caxis.comp - problems

Background posts:
forum.linuxcnc.org/10-advanced-configura...-order-matter#289826

forum.linuxcnc.org/10-advanced-configura...ling?start=60#288543


Homing and axis position commands are now working, although they're behaving very strangely.

I _think_ it's a PID problem now.  Once I changed PID from open-loop homing started working.

Behavior (either homing or an axis comand like G0 C180):

• Spindle immediately rotates in the correct direction
• Reaches target position and slows WAY down
• Continues to rotate past target (~90-180 degrees) at very slow speed
• Reverses direction and rotates back to target at initial search speed
• Nails target position with no hunting.

There's no vibration or oscillation.  If I turn P up past 0.2 it starts shaking.  If I fiddle with max_output, axis/joint max_velocity the amount and speed of overshoot changes... but the basic behavior remains.

Test settings
P = 0.15
I = 0
D = 0
FF0 = 0
FF1 = 1
FF2 = 0
FF3 = 0
maxoutput = 1
deadband = 0.1

Given my limited experience, I've not seen behavior like this.

Anyone have an idea why it's rotating past, slowing way down, then reversing to destination?

If this is a position loop with a velocity controlled drive, set your p=0 and then tune the ff1 so that you achieve close to the velocity that is commanded during a jog of the c axis. Then increase p slowly

It's a step & dir controlled servo.  Stepgen is set to velocity mode.

At the moment I can't really jog the c-axis.  I think the way the comp is written, when I jog it - at least using Axis GUI controls - it returns to the last position as soon as I release the continuous jog button.

I've not tried P = 0 yet with FF1, so I'll give that a shot.

After twiddling the PID values I've wound up with the following:

P = 0.2
I = 0
D = 0
FF0 = 0
FF1 = 0.008
FF2 = 0
FF3 = 0
maxoutput = 1
deadband = 0.1

[AXIS_C] and [JOINT_2]
MAX_VELOCITY = 360
MAX_ACCELERATION = 1000
STEPGEN_MAXVEL = 1000
STEPGEN_MAXACCEL = 2000
[TRAJ]
MAX_ANGULAR_VELOCITY = 500

It's still behaving strange.  C-axis MDI command accelerates very slowly, reaches the command position and stops.  No overshoot unless I bump FF1 or P values up.

But...
When it switches from spindle mode to position mode (M3 to M5), the automatic 'return to last position' function works perfectly at a speed and accel that appears to match the INI MAX_ACCELERATION and MAX_VELOCITY values.

I don't understand why an MDI G0 position command behaves different than the M5 automatic return.

andypugh wrote:

Halscope might help to see if the slowness is in command or response.
 

Which pins would you suggest I monitor in halscope?

There are a bunch available...
caxis.0.pid-in
caxis.0.position-fb-out
pid.c0.command
pid.c0.output
joint.2.pos-cmd
joint.2.vel-cmd
encoder.04.velocity-rpm

And some others I'm sure

subscribing, need to get this working on my lathe.

smc.collins wrote:

subscribing, need to get this working on my lathe.
 

In case it helps here's the hardware i'm using:

• Drive
  • Copley XSL
  • Step & dir command
  • Internally auto-tuned
• Motor
  • 3kw BLDC converted to servo
  • CUI encoder @4096ppr (16348 post-quad)
  • ~1.389:1 poly-v belt drive to spindle
• Spindle encoder
  • 600ppr (2400 post-quad)
  • 1:1 on spindle via HTD-3M belt

I don't know if this will be useful, but I've got two halscope captures

First is running the spindle at 500rpm and stopping it.  Capture is of the auto-repositioning move to C0 at M5.
 

Second is a back and forth G0 C0 to C90.  No settings changed.
 

halscope is zoomed out on the second one, but that's the only difference.

No idea why the moves are so diferent.

Notice how pos-cmd in the first capture remains constant while it changes in the second.
I have no idea really but it looks like two different modes of the custom component, maybe?