My X-Axis is losing steps!

[cncnovice2]

I'm having a problem with my X-Axis losing steps Or is it microsteps and not returning to zero.  The mach3 says it is back to zero but I lose about .0005 or so each time I run this program:
F20
G20G90
M98P1234L3
M30
O1234
G1X4
G1X0
M99
  System consists of the following...  Computer 1.83Ghz Athlon, 1Gig memory, PMDX 120 Breakout board, Geckodrive G212, Homeshopcnc 1200 oz.in. Stepper motor, 1:2 timing belt, 5 TPI ballscrew on 9x42" table.  
  When I run the above program with a dial test indicator zeroed on a block fixed to the table, it reads a lower number at the end of the program run, ie. the machine is slipping toward the positive direction.  I don't know what is causing this as changing the feedrate, acceleration, step pulse timing, etc. has not helped.  I have gone through the Windows Optimization procedure, and that didn't help.  What should I try?  I'm not even sure it is software or hardware related...  Thanks for any help or suggestions you might have, Don

[Hood]

What does the driver test look like?
 Have you tried changing the active sate for Dir?
Have you got the correct current set resistor in the Geckos for your motors?
Is it just the X axis? If so have you tried swapping drives? have you checked there is no binding in the axis? Pulleys etc all tight?

Hood

[RICH]

Just echo what Hood has posted. Easy to get to some tolerance then the "system " comes into play with
each part, mechanical and electrical, even the indicator,possibly having an influence.
Be meticulous. 
RICH

[cncnovice2]

First a clarification. The x Axis is slipping to the positive direction only.  If I se the program to go negative it behaves and repeats perfectly.  The Y-Axis is also doing it only it only goes toward the negative direction...  The MPG when set to .001 inc. repeats perfectly on both axes, but when set to .0001 inc. doesnt repeat in X.  Again, it loses steps to slip toward the positive in X.  Making repeated back and forth moves yields mor and more error. The indicator is mag base mounted on the head and touching the fixed jaw on the mill vise.
Step is active low for all axes.  Direction is active low only for the X Axis.  Could this be a mechanical problem?  I will try to post pictures of my setup. This may take another post since my card reader is finiky. Thanks, Don

[Hood]

Do you have a Breakout board with optos on it? Could be the optos are too slow. Also what are your steps per unit for the problem axis. Try setting the active state the other way and then you will have to reverse the axis direction in Homing and limits to correct for that.
Hood

[cncnovice2]

Hi, again.  Breakout board is a PMDX 120.  Steps per unit is set to 10000.  Kernal speed is 45Khz.  I should be able to post/attach a picture of the control setup now.  the cable in front of the BOB is the E-stop sense input from the BIG RED BUTTON.  Computer is off image to left. Gecko powersupply is upper left.  800VA 70V.  The Current set resistors are set per Geckodrives calculation formula and the motor spec.  The signal cables to the G212s are 4 conductor Shielded Com cable.  The shield is grounded to the aluminum plate at the Geckos only.  The SJOOW wires at the bottom are the outputs to the motors.  I've been using this machin for about 2+ years now and just recently reinstalled the Windows, and Mach.  Yes, It has been "optimized" per the Mach Support instructions.  Thanks, Don

[cncnovice2]

I reversed the X axis in Homing/Limits then ran the mirrored version of the program with the result that the table is still slipping to the left.  Sounds like something other than the computer?  Don

[Hood]

OK your steps per unit are 10,000 and I am presuming you are in Inches, if so then that is a resolution of 0.0001 inch. As part of that is micro stepping then you are never going to get things that accurate, it shouldnt really be cumulative though.
 Have you tried dropping the acceleration a bit to check that its not just skipping a beat every now and then.
Hood

[cncnovice2]

Yes, I have tried accel values as low as .1 (ramps up very slowly) and got the same results as accel. 4 which is what I'm using now.  Don

[Hood]

Yes sorry just re-read your opening post and saw that. Dont see if you have tried changing the active state of the Dir pins but I tend to not read things properly so apologies if you have
Hood