upgraded from 2.5 to 2.6 tuning issues

My mother board bit the dust so I upgraded to an ASUS J1800I which required Wheezy and 2.6

I updated all the 7i64 pins and fired up the machine knowing I had to retune.

The axis were much louder and on the edge of being unstable during jog and homing.
I loaded the tuning repeat program and started to tune. To get the machine stable the P values are 1/4 of what they used to be and the both the motor and plots are very noisy.

I tried setting pins pid.N.error-previous-target back to false like in 2.5 and this impacts the FF1 but everything is still noisy.

Is there something that I'm missing?

files attached

There should be no difference at all between 2.5 and 2.6 tuning wise except for the pid.N.error-previous-target.

There was a bug in muxed encoder firmware where encoder velocity was not calculated properly (scaling was wrong)
this would cause your D term to need to be updated if you used encoder velocity pin for pid.N.feedback-deriv
but your hal file does not seem to use the encoder velocity pins so I dont know whats going on.

You might try connecting/tuning the D term with pid.N.feedback-deriv connected (which should be superior
to the d/dt created velocity used by the PID comp if the pid.N.feedback-deriv is unconnected)

So just add the following for each axis?

setp pid.N.feedback-deriv hm2_5i23.0.encoder.0N.velocity

So just add the following for each axis?

setp pid.N.feedback-deriv hm2_5i23.0.encoder.0N.velocity

if your encoder channels match your PID component numbers it would be like:

setp pid.0.feedback-deriv hm2_5i23.0.encoder.0.velocity

setp pid.1.feedback-deriv hm2_5i23.0.encoder.1.velocity

setp pid.2.feedback-deriv hm2_5i23.0.encoder.2.velocity

Ok I will give it a try later today

net motor.02.vel-fb hm2_5i23.0.encoder.02.velocity => pid.2.feedback-deriv

added the above line to the hal file.

no help any other ideas?

was there maybe a filter change on the 7i48?

The 7I48 filters are hardware so other than component failure should stay the same

I would not expect _any_ tuning change going from 2.5 to 2.6 other that needing to set pid.N.error-previous-target. false

maybe something else in your system changed

Note that on the axis you showed in the HALScope plot you have no D term (while the other axis have significant D term)
net motor.02.vel-fb hm2_5i23.0.encoder.02.velocity => pid.2.feedback-deriv
will have no effect unless you have some D term

I tried D from .001 up to 1 and it doesn't change the amplitude of the noise I started with the Z axis because it was acting the worst.

I even tried switching the 7i48 over to cable power to isolate it from the machine power supply and put it on the computer power supply.


The plot shows a constant 300IPM move and as expected when PID output (pink) is high F-error (red) is also high. and when one is low so is the other.


What can be causing this?