7i92/7i77 Yaskawa Servo

Hello everyone, I can not thank everyone on here enough that puts the time in to help other users, especially me! Greatly appreciated.

I have some questions about the setup of my configuration/retrofit. I am using Yaskawa Sigma 1 drives for velocity control. I know the servos themselves have PID settings within the drive. I read elsewhere that the setup I have is a loop within a loop. The servos have their own PID loop for velocity and then the LinuxCNC loop is for positioning? With that being said will I have to tune for the positioning loop? I have also noticed that once I enable the system the axis starts to runaway, but very slow and nothing violently until I get a following error. Encoders seem to be working because I am getting feedback on the DRO and I checked with HALscope. Just need direction of where to check first for the runaway issue Thanks!

Dan

For velocity mode servos, the current (torque) and velocity loops are in the drive itself and the position PID loop runs in linuxcnc

If the drives only creep when linuxcnc is in the machine-on state (so you have the drive enables wired properly),
this sounds like the PID P term is 0 (and yes you have to tune the position loop)

You can adjust the PID terms live via the calibration screen in axis ( Machine --> Calibration )

gnipsel.com/linuxcnc/tuning/servo.html

(thanks JT)

Note you may find that the drives immediately run away as soon as you enter a non 0 P term
this means that you will have to reverse the analog out polarity in HAL

Ok so maybe I have an issue with my wiring.....I am trying to duplicate the wiring from the original spec but using the I/Os on the 7i77. With servo on and enabled I disconnected the analog out completely but still get the creep. I do question the original wiring from the manufacturer as they have the wiring reversed for the /SON circuit.. I attached the schematic of the /SON circuit from Yaskawa. Let me know. Thanks.

A little creep with the drives powered and enabled, but no control input is not unusual and maybe to be expected. A little fine tuning of the drives internal tuning parameters should be able to zero that out, but it shouldn't be necessary. Once the position loop is connected and tuned in Linuxcnc, it should straighten it out.

On your drawing, is ENA0+ connected to ICN-40 and ENA0- connected to 0V?

( That would be the correct wiring for a active low enable )

If you disconnect ICN40 is the drive disabled? (creeps stops and motor shaft turns freely by hand )

Now I seem to get nothing. 0V can be sourced from where? I take it I can not use the field voltage ground as 0V? Thanks Peter always a great help.

Dan

If ICN-47 comes from field +24V then yes ENA- goes to field gnd

Ok wired to what we discussed now I am back to beginning issue of axis creep, yes P value is at 0. I am confused as to why it still creeps even if I disconnect the velocity reference? Thanks Peter!

This is expected and harmless.

Expected for a couple of reasons:

1. With a analog velocity command, there's always some offset voltage so the commanded velocity is never exactly 0
2. Even with a digital command its not easy to detect very low velocities accurately so random creep still occurs

Harmless because when linuxcnc is not running and in the machine-on state, the drives should be disabled and cannot creep.
When linuxcnc is running and in the machine-on state, the position PID loop is closed so the drives cannot creep.
(This does require a non 0 P term of course)

( this is also a safety issue so its very important that the drive enable work )

Ok so a -10 value on the P seemed to solve the issue but my DRO is reading opposite of the command, this all sounds like maybe I need to switch A and B channels on the encoder or invert analog out??? Let me know. Thanks.

Dan