Z axis creeping up!

Just got my machine together and tried a sample file. The X and Y axis seem fine but when the job is done, the Z axis is about a half inch above the workpiece and Mach 3 is indicating zero on the Z axis??? Obviously the job is screwed up as well, but only in the Z axis, X and Y seem to be fine. I've tried slowing the feed rate way down, but it doen't make any difference. My initial guess was that it's missing steps? I'm using Mach 3 with Gecko G320's and a PCI parallel card. My driver test seems ok, not quite 24,600 but pretty close and steady too. The Gecko's are not showing any faults so I assume the motors are keeping up with the drives?

Any thoughts would be welcome.

Thanks

Look for noise.

Look for binding.

Lost steps are the result of: signal not sent, signal not recieved, signal screwed up, friction overriding signal, the perverse nature of inanimate objects.

When you assume anything, you're usually wrong sooner or later.

Since your using servos, it's most likely noise on the encoder lines or step/direction lines. Are your encoder wires running next to the power cables? What version of Mach3 are you using?

I checked for any mechanical interfernce last night and everything moves freely. I ran the Z axis through tests and when I tell it to go to various heights it's spot on. But when I run a program and the other axis are moving at the same time, that's when it seems to loose steps. So it might be an interference issue. I used the encoder wires from US digital which are shielded so its more likely to be getting interference on the step line from the breakout board. I'm going to try running the program with the X and Y axis disabled and watch the Z to see what happens, that should help narrow it down. I'm using the latest version of Mach 3, 1.84 i think?

I had the same problem, only it was lowering the z.
My problem turned out to be backlash due to a loose set of bearings. Mine was going down because the head was heavier than the counter ballance weight. If your counterballance is heavier than your head or you have a gas strut that is to strong this could be the issue.

run a set of g-code through mach3.
Put a digital caliper in a vice and lower the head on it and zero the caliper. set the machine position to z-1.
Then run g-code like this
z1
z-1
z1
z-1
etc. like 20 times.

Watch your ballscrew and thrust bearings for slop

I had a similar problem with the Zaxis creeping during program run. It turned out to be noise on the z encoder for the servo. I had used the supplied encoder with the motor which was a six wire encoder (renco) instead of a four. It had an a and b channel but also had a not a and not b channel. I just hacked them off and used four channels. Apparently, this makes the encoder very sensitive to noise. Long story short, us digital sold me a replacement correct encoder and I rerouted my shielded cable and the problem was resolved. Just something to think about.
Good Luck
Billyjack
:cheers:

I think I've traced it to the encoder. I separated the power wires as far from the encoder wires as possible and still had the same problem. As mentioned, I have the fully shielded encoder harness from US Digital so was not expecting a noise problem anyway. So other than noise, what else could cause an encoder to be at fault? Or is it maybe the Gecko drive has a fault? I know the X axis works fines so I used the step and direction lines from it to drive the Z axis but with no improvement so I know its not the breakout board or Mach 3. Is there a way to "test" the encoder?

Thanks for all the replys so far.

Sure, but one needs to be using common encoders on all the axis.

Move the encoder to another axis and see if problem follows it. Big problem if you're not using common encoders but you sort of set yourself up for the headache by trying to get bye with the decision to use what you had on hand in the first place.

This is how the tech diagnosed the servo amp and servo itself on my CNC mill. Moved the stuff around until he found the true culprit.

In the end, I had all three amps and servos serviced but that's a whole 'nother issue altogether.

Well I've finally found the problem. Wasn't the encoder, turns out to be a bad drive. I swapped the X axis drive for the Z and it worked perfectly. Sure was hard to trace, I assumed the drive would either work or not, didn't think it could miss steps.

Thanks to all those that helped.