Odd issue

Hello,

I performed a retrofit on a Kasuga VM-750 mill a while ago and never got around to fixing this one last issue. I can not seem to figure out what I need to correct if anything to solve a stability issue with the DRO readout.

1) When mill is at a standstill the DRO will bounce about .0001" on all joint.

2) When I am running a file the pink/purple/red line that shows where you have cut, bounces off the white line that represents the path needing to be followed. It appears it moves off and on the cut path a very slight bit. I have not been able to visually see anything in the finished part but I feel its there still since the DRO shows it happening. Could be wrong....

A bit about the equipment

1)5I25 with 7I77
2) AC servos and drives (MSI)
3)2000ppr encoders
4)1987 Kasuga VM-750

The old control system only went out 3 places (.000) and did not show any bouncing in the readout, but maybe thats because it only wen out 3 places and not 4 (.0000). Old controls was a Delta 10 setup.

The mill and equipment has been functioning well, just the readout bounces and feel I am overlooking something to fix this.

If anybody could help me out that would be great.

Here is a video of what it is doing.

A little servo dither like that is not unusual. There may be some fine tuning you can make to the servo tune or some other tricks that may allow you to reduce it below detectible levels, (but maybe not).
What are the counts/inch on the encoder for the axis in question? It may not be possible to reduce the dither below 2-3 counts and 5 or 6 might be more realistic.
Do you have a dead band set? It sometimes helps to have the dead band set for 1.5-2 encoder counts.

Thank you both for the responses.

I adjusted the dead band a bit with no luck. I changed the value back to where I had it originally. I am thinking that with the age of the drives, motor, encoders, and the lower ppr count on the encoders that this may be as best as it will be. Like I said the mill is running fine and I do not see anything in the cut quality that reflects this movement so I will just live with it.

I did however change from actual to commanded position and no more fluctuation so that is where its going to stay now.

Thanks again

0.0001" is under what is normally is expected.
That is not unusual and just shows the small amount that the motor is bouncing between two positions.
I have commercial (heavy/big/almost new) machines that also shows 0.001 (mm in this case) bounce.
From my experience when you try to create a dead band, you will get a worst dynamic response and have a larger following error (by definition the dead band is just this, making the system less responsive in a determined range). In resume the advantage gained is not valuable over the worst response.

Yeah when I adjusted the dead band the axis seems to become sluggish and not very responsive.

Well it looks like a fluctuation of .0001 is good for me then since its a 1987 machine....

Thanks everyone!!!

Really the only way to reduce this dithering is to increase the encoder resolution.
(not that this will actually improve machining accuracy significantly)
Your nice linear feedback control system pretty much turns into a
bang-bang crunchy mess when you are dealing with single digit position
error counts.

High end systems have up to 16M count/turn encoders now

Makes sense to me.....

I am going to leave it like it is. No need to change my encoders for a little dither on this old of a machine.

Thank you

Thank you for posting information about specific parts you used in your conversion. is there anything you would have done differently?

I will be converting the same mill soon. And documenting on this forum. it will be delivered to my shop this week and i anticipate the bandit3 controller will not survive the 70 mile trip even though it is working fine at this moment.

when i get it set up i will see if it has the same wandering servo issue.


-