Velocity slow down and arced cut

Do you think the solution was going back to 1.03 or lowering the feed rate?

1 Like

can we get some debugging here.

please make a version of the firmware that tells us what power setting it is using, and what the various errors that the PID loop is trying to correct are.

If the motor is stalling out, the left motor should be at max power during the entire cut and the PID errors should reflect that it’s falling behind.

If this is the case, and slowing down lets it keep up, then we need to detect this case and slow our feedrate (or report that we can’t make the cut) I think slowing down should be at least an option if not the default

However, if we are not maxing out the motor power (which I suspect is the case, because it does catch up at the end), then we have something else going wrong in the PID loop that’s not detecting the error properly. This could be the kinematics calculations, or it could be something else.

I’m pretty useless for debugging right now because I’ve got everything all taken apart working on the frame build instructions :frowning_face:

Definitely going back to 1.03 . Lowering the feedrate using 1.06 followed the same path, even without the weight of the router

2 Likes

Excellent! (well not really, but at least we know where the problem is)

was there something in the PID loop or triangular kinematics calculations that changed here?

@bar it’s ok!! I’m perfectly fine W 1.03 right now😀 when I get ready to put this bad boy on your bolt frame may need then so I can run chains opposite direction on the motors

2 Likes

Here is the list of things we changed…unfortunately they all seem like culprits:

Would you be willing to share the .nc file so that I can try to replicate the issue?

actually, I see two significant categories of things to check

  1. PWM changes
  2. which pin the PWM is on

But before we start going and trying to revert any of these changes, let’s see if the power to the motor is being maxed out or if something else is happening

also, if 1.04 still has the problem, we may need to go back further.

1 Like

Adding an extra pulley near the motor and attaching a cord with a weight in place of the bungee should work

ttrapp2.nc (14.2 KB)
This is the slowed file, 20 on feedrate all around

1 Like

Thanks! I will update when I know what’s going on.

I’ve filed an issue on github as a place to investigate

just weird no one has really had a problem yet. Just makes me feel like its something on my end, maybe w install?

Maybe, but it feels like a bug to me. The fact that going back to 1.03 fixes it seems like it wouldn’t be an install thing

1 Like

Going back to 1.03, did you do any re-calibration or just load up the software and run it?

1 Like

@rjon17469 I ’m glad you asked that because I failed to mention. The chain re calibration pop-up did show up when I went to 1.03. So I auto recalibrated with no problems And was positioned in the middle of the work this time.

1 Like

I think you said you had moved the ‘Home’ location (must have, it extends off the sheet :grin:). To aid replicating the issue, where on the sheet was this object placed, or to ask differently, where is the XY 0,0 mark with respect to the object?

Also, when the issue occurs, do the horizontal cuts extend past the outline cut? Are the horizontal cuts in the correct locations, or are they staggered so that they stay within the curved outline? There is a lot of cutting in this file before the dramatic curved outline happens, I’m looking for an early indicator that the issue is present. Does eliminating the many horizontal cuts seem valid for testing, or are they a good indicator that the issue is present?

Finally, I think you indicated that using GC and firmware v1.03 work correctly. As @dlang pointed out, some major change since then has been in the PWM handling. The heart of those changes are in the file ‘motor.cpp’ .

My questions above are leading up to trying the v1.06 version of motor.cpp in the v1.03 version of firmware. I know that simply replacing the file Firmware-1.03/cnc_ctrl_v1/motor.cpp with the one from Firmware-1.06/cnc_ctrl_v1 works (will run the motors correctly), I have a routerless setup here ‘in the warm’ and have just run it. But it’s ‘wool socks’ weather out in the shop, though, so I’m looking for a way to do a real test succinctly.

1 Like

(Below images) Left cut has arcs 1.06, right cut 1.03 was perfect to within .5mm!!! On every single measurement! The piece of ply is 5ft wide positioned centered on the 8ft backerboard. Might give you an idea of my starting position.
As for the 14 horizontal pocket cuts, X was off shifting with the arc. So everything was skewed it seems.
If you want me to, I will put the 1.06 motor file you mentioned in 1.03 on my machine and do a test run to see if it singles it out as the culprit.

3 Likes

That’s just the info I was looking for. If you could run the v1.06 file in the 1.03 setup that would be very helpful. I’ll set up and run it here as well.

2 Likes