How does my encoder feedback look?

I'm tuning the steppers after adding encoders to X, Y & Z axis. I have it working and I think this is about as good as I'm going to get. This is my first time with encoder feedback. Please take a look at the halscope graphs and let me know what you think.

The machine is a home built vertical mill with nema23 steppers, mesa 7i76e with the 7i85 for the encoders.
One of my concerns is the Y and Z axis will hum when they stop moving. This only happens once in a while and the hum is not very loud and I don't see any movement visually or with halscope. 

   

On the first and last plots, a tiny bit of FF2 may help
correct the acceleration dependent error.

On the middle plot and very small change of FF1 from 1.000
to say .999 or 1.001 may help the velocity dependent error.

Thanks guy, I tried both of the above with very minor improvements. I can center the error during the cruise part but I can't seem to improve the accel/decel.
I have been running all of my tests at rapid travel rate of 1500mm/min. Should I be happy with what I've got?

You should be able to make the errors during accel be above or below 0
by tweaking FF2

Is P as high as it can get without oscillation?

When everything else it optimized you can try adding as much I term as is stable
(watch out for large magnitude oscillations with too much I term)

PCW wrote:

You should be able to make the errors during accel be above or below 0
by tweaking FF2

Yes, I am able to move this.

Is P as high as it can get without oscillation?
No, I can increase P to 50-60 with no oscillation but it make it difficult to remove the following error.
I thought it was best to keep P as low as possible. I will play with higher numbers tomorrow.


When everything else it optimized you can try adding as much I term as is stable
(watch out for large magnitude oscillations with too much I term)

I'll try adding some I also.
Thanks

 

andypugh wrote:

charleyann wrote:

I have been running all of my tests at rapid travel rate of 1500mm/min. Should I be happy with what I've got?
That depends on the machine. That seems a bit low for plasma or a router. 
My old knee mill does 45 mm/sec which is 2700mm/min so you do seem very much in the low side of normal. 



Hi Andy, it's a small machine. 400x300x250 I cut mostly aluminum and some mild steel.
I can get 3000mm/min with no problem. I just prefer to keep it slower.

 

After a couple of hours testing this morning I found I can increase P to over 100 with no oscillation.
Setting P at 76 took care of the occasional humming from the steppers.

Here are my final results. No matter what I do I can't improve the accel/deccel

Z-Axis
Y-Axis
X-Axis