Bottom right arm behaving erratically

Bottom right arm limping,

Here’s what happend. Calibration went without any problems. Then, when I moved Maslow around the bottom right arm seemed to act different from the others. Spooling out way too much belt eventually resulting in an emergency stop. I swapped the arms around (bottom right became top left) but the results are the same. Calibration not a problem but when normal operation starts bottom right now and then spools out too much belt.
Here’s the calibration log and the log of the operation sequence that resulted in the emergency stop. Also an image of how the belt looks like.
Suggestions on how to proceed are more than welcome.

log4.rtf (47.3 KB)
log5.rtf (5.6 KB)

1 Like

This looks to me like there is something funky with the coordinates for the lower right anchor point maybe?

Like if the machine thinks that it’s much further away than it really is then it’s going to extend too much belt to try to reach it.

Can I take a look at your maslow.yaml file?

maslow.yaml (3.2 KB)

1 Like

Nothing off looking there.

It seems like there is maybe something funky going on because you aren’t the only one seeing similar things.

I’m going to run through the whole calibration process right now to make sure that it’s working right for me at least.

allright. To me it really was a problem with the lower right arm which made me think it was a hardware issue. But I double checked the arms and swapped them around. Having said that, could it be something on the controller board? I also noticed that pressing the z-axis up button sometimes resulted in the bottom right belt unspooling. Just relaying my user experience. Maybe it helps. For now I’ll wait for your input on how to proceed. Let me know it there is anything I can do in terms of user testing .

1 Like

This feels like a software thing to me, but I’m not quire sure where it would be. I think that something in the way the position is being calculated or the calibration process. I’m going to keep digging and I will report back if I find anything.

1 Like

OK I ran through the calibration process and found nothing…which is good, but also bad because it means that we didn’t really learn anything.

1 Like

I’d say the calibration process worked as it should. The “only” thing it does is measure the frame accurately. And the numbers it came up with are in line with my manual measurements.
What would happen if we would deliberately feed Maslow with the wrong frame dimensions. Wouldn’t it still operate (move) normally only with dimensions and shapes of the output warped? Or will it (while operating) also still triangulate its position and find out that its dimensions are off. And when it does, will it ignore/compensate/abort. And what is the threshold for that. Could we override this in a test setup?

If the dimensions are wrong the machine will feed out the wrong length of belts. If the wrong numbers are too small the belts will be slack, if the wrong numbers are too large the tension will be too high and you will get an error (or break something)

It will abort if any of the belts have an error that exceeds 15mm while running. We could override it if we want to.

What are the behaviors that you are seeing now?

I see from the file that the overcurrent threshold is only 1300. Perhaps the OC circuit for that arm on the PCB is a little low - could that explain the results?

Ill record a video

1 Like

Videos are super helpful :grinning:

Is the belt material bespoke or is it an off the shelf item?

Not sure on dimensions but it’s just a Timing belt for CNC I believe. Check Amazon and there are tonnes. When I get dimensions from mine I’ll be ordering spares as I’m seeing a few people having issues here.

1 Like

We should have spares on our website imminently.

It’s regular steel belted 2GT belt, but we tested all the manufacturers and the one that we are shipping is the best one so it’s better than the options on Amazon (although those will work)

1 Like

best in which way? less stretch?, better fit? other? (so we know what to look
for over time)

David Lang