Re: Measuring backlash more accurately?

At 05:45 PM 2/16/03 -0600, Jon Elson wrote:

>So, if the error IS accumulating, then you are either losing steps in
>some manner, or there is mechanical slippage in your machine,

Thanks for the reply, Jon. I'm using three General Controls 29A55
single-axis drivers (half-stepping drives) with CNCPro. Leadscrews are 2mm
pitch "anti-backlash" ballscrews (it's an older Techno Series I gantry
table,) so there are 2540 steps/inch. The shaft couplers are good and
tight. I have the acceleration set pretty low--.5 in/sec^2--because of the
relative flexibility of the machine, but the motors on all three axes are
capable of 2+ in/sec^2 (but the machine flexes more at that acceleration.)

I've also backplotted the gcode I was running in editNC, and know that it
is "good".

I did an experiment after I posted. I had the Y backlash set at .001".
Repeatedly jogging out and back .250" gave a reading of 0.000" on the
indicator, sometimes varying .0005" or so.

I set the backlash compensation to .0011", and was suddenly "losing" about
.001" of travel each cycle--i.e. the Last Word was being deflected about
.001" less each cycle.

I set the backlash back to .001", and the movement was back to being about
even. Same when I set the backlash to .0009".

When I set the backlash to .0008", I was suddenly "gaining" about .001" of
travel each cycle--the Last Word was being deflected about .001" more each
cycle.

I'm wondering if there is some interaction between the relative coarseness
of the stepping, and how CNCPro inserts its backlash compensation. The
steps are almost, but not quite half a thou, and trying to apply an even
thou of backlash compensation might accumulate rounding errors(?)

Best regards,

Randy

Randy Gordon-Gilmore ,----.___________ ______________ _________________
ProtoTrains // = = === == || == == == = || == == == = == =|
Rio Vista, CA, USA /-O==O------------o==o------------o==o-----------o==o-'
zephyrus@... http://www.prototrains.com