Interpolation error

This is only local stepgen tuning and is correct.for a 1ms servo thread
(P = 1000 and FF1=1.0 )

It will show

Fat fingers.

So Chris, I'm using a 2KW servo, 9.55 Nm model SD-A3W2000V2I-HM, I'm sending the manual link (3.8MB). What P values do you refer to, the P of the PID or the P of the G64 P ...?


drive.google.com/file/d/1Nyh1VrkeqeCwwYh...s7s/view?usp=sharing

So Scotth, the machine was in X with 0.04mm and Y with 0.05mm clearance. I demonstrated, revised the axes and it was less than 0.01mm and still compensated with backlash

Can I use (P = 2000 and FF1 = 2.0) or (P = 3000 and FF1 = 3.0), or will it have no result?

I tried to use the halscope just to check these reversal points between the axes, but I'm a very beginner I couldn't analyze

Hi Ingo,
your servo has a lot of power.
Your problem may not be solved by setting 2 parameters
I think that your parameters for the control loop are not correct. Servo tuning in the open loop is a game of luck. There seems to be a PC software that you can use to tune the servo. I would try that first.
Parameter 113 and 114 is your auto adjust position gain level. There may be a scope in the software. So that you can see the follow error. It would be really good if the software can auto-tune.
Otherwise switch off the autotuning and set the parameters 115 - 119 by hand.

if that does not work
In your manual on pages 71-72 is the CN1 interface. You can read your encoder signal from the amplifier. Get a 7i85 card to which you can connect the encoder. So you have full control over this servo.
In any case, I would reduce the P in linuxcnc and increase the gain in the amplifier.

Draw a circle with G2X0.000Y50.000I0.000J-50.000 on paper. So you also have an overview of what is happening.

The halscope only makes sense when the encoder gives feedback to the nc. Otherwise you only see what the nc outputs and that's always good. only with the feedback can you compare feedback and command position in the halscope. (follow error)

I think that this is not a problem on the linuxcnc site.
It seems that the velocity control loop in the drive is not sufficient tuned. Something is osscilating or do you have an positional controller in the servo drive engaged? than its p could be to high and overshoot the target position.

Okay, I will test tomorrow and give feedback

I don't have a position controller. I made an external contour and it was perfect, could it be that?