Servo Oscillation Issue at ±6.6V 5i25 + 7i77, Analog Control

That looks like maybe an electrical/wiring issue with the encoder.

Is differential signaling used?
 

Yes, it is a differential signal.
I am attaching a screen manual from the servopack.
 
 Two servos are Delta asda-b2 750w and one on the z axis is the Chinese a1-svd 750w and in total there are such symptoms on each axis.
The encoder cable is shielded and a separate analog cable is also shielded, the shield is only connected to the servopack.
Maybe I need to use resistors?

 

Is the 7I77 jumpered for differential inputs?

It may be a noise issue. One way to check if it's a common mode noise issue
(noise that exceeds the common mode range of the RS-422 receivers)
is to enable encoder error reporting:

setp hm2_5i25.0.encoder.00.quad-error-enable true

Once this has been done, inputs that violate the quadrature sequence
(often noise will do this) will set the quadrature error flag:

hm2_5i25.0.encoder.00.quad-error



 

It checks your method and shows no error. I will try to turn on linuxcnc and control from the servopack to read a higher speed and exclude the encoder reading.
Because it still seems to me that it is not related to the encoder reading but to the analog power supply.
What still surprises me is that if I drive the axle in the positive direction, I can achieve 10 m/min, and if I drive the axle in the negative direction, 8.3 m/min.

It sure looks like an encoder issue, especially the speed limit
(in one plot the encoder velocity dropped to 0 in basically an instant)

Did you set the encoder filter to 0 on all encoder channels?

No, I set it on one axis I was testing. Should I set it on all axes?

I changed the encoder filter to 0 on each axis, it had no effect. I wonder if I can somehow drive the servo without a closed loop and see if it works.

Yes, if you decouple the linear axis (and comment out the net statement that connects
the PID output to the 7I77 analog output pin, you can simply setp the output voltage
with halcmd.
 

After a long break I'm back to the topic. I did some research, tried to swap the wires between the mesa and the servopack, but no changes. It's not a problem with the analog voltage as in the topic because if I set P=0 I can normally drive at high speed without a closed loop.
What should the encoder accuracy be set to?
Maybe I have set the scale too large or too small?
But I tried changing it to less and it didn't have any effect.
I'm still wondering if the analog wire can go parallel to the encoder wire, maybe it creates some interference.

Yes, looks like an encoder issue of some kind (especially the sudden drop to 0 velocity)

I would try lowering  the emulated encoder output scale, as
maybe the drive has an issue with quadrature generation at high
speeds and high  resolutions.