Calibration Issues Maslow 4: Calibration Fitness Too Low

[Markus_Rinn]

I now tried to make a dry run of the maslow logo. That worked first like a charm but then stopped suddenly. I got no error message in the browser, but a red led (LED2) started blinking.

I made a video, and I will upload it and post the link here (private video).

I hope that helps. I’m tired (22 o’clock in germany).

Bye Markus

[Markus_Rinn]

Red blinking LED

[bar]

I think that we want to get through a much larger calibration grid before cutting will work reliably.

The default grid is 10x9 points and 2000mmx1000mm and I would really like to get that working for everyone

[ronlawrence3]

or 9x10 and 1000mmx2000mm for those of us in portrait mode

[ronlawrence3]

@smarcus3

the calibrated size is about 200mm off from what I can measure with a tape measure

FWIW, I have this same difference (about 200mm difference between measured bl and br anchor points vs a “good” calibration).

[smarcus3]

Both of these led to huge amount of lose belt which almost destroyed my belt. Both were way too scary to run.

[bar]

I’m working on a fix which should eliminate the lose belt during issue, I’m hoping to have something working tomorrow.

[bar]

@smarcus3

Both of these led to huge amount of lose belt

Was one of them this one:

@bar

OK, here is phase two of this.

This is the earliest prototype of the system to eliminate the extra slack during calibration

[smarcus3]

Yes the two that were requested to run earlier:

Interstitial Firmware Releases

Here is 0.68.1 which works to improve the math behind the calibration process. It includes a re-write of the algorithm which does the computation (not 100% sure it’s actually different, but I think that the code is more clear) and a new system for making the initial guess used to begin the computation. I believed that the initial guess was not important (in theory it shouldn’t be), but it appears to be important so now instead of using a random hard coded initial guess we are using the existin…

Help requested: Improving the Calibration Process

OK, here is phase two of this. Now we have an initial small 2x2 calibration pattern which is done with the belts slack, and then that information is used to get a good enough sense of the frame dimensions to do a slightly larger 4x4 grid which is then again used to run a slightly larger 6x6 grid while the machine automatically keeps tension on the belts. Basically at each of the red circles we stop and recompute the anchor point locations and improve our understanding of the world. Then we use…

The first was a bit scary the second one quickly went off the rails with similar effects at the second point with over current warnings which I’ve never seen before. I have a video of the first which is still uploading. Will post again when its finished.

[bar]

Thanks for the video, that is super helpful. And thanks for trying those, that feedback is really valuable.

[smarcus3]

0.68.1 Video:

New video by Steven Marcus

[bar]

Oh god, something is going wrong there for sure. It shouldn’t be extending nearly that much belt!

[smarcus3]

Yeah it’s never done anything like that before prior to these two versions. Willing to try anything new to try when something else is ready.

[Markus_Rinn]

WOW! I’m really impressed by the progress. I just made a calibration, only on a very small 300mm x 300mm grid with 3 x 3 points. Now for the first time my measured frame dimensions fit to the measurements found by the calibration. Measured with a laser 4260mm x 2490mm.

LOG:

Computing… This may take several minutesFitness: 1.2744082 in 100
Fitness: 1.3262010 in 200
Fitness: 1.3703707 in 300
Fitness: 1.4141947 in 400
Fitness: 1.4543760 in 500
Fitness: 1.4913931 in 600
Fitness: 1.5270815 in 700
Fitness: 1.5600575 in 800
Fitness: 1.5870276 in 900
Fitness: 1.6110889 in 1000
Fitness: 1.6296933 in 1100
Fitness: 1.6416067 in 1200
Fitness: 1.6500619 in 1300
Fitness: 1.6563623 in 1400
Fitness: 1.6586824 in 1500
Fitness: 1.6586824 in 1600
Fitness: 1.6586824 in 1700

Calibration complete
Calibration values:
Fitness: 1.6586823228585335
Maslow_tlX: 5.3
Maslow_tlY: 2490.0
Maslow_trX: 4260.4
Maslow_trY: 2490.4
Maslow_blX: 0.0
Maslow_blY: 0.0
Maslow_brX: 4273.7
Maslow_brY: 0.0
These values have been automatically saved for you.
You MUST restart your machine for them to take effect…I know that is annoying, it’s getting fixed ASAP.

Now I will go for 550mm x 550mm with 5 x 5 points…

[Markus_Rinn]

550mm x 550mm with 5 x 5 points:

LOG:

Computing… This may take several minutesFitness: 1.0772704 in 100
Fitness: 1.0772704 in 200
Fitness: 1.0772704 in 300

Calibration complete
Calibration values:
Fitness: 1.0772703391602527
Maslow_tlX: 6.2
Maslow_tlY: 2490.0
Maslow_trX: 4260.4
Maslow_trY: 2490.4
Maslow_blX: 0.0
Maslow_blY: 0.0
Maslow_brX: 4273.7
Maslow_brY: 0.0
These values have been automatically saved for you.
You MUST restart your machine for them to take effect…I know that is annoying, it’s getting fixed ASAP.

Does the fitness “test” end, when it fits three consecutive times the same value? Because this time it took only three iterations.

[bar]

Thanks! It does feel like we’re making slow but steady progress. I’m sure we’ll have it working smoothly for everyone before too long.

Thanks for your patience!

[smarcus3]

What version are you running so I can retest it as I haven’t received positive results yet other than very small configuration patterns.

[Markus_Rinn]

The newest one: 0.69!

[Markus_Rinn]

Yesterday with the interstitial firmware 0.68.1 my fit was above 1. After calibration I made a video of the dry run cutting the maslow logo. It made some weird crackling noises and shortly after the video it stopped with the red light blinking. Here is the link to the video:

calibration with firmware 0.68.1

[Markus_Rinn]

Some questions (slightly unrelated to the topic):

• During calibration I found that even when all belts are tight one motor seems to spin. Looking closer I saw it was only wiggling around. Is that normal? Is that part of the measuring process?

• I set all my anchor points some cm to high, because when I measured the height of the spools my z-axis was not completely down. I made the calibration with the z-axis all the way up. Is it possible to insert negative values for the z offset of the anchor points in the yaml file? Then I can rerun calibration with the z-axis all the way down.