Another servo tuning thread... New rotary table

Hi all,

Trying to tune my new A axis, I'm not really sure if I'm at a good/reasonable spot or if I just don't know how to optimize this further.

I've attached a screenshot showing a a short move of the rotary at about 2/3rds of max speed. I've tweaked in FF1 to bring the following error down but I find that the error gets to be fairly significant at lower speeds and I'm not really sure what to do about that. Do I just split the difference with the FF1 value? or is there some way i can apply more correction at one of the spectrum that isn't FF1?

Also curious about the acceleration/deceleration humps.I can tweak them a bit with FF2, but the deceleration hump is hard to affect, also isn't it actually on the wrong side of the line? Shouldn't I expect the deceleration lag to be the opposite of the acceleration? Not sure if this is just a property of the big spinea unit I'm spinning up with a not particularly large servo motor or what, Inertia mismatch?

I'm also a bit lost on the I term, I played with it some but really didn't notice much difference.

If this is a velocity mode servo, i would first zero the D and Bias terms since they really make no sense. I suspect the relatively huge bias term accounts for assymetry in the accel/decell tuning

The integral term will not make any noticeable difference unless its _much_ larger

Yes, this is a 7i77 and it's in torque mode.

I can't seem to get a stable P over 60ish without a bit of D. I can remove D and reduce P and start again.

The bias was there to remove some weird asymmetry. I'm not sure what the source is.

No D for _velocity mode_

You _need_ D for torque mode

I would start with a low P term and then set D as high as possible without buzzing
then set P as high as possible without oscillations (no Bias no FF1)

Then set FF2 to null the lag/lead on accel/decel

Then lastly set I as high as stable then set to about 1/2 of this value

Thanks. I'll head out to the garage and give it a go.

Alright, I ended up at something like:
P: 100
I: 12000
D: .2
FF1 .213
FF2 .0008

Seems to work quite well. I'd have liked more electrical resolution, but I just don't have as many counts per degree as I do have counts per inch on my other axis.

That I term does seem a bit nutty to me but it seems to drive out any following error right quick.

For whatever reason the bias seems to have gone away. It seems like the Spinea gearbox takes a bit of a set when it makes a few rotations in one direction or another, It appears to come packed super full of grease from the factory, so I imagine grease mushing around into one side of the mechanism vs the other.

As a note to anyone else wanting to use a high ratio Spinea Twinspin drive as a rotary axis: So far the rigidity seems awesome. But there is a lot of drag. This might not be the gearbox you want if you're a hobbyist and trying to use a sub-HP servo or stepper. I have a 750w servo on a ts200 with a 169:1 gear ratio and had hoped to be able to get 3000rpm/169=17.7rpm out as an indefinitely sustainable top speed just in case I ever wanted to maybe turn something lathe style very slowly or whatever use case, but the internal resistance of the drive is quite high and running at 8-10 output rpm puts my 750w servo at 200-300% load, which is of course only sustainable for a few seconds. Perhaps the situations will improve some when it's 30c in the shop instead of 3c and the grease is a bit thinner, but the math suggest it won't make that big a difference.

The need for FF1 does suggest a velocity dependent torque
which might come from viscous damping