I have a 16tpi lead screw with steppers set to 1/8 and the usual 200 step motor which I calculate to 25600 steps (i.e. pulses).  However, when I use the automatic calculation routine within Mach3 I get something like 25830.69158, so I know it's close. I've got the kernel speed set to 25k.

I have the backlash disabled and therefore always move the axis in the negative and then to the positive before telling Mach to move to some desired distance in the positive direction. My caliper is fixed to the bed of the mill and the holder for the sliding portion of the caliper is fixed in the spindle. I'm currently using (+) 3.5" for the X-axis.

Here's what odd: I'm getting random measurements from the caliper. Sometimes it will land on 3.501", 3.497", 3.522", etc.  Note, I always move the bed negative, then positive, then rezero the caliper and the Mach3 DRO for the X-axis and before moving the prescribed distance. (This is to alleviate any backlash).

I haven't changed the pulse width or the kernel speed just yet and really don't know what to expect when I do other than having to retune the steppers which I'm trying to avoid.

I just can't figure out what the issue is.  Any help would be appreciated.


BTW, has anyone found that when you use the automatic calculation routine you have to input G90 in the MDI in order to get the steppers moving again?

Should be rather constent. A few thou i would expect but .022" is like a 1/3 of a turn of the screw. Sherline screws are 20 tpi and are made rather well
but even with anti-backlash device on them you still have backlash. The screws though are quite accurate in their travel and hold within about .001" for travel when new. Just giving this as something to gauge your screws to.

Not knowing the quality of your screws ( ie are they threaded rod? )
I would slow the velocity and acceleration down and run some more movement tests and see if the results are within some range.
Also check movement for 1",2", and 3" starting from the same spot to see if they are gaining consitantly or whatever. The caliper should work fine if held nicely and assume you checked it.

The kernel speed is ok. Have a look at your driver info to see what pulse width they recomend.

If the Y axis is fine,  you may want to just swap the x with the y and see if that causes any change to the Y movement ( off course use the Y settings when you do that).

RICH

Using two boards:

Xylotex XS-3525/8S-4 which states use a minimum of 1.0uS.
cnc4pc C11 which states start with 3.0us, but up to 5/0us can be used.

I'll play around with the different pulse widths and see what happens.

Haven't started on the Y-axis yet.

Okay, I ran the DriverTest.exe and was able to get up to 65kHz, but when I selected 75kHz I got a message that stated it was pulsing too slow.  Which seems to be the opposite of what I would have expected the message to say.  Maybe I don't have my thinking cap on straight since it seems I'm pulsing faster with 75 than 65.  Unclear about this and would appreciate some clarification.

I will note also that even at 65kHz, I was getting a nice flat line with only an occasional minor spike.

In Mach3, I tried different uS settings for the step and direction but didn't seem any changes in the erratic readings while moving identical amounts as described in the first post.  Do I  have to close and restart Mach3 when I make these changes? I didn't see it in the manual.

Thanks in advance,
Vogavt

Try 4us and check the x and y axis. Say the x axis steps per unit work out to be good and are repeatable, then swap the cable from the x axis to the y axis and see if the y axis movements are ok. ( do that when everything is turned off )
I will assume that the steps per unit are the same for the x and Y and that the accel and velocities are fine for both axis.
What you are doing is checking the BOB and drive to see if you are loosing steps by using the known good pulse path to check a possible bad one.
If the y axis is still no good then it may be mechanical.

One thing at a time .......
RICH

Just a thought but in the first instance I'd forget your caliper. As Rich says, the 22 thou is a THIRD of a turn which should be obvious to the naked eye - stick some tape on the screw with a pencil mark and convince yourself first whether you are ACTUALLY getting this sort of rotation error.You could try commanding an integral number of turns back and forth over and over and just make sure the pencil mark looks like it stops in the same place each time. First find the problem THEN look for the solution.

Yes, threaded rod. Harbor Freight 44991 standard issue with no modifications to the leadscrew, replacement or otherwise.

Good point about the penciil mark. I'll try that this afternoon and see if I can land back on it. Might be difficult though since I've got backlash disabled and am roving negative then positive to take up any perceivable backlash.  I'll have to use two different marks (colors) to make certain I get at the same starting point each time and a different color for the ending point.  Green and Red seem like obivious choice of colors! LOL!

Thanks for the comments and support!

Vogavt

I wouldn't worry about backlash for the moment. As you havn't said otherwise I'm assuming your steppers are directly coupled to the screw. Whatever - why not stick your mark on the stepper shaft - then you're definitely isolating any backlash.

Similar to marking your stepper shaft you can mark you hand wheel. Your just magnifying movement.

If the handwheel is say 3" dia then circum is 9.424". So you wrap a piece of tape around the hand wheel and then with a razor blade cut it. If you remove it and put a mark say every .050" along the tape then relative to a stepper ( 9.424/200=.047"  dosen't take into account the screw ) but you get the idea. Thus you can see steps roughly or linear movement depending on how you mark the tape. You can even make a big paper circle wheel.
For backlash rough check, you just tighten the gib and rotate the handwheel one way and then the other and you are seeing the backlash from the screw and bearings with the pointer set on mark. If you tighten the bearings / add preload and there was play due to them then you can see the
the adjustment affect of preloading the bearing. Idealy you would be checking the torque while you turn the handwheel but it's a poor mans way
for isolating things and seeing what is influencing the lack of movement.


Here is another example:
6" disc=18.85" circ  
18.85 / 200=.09425" one step of motor on the disc
Thus you can see affect of one step to movement.
You can put one on the motor and one on the handwheel and compare them if you wish.

Probably clear as mud....FWIW,

RICH

Good replies!

Yes, it's directly coupled and I'd planned on putting the marks on the shaft and on the base of the motor.

Regarding not worrying about backlash: I'm should be eliminating the backlash since I'm moving neg, then positive.  Maybe I'm not making myself clear on what my question is, so I'll give it another whirl. 

What is happening is when I bring the x-axis to the negative and back to the posiitive AND zero both the DRO in Mach and the DRO on the caliper, then move the bed in the positive I get a specific number.
I repeat the process as stated in the previous sentence over and over again, yet come up with different numbers each time, even though I'm not changing pulse width, steps/rev, etc., nothing other than moving the axis back to zero, rezeroing both and starting again.

You're correct, I'm not worrying about the backlash at this point. I'm just having difficulty repeating the measured movement.

I will try the marking approach here shortly and see what I find out.

Thanks guys, hang with me on this please.

Vogavt