Router losing steps, operator (me) going mad

All of a sudden my gantry router starts loosing steps on the x-axis and as a consequence I am losing my marbles.

I had been milling the same part for about two hours, and at the end of that time the router lost steps while accelerating to its last position, away from the piece. This happened twice in a row and once yesterday.
The drivers are not at all hot, the x drive seems just a hint warmer but by no means extreme.

I opened up the ini file and lowered MAX_ACCELERATION from 1600.0 to 1500.0 (metric set-up)
I left STEPGEN_MAXACCEL at 2400.0 which is what I have because of backlash compensation, only on the x-axis.

After starting the machine again it stalled as soon as it had finished the first part of the program instead of after finishing, ruining the piece.

So before I start a wild trial and mostly error streak, is there any knowledge I should use instead of just trying stuff?

Oh, it seems that it looses steps when speeding up for a longer trip, up to 100 mm where most in the program is smaller moves. Maybe this is a top speed issue as well as acceleration?

I opened up the ini file and lowered MAX_ACCELERATION from 1600.0 to 1500.0 (metric set-up)
I left STEPGEN_MAXACCEL at 2400.0 which is what I have because of backlash compensation, only on the x-axis.

Those figures are HUGE for a stepper powered machine.

Have you done the calculations to see if they are even in theory achievable?

I would suspect it is no co-incidence that the missed steps occur on longer moves, the shorter ones will not have time to exceed the rate at which pulses can be generated.
Does the problem co-incide with a workpiece which requires longer rapid travels?

Personally I would divide them all by 10 and see if the problem still exists, once you have eliminated any mechanical problem

regards

Thanks guys.

I realised that there is a big change from from previous sessions, this is the first part that I'm milling with G61 (exact path) instead of G64.
But is exact path used for G0 moves?

I have now reduced Max velocity to 65 from 70 and the first piece went well.

I've been running this machine for like 2 years with this configuration, it has 16x16 mm ball screws and 6.5 Nm motors. Max speed was 4200 mm/min, now is 3900 mm/min. But I have had it up to 7000 mm/min but then lost steps too often.

I have not done any calculations, have not been in that deep yet to learn. All trial and error...

I'll make a new configuration using stepconf and see what numbers that brings me.

About exact path, would it be ok to switch back and forth between g64 and G61 within the program?

Edit:
I do not notice the difference in milling time so choosing G61 does not bother me that much. Except off course if it is the cause of losing steps...

Hi Guys,

Been doing a bit of testing in stepconf.

There seems to be something mechanical going on but I can not figure out how.

The stalling happens in a specific area of the x axis, also the same area where I had trouble yesterday.

If I let the x axis run back and forth in another area it does 155 mm/second and 2200 mm/s2 over 100 mm area nicely.
But when I go to an area between 1/5th and 1/3rd away from the front end of the x-axis the motor wants to stall, especially when moving away from the front end.

The machine is based on a 100 mm4 inch thick torsion box that was glued on a slab of granite. On top of that 60 mm steel box section, on top of that 25 mm supported shaft.

I thought that it was not bolted to the steel street so I loosened the bolt holding it down. No difference.
Checked for loose connections, none found.

Any thoughts or experience?

Edit:
Got a result. In the wrong direction but still, a result.
After loosening the connection between the steel supports and the torsion box bed, the whole lot stalls at lower numbers along all of the x- axis...

First guess is that tightening the mdf torsion box allow deformation and thus tensioning in more ways than I want.