When cutting sled, it just wandered off

GC/FW 1.02
I didn’t see any “Buffer overflow” errors, but when I was air-cutting the sled (temp sled, no bricks), it would be going fine and then would “wander off” severely during a cut:
Home was at: -25.75, -16.31, Manual Z Control, Rectangular config.
Line of NC where weirdness happened:
56 - crosshairs diverged then router came back from the error
156 - crosshairs diverged then router came back from the error
1851 - crosshairs diverged, bad cut!
2047 - both combined for a moment off path. then divergence – this would’ve been a terrible strain on the router if really cutting (3/4" in, going a path that it had never gone before)
sledB.nc (77.5 KB)
log.txt (1.4 MB)

What else can I give you/should I try? I just had the apostrophe that I could probably zip-tie a couple of “bricks with holes” to the router – would that help?

1 Like

Your feed rate looks like it set to 40 inches/minute. I found on mine the motors can’t reach that speed and it will create problems like what you are seeing. Try slowing down to 30 inches/minute.

I agree with @Bryan_Pollock about the feed rate… Slow it down to 30 ipm and try again. Also, just as a FYI, the default safe height in makercam (0.5 inches) is, in my opinion, too high and you waste a lot of time waiting for the Z-axis to raise. I usually set it to 0.1 inches. Just little tricks to help speed up your cutting :slight_smile:

I see several Buffer Overflow in the log file :thinking:
Surprisingly they show up in identical pairs. This is not caused by amount of digits, so the feedrate seems plausible.

It looks like the manual Z-adjust issue that is fixed in FW1.03 Buffer overflow when using manual z-axis · Issue #366 · MaslowCNC/Firmware · GitHub

1 Like

Ok… will try FW 1.03 and 30 IPM and get back to everyone

1 Like

This looks like the error that we fixed in 1.03 to me…sorry for the trouble @Thormj

1 Like

I wish I’d hit the lottery so I could buy everyone a Z-axis kit :slight_smile:

1 Like

I have a Z-Axis kit, but my 3D printer is down so I don’t have a good way to mount the motor to the router yet…

Update: Still has overflows, still wanders off:
30 ipm, new firmware, old groundcontrol
Revised NC code (took out the countersinks on the sled): sledC-30.nc (49.5 KB)
Log: log.txt (1.0 MB)

Is there a way to up the time on the zero-chain-lengths -> Extend chains?
My Right-Chain tends to nautilis itself but it’s an effort to jump over there before it starts.

Crap- here are the events:
888 -Crosshairs diverged and it looks like GC got way ahead of the hardware (it was telling me to adjust Z while it was still moving to the hole)
1117 went way off course, then corrected by 1124
1312 way off course - buffer overflow corrected by 1324

– The first was at linear rapid move. The other 2 were at cutting depth on the circle…
saw other overflows, but didn’t seem to matter…

I am able to replicate the same issue that you are seeing. I’ll start digging into it right away to see what I can find out

1 Like

I’ve re-opened the GitHub issue where we were seeing it here:

1 Like

However… upgrading to GroundControl-master fixed it. No overflows visible, no bad cuts.
Log: log.txt (616.9 KB)

Sorry about not testing it earlier - had to get Python fully operational first,


Great! I found that it is fixed on master too. When I was able to replicate the problem I was on a branch which I hadn’t updated :roll_eyes:

I’m sorry for the trouble, the fix should be rolled out to everyone tomorrow!

1 Like

If you are seeing buffer overflows, all bets are off.

You need to get to the newest version that fixes some bugs with manual Z
handling before anything else is valid.