7i92/7i77 Yaskawa Servo

OK I will try to get you something today, I am not at the shop anymore. Thanks for putting in the effort on this.

Dan

Oh I forgot to mention I did put the ferror value at a high value. That is why the over run of 25.xx mm for a 1 mm incremental move. If I set ferror lower to like 1 I get the following error instantly. Again I will provide the requested files tomorrow. Hope that helps. Thanks.

Dan

See attached files, I am still trying to fully understand output scale and output min and max

The axis I am working with now has a rapid traverse feed of 40,000mm/min with an accelration of 900mm/(s*s). They also have a parameter, rapid traverse voltage = 9V.

The specs for manual feed are 10,000mm/min with an acceleration of 500mm/(s*s)

The old controller also references servo loop gain at a value of 1.8 and a stand still servo loop gain of 1.8, I am thinking I can use these as a start for my P value and FF1 in LinuxCNC. Let me know. Thanks.

Dan

Some general hints:

Unless you understand exactly what you are doing:

Do not touch the output limits, leave them at -10 and 10
Do not change the default output scaling, only change the polarity (from 10 to -10) if needed
to get correct feedback direction.

Usually the only reason to change the output scaling and limits is scale them in machine units
This is a generally a good idea (because it makes the PID scaling the same for inch or mm machines)
but probably more than you want to take on right now

When testing set both the min_ferror and ferror limits to say 25 MM or 1 inch

An approximate FF1 value can be calculated from your numbers 9V @ 666mm/s = 9/666 = .0135

(when I suggested 1 as a starting point I assumed an inch machine)

A P value of 1.8 is probably way too low, heres what a P gain of 1.8 means:

With a error of say 1 mm ( a huge error), the PID loop would only make a
velocity change of 1.8mm/S to correct the error

(oops thats wrong I was assuming that your numbers were in machine units, in volts, a P gain of 1.8 and a error of 1 mm
means 1.8V ( so a commanded speed correction of 133 mm/s would be applied, this may be about right or even too high)

So I am getting ferror with a trail off (see attached), this only happens in the negative direction of the axis.

can you post your _current_ hal and ini files?

looks like a hal or hardware problem

see attached, thanks!

in the hal file, remove these lines

setp pid.x.maxerror .0005
setp pid.y.maxerror .0005
setp pid.z.maxerror .0005

In the ini file the y output scale is 1 instead of 10 plus the
FF1 value probably needs setting as well
(FF1 can be fine tuned more easily if P is temporarily set quite low)
FF2 is used to get rid of the bumps during accel and deccel

Sorry forgot to mention I am only working with the X axis currently. The scope images were of the X in both directions, just questioning why the difference in the positive direction? Let me know your thoughts on the positive direction, again its only on the decel of the analog out. Thanks

Dan

The problem is the

setp pid.x.maxerror .0005

line

Edit:

This handicaps the P (proportional) part of the PID loop to only allow velocity corrections of
.0005 *10 = 5 mV maximum change in the output voltage on top of the output controlled by the FF1 term

Its actually rather amazing that it works at all and shows that FF1 is quite close
(since FF1 is doing pretty much 99.8% of the work)

The difference between directions is probably due to a few mv offset in the drive and 7I77 output