Steps per unit calibration problems

[idtkid]

I am running a spring loaded rack and pinion setup (like on Shopbots) and I am having some weird issues with Mach 3's steps per unit calibrator. For example, I choose the X axis, tell it to move 1 unit and it moves 1.25. I enter in 1.25 for the actual move and repeat the test again only to not get a move of 1 unit. I can repeat the test 10 times over, everytime getting a result that is +/- ~.1"

My spur gear is spring loaded to the track very tight, the gear is secured on the motor shaft very tight, I cant see any backlash whatsoever. Any ideas as to why I can't get precise moves? Thanks!

[Konstantin]

Well what kind of number Mach suggests?

[idtkid]

All irational numbers ranging from ~525 to ~560

[Torchhead]

Do it manually with a tape measure.

A sharp point in the tool holder helps. Establish a zero for one axis and reset the axis DRO to zero. Open the MDI window and send the axis out a short distance: say 10inches. Measure from where you started to the stop. Lets say it was 10.25 (10 1/4"). Take the Commanded distance and divide it by the actaul distance (10.00/10.25) = .97561 now multiply the Steps per inch you have in motor tuning by the ratio. Say your setting is 8000 steps per inch. YOu new setting would now be 7804.87 steps per inch. THIS IS important move the axis back to the zero point BEFORE you change the setting in MACH. Now do it again. If it's pretty close then do the same test of a much greater distance and once again work the ratio. If you can get dead on at 48" on a tape you will be very close at 1". Do this with each axis. Use a wooden feeler block with Z.

Here is a formula:

DT/DC = CSPI/X

where DT = Distance Traveled [actual distance measured]
DC = Distance Commanded [how far you told it to move]
CSPI= Current Steps per Inch [Current setting in MACH "Steps per Unit"]
X = New Steps per inch [What you change to for the next try]

Just rememember not to change the Steps number until you have returned the axis back to zero (e.g. G00 X0) each time. The first time the error may be large so use smaller moves. As you get closer and closer make the moves as far as you table will permit.

If your table is accurate enough and you are compelled to be dead on then do the final calibration with a dial indicator and use the same reference method but you will be in fractions of steps so work to three decimal places on the steps.

Tom Caudle
www.CandCNC.com
Totally Modualr CNC Electronics