Also calibration issues

So we’ve run into a few problems after several hours of trying to calibrate, nothing successful yet. Figured I’d mention some of this just in case useful info comes out of troubleshooting.

First problem was when trying to retract. We’ve never had a successful full retraction on any belt without numerous retries (max was something like 40). Some of those retries worked with hitting retract again but more often than not it required a full reset of the board.

We also would run into one motor that just wouldn’t retract at all, unplugging and plugging that connector back in seems to help.

AP mode drops frequently. We don’t have wifi available in the building though I may bring in a router so we can at least try connecting to that.

Twice we got the Maslow to finally reach a point where all 4 belts were retracted, then we’d do the extract step and hang it on the frame. The first time attempting to calibrate, it seemed to forget which mode it was in and it said we the belts had to be retracted first. The second attempt it did hit 90 points, which took about an hour and a half and then went into calculating. Unfortunately, after 45 minutes of calculating we had to give up.

We’re using a vertical frame, roughly 20 degrees from vertical. 3556mm wide, 2336mm tall.
We started with the V0.67 firmware files
I suppose my first question: should the retract step take multiple to dozens of attempts to get the belts to fully retract?

MakeShift Lincoln wrote:

I suppose my first question: should the retract step take multiple to dozens
of attempts to get the belts to fully retract?

No, but initially the belts were wound by hand and we’ve seen that extending and
retracting them lets the machine wind them a bit better than we do by hand.

It may also allow the moving parts to break in a little bit (hopefully they
won’t continue to wear a lot, but initially if there are a couple tight spots
this can smooth them out)

We have had a number of people who have had to go into the config and change the
max current where it gives up trying to retract from the default of 1300 1500 or
even to 1800 to get reliable retraction

but before you do too much else, just cycle it a few times and see if that

David Lang

1 Like

Thanks! We’ll start with increasing the current values to see if we can get a solid retract on all four belts.

Next question, though may be irrelevant if the calibration issues were related to the belt retraction…

Is there any indication of when the calculations are reaching an end? The fitness value fluctuated in a loop between ~0.67 and ~0.69 for a good 40+ minutes before we pulled the plug for the night.

1 Like

Yes, it should terminate automatically after a maximum of 200,000 cycles. What did it get up to?

I wouldn’t expect it to take more than 15 minutes to get there

I don’t remember what it went up to but it was definitely a loop. The calculating messaged started off with what I would guess to be a normal close-to-zero number but after a few minutes would eventually loop between that 0.67-0.69 range, including the cycle count.

1 Like

Now that sounds like a bug for sure

Alrighty, so we got it to retract/extend reliably once we set the current values to 1700.

We were able to go through calibration and even got it to finish calculating. We then entered the values in the settings and restarted. After that, we were unable to get the Maslow to move in any direction and instead just got “error: 8” in the console.

So I then made the mistake of hitting the slack button. After which, it wouldn’t let me hit “take slack” until I first retracted then extracted. A little irritating that I have to pull out the ladder to get up and disconnect the corner while trying to hold it by the belt in one hand but presumably we wouldn’t have to do that very often.

I was able to get a post calibration retraction after removing it from the frame but it seems each time it has to retract all 4 belts, it kills the connection and I have to restart the Maslow to reconnect.

I hit retract once again, extended the belts and hung it on the frame but the FluidNC page became unresponsive and I had to do the above all over again two more times. I’ve yet to get it to stay up long enough for me to get to the “take slack” option and so here I am again.

What should we be able to do after the calibration is successful, values manually entered, saved and restarted? Is there any logging/debugging I can do to troubleshoot what’s going wrong? Also, what is “error: 8”?

Edited to add that we’re still on firmware 0.67…0.68 gives us a trojan warning on attempted download and Windows Defender dumps it.

This sounds like an issue for sure. I’m not sure what could be causing that though.

A copy of what you see printed out here could help us to figure out what is going on:

Error 8 means that FluidNC thinks that the homing process isn’t complete which is retracting and extending the belts. When you are extending them are getting a message printing out that they are finished extending and hearing the fan turn off?

Well, baby steps at least. We got a solid retract, extend and belt tighten. We were able to get it to manually move left (clicked move left once, supposedly) but had a couple of issues:

  1. the bottom of the sled was lifting off the waste board as it moved
  2. the connection dropped and the Maslow went right off the waste board (from the middle) but we stopped it with the emergency stop button we added.

I suppose I’ll need to confirm that the arms are in the correct order…

1 Like

FWIW, the X-Carve version we have has no endstops. We move it via the interface, set it’s home position and move on. The Maslow at least has the current detection to know if it’s retracted a motor too far whereas the XCarve would just crash itself. That would be a nice feature for the Maslow since having to do the take down → retract → extend before we do anything at all and for every attempt to move it is a bit irritating. Especially for an 8’ tall frame where we have to try to hold the Maslow by it’s belt, while on a ladder for the first upper disconnect.


I think that this is an excellent thought on where to look. I have been trying to figure out what could cause the drop off the bottom of the board, and I think this is a good guess.

Edit: In the in person build day yesterday several people said that step was confusing and gave us some excellent ideas for how to make the assembly instructions better.