Z Axis Drift......

[bbristow]

Please HELP! I am pulling my hair out…
I have been having this problem now for a while, but have been able to avoid it until now. It seems that my Z-Axis is drifting or loosing steps. What makes it more difficult to figure out is that it seems to be a math error of some kind as the results I can duplicate accurately.

I am thinking this is a Mach 3 issue since I can duplicate the results.

Here is an example…. I am cutting a 3d part, I can let it cut the file out and when it is finished, it thinks the z axis is at 0.0, when it is actually at 0.25. I can cut the part again after re-zeroing it and I end up at pretty much the same offset. I am doing parallel finishing; I can see the bit is gradually and smoothly cutting lower and lower. There is no big jump; it just seems to smoothly drift.
I can also zero the machine and toggle the head up and down about 15-20 times and will check zero and it will be off again. I can do this at any speed setting and I let it completely stop before changing direction and I still produce the same results.
This is what I have tried so far with no change….
Tried changing frequencies to 3 or 4 different settings.
Tried the shirline ½ step mode.
Tried setting the pulses under motor tuning to 5, then to 15, then to 20.
Tried changing the number of pulses per inch from 3200 (Setting I was running at) to 2000, to 1600, to 1000 and I still see the same results.

I am running the Joes 4x4 hybrid machine. You can see it here
http://www.saberfire.com/gallery/Projects/CNCMachine/

I am Running Mach 3 with the Hobby CNC Pro kit with the 305/oz motors. I have the jumpers set to quarter step.
I am not sure what else to do here?

Thanks,
Bobby

[PaulRowntree]

Could the Z drive electronics be picking up spurious pulses ?

Does the Z drift when you are not activating the Z motion, (ie just moving in the XY plane)?

Does the Z drift when you are not activating any motion (not X, nor Y nor Z) ?

Does it drift if the Z motor is unplugged (ie an air carving?)

Could it be an accidental short between the PP and drive components?

If it is not electronic, perhaps it is mechanical : Does it drift if the Z motor is unplugged (ie doing an air carving?)

Good luck!

[bbristow]

-Could the Z drive electronics be picking up spurious pulses ?

-Does the Z drift when you are not activating the Z motion, (ie just moving in the XY plane)?

Yes. It can be off just by moving the head up and down as little as 10-20 times an any speed by itself. It is always off by the same amout. i.e. If I move the head up and down 20 times, 2 inches up, 2 inches down, it will be off .035 (not auctual number, I forgot since I ran the test). When I do the same test I will end up with the same results. If I cut a part out and zero after every part, all the parts will be identical with the exact same slow z cutting further and further into each part.


- Does the Z drift when you are not activating any motion (not X, nor Y nor Z) ?

No, it holds it position without any problem when just sitting.

- Does it drift if the Z motor is unplugged (ie an air carving?)

Yes it does.

- Could it be an accidental short between the PP and drive components?

- If it is not electronic, perhaps it is mechanical : Does it drift if the Z motor is unplugged (ie doing an air carving?)

Nope. It holds it poisition just fine.

Thank you for your help!

[PaulRowntree]

I don't quite follow : you say that it drifts in Z when the Z motor is unplugged, then you say it is holding position just fine.

What about the test where you hold Z constant and move in the XY plane?

BTW : I am very impressed by your build documentation and the final machine. I should be so lucky ...

Cheers!

[awerby]

Please HELP! I am pulling my hair out…
I have been having this problem now for a while, but have been able to avoid it until now. It seems that my Z-Axis is drifting or loosing steps. What makes it more difficult to figure out is that it seems to be a math error of some kind as the results I can duplicate accurately.

[If it's a math error, the z should come back to zero, when you give it a MDI move: G00Z0. Does that happen? Lost steps are pretty random; you won't see the axis end up in the same place twice. If it does go back to the zero-point you started from, then you need to check the steps/unit setting for Z - even a small error will accumulate. The math works like this: multiplying the steps for a full revolution of the motor (usually 200) times the micro-stepping factor (4) times the pitch of your screw in threads/inch or gearing (?) equals the number of microsteps it takes for the axis to move one inch.]

I am thinking this is a Mach 3 issue since I can duplicate the results.

Here is an example…. I am cutting a 3d part, I can let it cut the file out and when it is finished, it thinks the z axis is at 0.0, when it is actually at 0.25. I can cut the part again after re-zeroing it and I end up at pretty much the same offset. I am doing parallel finishing; I can see the bit is gradually and smoothly cutting lower and lower. There is no big jump; it just seems to smoothly drift.

