Resolved: V0.85 calibration step 8 randomized?

Software
1

Did the step 8 of 9 cut test get randomized?

I’m seeing some really wild swings in where the cuts are being made, also they start to cut more and more at an angle. Usually the cuts are pretty close to each other, and offset by a little bit, and horizontal to the floor but these are off by a lot and each cut appears to angle clockwise more each cut.

Typically I am good to go by the 3rd cut at 605ish mm, but I’ve just done the 3rd cut and its 49mm difference between the horizontal and vertical measurements.

The difference in distance from the top left of the first cut to the second was 305mm. The difference in distance from the top left second to third cut is 85mm. This is not the behavior I’ve seen previously.

I noticed this “One Machine Position Report Command Misread” every so often during the calibration cuts, here is snippet from the command line output:

Sending: G40  
Sending: G0 Z5  
Sending: G0 X0 Y0   
Sending: G17  
Sending: G0 X36 Y-36   
One Machine Position Report Command Misread
Sending: G91  
Sending: G0 X-300 Y300   
Sending: G1 Z-7 F500

It finally got to the same distance horizontally and vertically (615mm), it took 9 re-cuts to get there.

Also now that I think about it, the previous step moved the “center” when the left and right chains are let out and you hook up the sled, but it was well below and to the left of where the center should have been.

The cuts are angled a little like they are twisted clockwise by a couple degrees. I don’t have time to cut anything tonight to see if anything adverse happened, but if anyone has any ideas on what could possibly be going wrong, please let me know.

The cuts used to stair step down and to the right, but my cuts were:
#1 in the usual place, squared off from the frame
#2 way left and way down from #1, also angled clockwise a few degrees
#3 up and right of #2, still angled
#4 up and right of #3, still angled
#5 up and to the right of #4, still angled
#6 down and to the right of #5, still angled
#7 down and to the right of #6, still angled
#8 down and to the right of #7, still angled
#9 down and to the right of #8, still angled

2

You’re using v0.85 for Firmware and GC both, yes? I just ran the Step 8 sequence in millimeter mode and zeroed in in three cuts, stair stepping down to the right. Does your sled go to the expected place when you ‘Return to Center’?
I’ve noticed those occasional lines about “One Machine Position Report Command Misread” too, but haven’t looked into them yet. There may be a clue in log.txt about what it refers to.

3

Yes, 0.85 for Ground Control and Firmware.

No, return to center does not go where I expect, it is left and down of what I would measure as the center.

Doing a cut now, and its totally slanted to the right, similar to how the calibration cuts were angled. After this cut I’m probably going to do the whole calibration process again.

Did not notice anything that stuck out in log.txt

4

Do you have a way to mark the topmost link of each chain when you recalibrate them? I wonder if one of the chains is jumping the sprocket. Marking them then observing where they move to when returning to center would answer that.

5

When I’m done with the cut I’m doing now (at 30% and about 3 hours in) I’ll do a full re-calibration and see if I can’t resolve things. Also I’ll do as you suggest and mark the chains at center.

6

Mark the chains when the motors stop running at the end of the ‘Automatic chain cal.’ step, before you ‘Return to Center’! Having marked at that spot, you can use the ‘Advanced/Manual chain cal.’ if you ever need to re-cal the chains and want to skip the part where the motors measure out the chains. It’s got to be after the motors measure out the chain and before you move the sled, though.
Also, do watch the chains for any jump at the sprocket during calibration. It’s as boring as anything to watch, but crucial to guard against. Fingers crossed,

1 Like
7

3 hours into my cut (about 50%), and something decided to reset my home randomly and start cutting into a space that was meant for future cuts, but at the place in the gcode where I had been. The bulls eye seemed to think I was were I was supposed to be, but the maslow was a fair bit right of that.

So… that cut is ruined.

I’m re-doing the calibration again and on my third round of calibration cuts and it is looking more like it should. Made sure to mark the chain as you suggested after the chain was let out.

I guess I can say things are resolved, because the calibration is working as I expect. Note to other maslonians, if your calibration is going crooked, stop and retry because it won’t go well for you if you keep going.

2 Likes
8

Good advice. I’m glad things are going well now!

9

Also I thought I would be clever and use painters tape to cover the calibration cuts, so that if the cut went over the same place I would still be able to tell where the cut happened. Turns out it was a bad idea because it gummed up the end mill with tape goo.

10

I recall in Beta testing that the GC has a finite limit on lines of Goode. If you go over, it will stop or do other things bad to your piece. 3 hour cut at 50 percent seems big. I don’t remember the limit, but @bar or one of the code readers like @blurfl can find it in the code I am sure. May be part of your issue, that is what happened to one of my pieces.

1 Like
11

Thanks, I realized the 1.4" bit I was using could plunge much more than I was so there were way more passes tha I needed. My 4th attempt at this piece I upped it to 7mm plunge per pass, and its finally done successfully!

4 Likes
12

I believe that it’s 50,000 lines right now

13

Should not be near that then.

14

Do you hold it all in memory? Seems like you should only be limited by your biggest drive’s maximum file size.

15

Yes, that’s the current approach. An opportunity for a contribution to the effort, there!