Another servo tuning

I will give it a go tomorrow but I have the same issue on all three axis so I am thinking I have some sort of systemic error somewhere.

I am having brain storm day and guys I just discovered an issue I think, maybe not related to the tuning maybe it is.
The machine came with Heidenhain LS403 scales and I am using EXE 610 to convert it.
I just went through the datasheets again and the max input frequency to the EXE is 50Khz, with linear scale grating 20um and if I am calculating right is 1000 mm/s.
It looks like I will have to look for faster EXE boxes.
Anyone knows of any faster exe boxes if I am understanding the datasheet correct?

Do you know what the exe boxes do if the input frequency is too high? Do they lose counts, or do they catch up later....

Have you checked whether your system is moving accurately (ignoring dynamic errors)

I actually have no idea what they do if the f is too high. I have been thinking about testing it but have no idea how. I have been having fairly decent accuracy but all up to 1000mm/s I just started with speed and accel increase lately with not much luck so far.
What I am about to test is to bump up accel but limit the speed to 1000mm/s and another test to run open loop as Peter suggested.
How do I measure ferror in open loop? do I have to create a script to calculate motor fb - encoder pos? or what would be the best way.
I am guessing that ferror in open loop config will be always 0, is this correct assumption? since there is no actual encoder feedback.

So this is what happens if I bump up accel to 300 and limit the speed to 16 m/min in hal

here is the openloop net x-pos-fb     <= hm2_7i95.0.stepgen.00.position-fb
but I don't know how the ferror is calculated in this case I had to decrease scale it 2m now otherwise I could not trigger.

Edit: I forgot to bump up the max speed again

is there any way to reload HAL without closing and opening whole thing again?