Velocity slow down and arced cut

Here’s what I see so far when I look at this issue:

  • My machine is set up as a bottom-feeder, so I chose to try the v1.03 motor.cpp in the v1.06 firmware framework. I made sure the PWM was set to the default 490Hz
  • I’ve simplified the trapp2.nc file by commenting out everything but the outline.
  • I’m plotting with 0.8mm pens instead of a spinning router bit. The green pen was plotting the v1.06 version of motor.cpp, the red pen with v1.03.
  • I’ve drawn a ruler-straight blue line between the two outside diagonals, anchored halfway between the two at the bottom end and at Y=0".
  • Both green and red lines are straight and parallel to the blue line below that midpoint. In fact, they are still parallel to the blue line until about y=4". The arc seems to start above that.
  • the red and green line are parallel along their length to less than 0.5mm, both arc equally. At the top end I measure both to be 5mm from their correct position.

Have to wait and see what others find, but provisionally the cause of the arcing doesn’t seem to be the changes to the motor controls.

4 Likes

That is excellent tracking of the issue @blurfl!! I it makes a lot of sense to me that this wouldn’t be an issue with the motor controls, but now we know for sure!

What’s next on the list to investigate? There have been a lot of changes since 1.03.

Do you replicate the behavior when running 1.06? (The normal version, not the modified version)

P.S. Cool pen holder! I’m not sure I’ve really seen a picture of it in the wild before :grin:

2 Likes

As a long shot I reloaded 1.06 firm and gc and this is what I got. I dont recall seeing the warnings before when I loaded but wasnt looking when I first installed

I think that ‘unable to find machine position’ is significant. that means that the forward kinematics calculation isn’t working

1 Like

The last image I posted above shows- warning older configuration version detected (14 instead of 19). Where is gc reading that info from (log)?

There is a version in the ini file

1 Like

GC keeps it’s version number in a file data.py, the firmware in Maslow.c. GC requests the firmware version number with a ‘B05 \r\n’ command, and GC adds its version number to the resulting printout as well.

1 Like

Definitely a ‘skunk-works-grade’ project :wink:

1 Like

Today I removed 6 pounds from my sled (became 18.5 pounds after removing) to see whether weight made a difference from yesterday’s curved cuts. It did not, the line perfectly matched the curved cuts from yesterday.

Then I re-rigged to the top-feeder setup and ran the trapp2 outline using firmware with both the v1.03 motor.cpp and the v1.06 motor.cpp. Both produce a perfectly accurate pattern. Ruler straight along the whole angled length. The difference from yesterday’s curved cuts was the change from bottom-feed to top-feed.
Tomorrow I’ll re-rig back to bottom-feed to make sure the change is repeatable.

5 Likes

I never touched the settings when switching back and forth from 1.03 to 1.06. From your findings with the top and bottom feeds, do you think its possible that there was an automatic switch to bottom feed when I switched to 1.06?

No, the switch involves some recalibration. It couldn’t happed without one knowing, and everything runs backward if it’s switched without moving the chains :slightly_smiling_face:

3 Likes

@blurfl looking forward to seeing the confirmation regarding the bottom feeding accuracy. If it is the same as your first test, which I expect it will be, then it looks like there is some tweaking that may be needed for the kinematics with the chain on bottom.

As one more test, could you try also drawing out the cut pattern lower in the work area, or say moved to the side? Trying to gather as much data as possible to attack this.

1 Like

@rjon17469, I’ll be glad to put it anywhere interesting.
As it’s a bit of bother switching back and forth, are there particular combinations of v1.03/v1.06 firmware and top/bottom feed that you want tested or compared? I’ll skip the uninteresting ones :grin:

3 Likes

It sounds like for you, no difference was noticed when using the v1.03 motor.cpp file, right? I suspect this is a different issue from what @clintloggins is experiencing, as your top-feeder configuration seemed to work well but the bottom-feeder didn’t. So given that, I would say let’s just use the pure v1.06 and go from there!

It may be worthwhile transitioning this to a separate thread also and/or making an issue on GitHub so we can ensure we don’t confuse troubleshooting efforts.

2 Likes

Good-o, start a thread and I’ll report there.

1 Like

A quick test today with 1.07 (reports1.08)
Straight as far I can tell.

2 Likes

Great @Gero ! This is with top feed?

1 Like

Yes, topfeed. However tried my files and the bows are back again. So be careful with weight and feedrate :frowning:

1 Like

did the warning for being unable to keep up make it into 1.07? @gero, does it show up in your logs when you have problems?

@dlang I answered in the post Straight line curves down at top of the sheet (partly solved)