluses and minuses. I'm skittish about putting a PCB that close to the business end of a hotend.

How about it David, what's it's track record so far?

[dc42]

I'm debating about going with the Delta Smart Effector but I haven't heard anything on their pluses and minuses. I'm skittish about putting a PCB that close to the business end of a hotend.

How about it David, what's it's track record so far?

The feedback we have from users of the Smart Effector is generally very good. A few users who use 40mm heatsink cooling fans instead of the E3D 30mm one have had issues either with the magnetic joints stopping the fan from working or with the fan interfering with the touch sensor, cured in both cases by moving the fan out a little.

Why not post a question at https://forum.duet3d.com/category/13/sm ... a-printers asking users what they think of it?

[rootboy]

I just signed up there the other day, so no reason.

But I thought it would be a good way to showcase a solution here that I wasn't even aware of until a few weeks ago.

After I got medieval with the plate the other day, my alignment is much better with my FSRs, still not great however. I will probably order one here shortly.

I'm using the ball and socket SeeMeCNC carriage, so magnetics is not an issue. And I have the adapter ring from 713Maker, so I should be set.

[dc42]

SeeMeCNC has their own version of the Smart Effector now, and that may fit your ball and socket arms better. The Duet3D one is designed to be used with magnetic joints.

[klaubauf]

thanks for the input dc42 and rootboy!

things are improved quite a bit after digging deeper into FSR response. the sensor at Y seemed to have a dead spot in the center for some reason while the outside was much more/equally responsive to the other sensors. i replaced it and things seem to be significantly better as shown below.

i'm not understanding that front dip, though. i've rotated the glass, observed normal tap/hop movements during probing at the coordinates, checked FSRs (more thoroughly this time). would that likely be binding that occurs at the FSR mounts of X and Y only when hitting the edge of the plate? there's a similar phenomenon at the Z tower although not as drastic.

[rootboy]

Anytime!

I de-attached one of the plungers and things are looking much better...

[rootboy]

SeeMeCNC has their own version of the Smart Effector now, and that may fit your ball and socket arms better. The Duet3D one is designed to be used with magnetic joints.

Good to know, thanks!

After knocking one of the plungers loose (the Z axis), the map has improved immensely. What I am thinking about trying now is to detach all of them and use bicycle tire patches as my mounting medium to the bottom of the bed. I'll glue the patch to the top of the plunger, and then peel off the backing and stick the patch to the bottom of the bed, This should give it the tiny amount of lateral play that it needs to move freely.

[klaubauf]

another previously undiscovered issue has been resolved and my mesh results are the best i have ever seen on my machine. the issue was slightly bent arm (the one the holds the eccentric nut) on the Y trick truck carriage. it seems that at certain coordinates, the carriage would lift away ever so slightly from the linear rail. i bent it back into flatness with gentle hammering and a board, then tested against a 1-2-3 block. results below:

[dc42]

The X-direction ridges and valleys at each end suggest that you still have a small issue with backlash, probably caused by either a sticky joint or play in asjoint.