Following Error When Homing

Hal file:

Encoder resolution on X and Y scales is a bit on the course side, 125 pulses per mm on X and 250 pulses per mm on Y. Z has 800 pulses per mm.

That looks normal function order wise

could you try with

setp pid.x.error-previous-target false
(this will change the tuning)

Otherwise to chase this down will probably require halscope
and a wide following error setting so a following error is not tripped

Then trace the position, following error, PID out and index enable at
say 2 servo thread periods per division when homing (single trace triggered on
index enable falling edge)

This is a Heisenbug. As soon as i attach the HalScope, those axis/joints I watch do work as supposed.

Got it. Don't know what to make of it. Following error explodes one servo period after index-enable is 0. Second screenshot is from second attempt, when it works. x-pos-cmd when zoomed in looks exactly like f-error.

tommylight wrote:

Those screenshots point to axis not being tuned properly.

You mean the following error increasing? It always stays below one encoder count. I think the stairsteps you see are an artifact of some kind of interpolation based on velocity and interval. One encoder count amounts to 8µm.

I'm pretty sure the tuning can't be much improved at this (low) speed (without changing to a higher res encoder).

Hmm something is different but I'm not sure of the cause.

In your plots, the encoder position reset happens one
sample later than the index-enable falling edge
In my simple test, the encoder position reset happens in the
same sample:

Maybe that has to do with pid.x.error-previous-target true. I will try again setting this to false.

Well it _should_ not (index-enable should be an input only to the PID component)

Was a solution ever found to this problem?

Reviving on old thread here..... So I am having the same problem as rmu. I see a big glitch on the following error one period after index enable goes low. Mill is a servo system with encoder feedback and using encoder index for homing. I have a mux2 plus a timedelay to delay switching over to a small FERROR after homing which works but I would like to get to root cause (Yes, I am an enginerd).