Re: [CAD_CAM_EDM_DRO] Balancing drivers to motors

Don Rogers wrote:

>I've been following this group for almost a year now and it has been a
>wealth of information.
>
>I have put together a Taig CNC mill. I am using Parker Compumotor OEM 650
>drivers, and Pacific Scientific Powermax II M22 motors on X, Y, and Z, and
>a M21 in reserve for A. My interface to the mill is TurboCNC. I have
>varied the acceleration and max speeds to the point that I can either make
>a 0.0005" move or a 10.000" move and both are dead on with the dial
>indicator after repeated moves.
>
>The Parker OEM 650, and OEM750 to the best of my knowledge, has the option
>of varying the wave form from a pure sine wave to a negative percent of the
>third harmonic. In addition, there are tuning instructions for Phase B,
>and Phase A offsets via trim pots. This coupled with the Microstepping
>ability of the driver up to 50,800 steps/Rev are leaving me some what
>confused. The old mechanic in me the said "if it ain't broke, don't fix
>it" is clashing with the thought "if I tweaked this, it might be better".
>
>

There is a fallacy in thinking that these motors can actually MOVE in
increments
of 1/50800 of a rotation. Just because the driver can produce drive
waveforms
that split full steps to this resolution doesn't mean that the motor can
move
with such fine resolution, or that it maintains such precision under
varying loads.

You might do some experiments with it, such as making extremely small
moves, and using a dial indicator to detect when it "jumps". I can almost
guarantee that it will jump in small increments, perhaps from a few tenths
to something approaching a thousandth of an inch, when you have the
program "creeping" in the smallest increments it can. In other words, if
you make moves of one microstep at a time, the machine will not move at
each microstep, but will suddenly jump many microsteps, when the increasing
force developed by the motor overcomes the static friction.

>I have yet to do very much constructive work with the mill, but what I have
>done so far seems to work well. I have got the backlash down to less than
>0.001" and the ways are tight, but not binding. The X will track to less
>than 0.001" full left to full right. The Y is dead on, and I still need to
>do some tuning on the Z, but overall it is ready to cut.
>
>My question is about the waveform shape setup, and the Phase A and Phase B
>offsets. So far, I have been leaving them to whatever the factory defaults
>were, or in the case of the offsets, what ever they were set to when I got
>the drivers. Is there some tuning that I should do or should I follow the
>"if it ain't broke don't fix it" rule?
>
>

I suspect you don't have the instruments to make these alignments. They
probably
use a super precise encoder to measure extremely small angular movements of
the shaft, and adjust the drive waveforms to minimize cyclical error within
the steps. This would require angular resolution down to the tens of
arc seconds,
but would be meaningless when frictional loads are added.

Jon