[If it's at .25 above the part when it thinks it's at 0, it's drifting upwards, not downwards. This can happen if it loses steps while plunging - are you sure you're using a center-cutting endmill? If not, that would account for this, since it would be hitting a lot of resistance on each plunge. But if it's really drifting lower, then it's something else. Did you mean to write -.25? ]

I can also zero the machine and toggle the head up and down about 15-20 times and will check zero and it will be off again. I can do this at any speed setting and I let it completely stop before changing direction and I still produce the same results.
This is what I have tried so far with no change….
Tried changing frequencies to 3 or 4 different settings.
Tried the shirline ½ step mode.
Tried setting the pulses under motor tuning to 5, then to 15, then to 20.
Tried changing the number of pulses per inch from 3200 (Setting I was running at) to 2000, to 1600, to 1000 and I still see the same results.

[You might be setting your velocity and accelleration too high. Due to the loads on them, Z axes typically need lower settings than the other axes. Go back to the original default frequency and pulse settings (or ones that seem to work on your other axes) and, in config/motor tuning, lower both the velocity and accelleration values for Z by about a third. Check to see the motor runs smoothly by pressing the up/down arrows on your keyboard. If so, click "save axis settings", and try jogging the Z axis up and down in the manner that produced the lost steps before.]

I am running the Joes 4x4 hybrid machine. You can see it here
http://www.saberfire.com/gallery/Projects/CNCMachine/

I am Running Mach 3 with the Hobby CNC Pro kit with the 305/oz motors. I have the jumpers set to quarter step.
I am not sure what else to do here?

Thanks,
Bobby

[That's a pretty nice-looking machine you built! Another thing to try is to swap the motor leads from the axis that's misbehaving to one that works okay. Make sure you've turned everything off first and given the capacitors time to discharge. See if the error follows along. If that's the case, then it's probably a problem somewhere in that drive circuit. If the error stays with the Z-axis, look for bad settings, a fried motor, or a mechanical problem. Sometimes a loose coupling will mimic a lost step error, only letting go when turned in one direction.]

Let us know how it goes, okay?

Andrew Werby
www.computersculpture.com

[bbristow]

- I don't quite follow : you say that it drifts in Z when the Z motor is unplugged, then you say it is holding position just fine.

Sorry, I thought you meant the auctual router being unplugged. If the z axis is not told to move, it is very ridged and does not loose height.

- What about the test where you hold Z constant and move in the XY plane?

Z is held just find with no movement.

- BTW : I am very impressed by your build documentation and the final machine. I should be so lucky ...

Thanks! I worked very hard on it.

[bbristow]

[If it's a math error, the z should come back to zero, when you give it a MDI move: G00Z0. Does that happen? Lost steps are pretty random; you won't see the axis end up in the same place twice. If it does go back to the zero-point you started from, then you need to check the steps/unit setting for Z - even a small error will accumulate. The math works like this: multiplying the steps for a full revolution of the motor (usually 200) times the micro-stepping factor (4) times the pitch of your screw in threads/inch or gearing (?) equals the number of microsteps it takes for the axis to move one inch.]

[If it's at .25 above the part when it thinks it's at 0, it's drifting upwards, not downwards. This can happen if it loses steps while plunging - are you sure you're using a center-cutting endmill? If not, that would account for this, since it would be hitting a lot of resistance on each plunge. But if it's really drifting lower, then it's something else. Did you mean to write -.25? ]

Sorry, I was confused what to call it hear, since the end result is that it is auctually cutting lower and lower.

[You might be setting your velocity and accelleration too high. Due to the loads on them, Z axes typically need lower settings than the other axes. Go back to the original default frequency and pulse settings (or ones that seem to work on your other axes) and, in config/motor tuning, lower both the velocity and accelleration values for Z by about a third. Check to see the motor runs smoothly by pressing the up/down arrows on your keyboard. If so, click "save axis settings", and try jogging the Z axis up and down in the manner that produced the lost steps before.]

If I toggle the head up and down about 15-20 times and will check zero and it will be off again. I can do this at any speed setting and velocity setting. I let it completely stop before changing direction and I still produce the same results.


[That's a pretty nice-looking machine you built! Another thing to try is to swap the motor leads from the axis that's misbehaving to one that works okay. Make sure you've turned everything off first and given the capacitors time to discharge. See if the error follows along. If that's the case, then it's probably a problem somewhere in that drive circuit. If the error stays with the Z-axis, look for bad settings, a fried motor, or a mechanical problem. Sometimes a loose coupling will mimic a lost step error, only letting go when turned in one direction.]

I will try this....

[jsheerin]

If swapping the z motor to another driver doesn't produce any results, then:

What is your acceleration setting in the mach 3 motor tuning screen for the Z axis?

What is the range of speeds you've been testing at?

[stevespo]

What type of screws/nuts are you using? Have you measured the backlash on the Z with a dial indicator? This procedure has been documented pretty well and amounts to moving the axis a defined amount and then actually measuring it to make sure that the values agree.

Does .01" really equal .01"? Is it the same in each direction? If not, you can enable backlash compensation in Mach and it should help.

At one point my Z was "drifting" downward and it was due to my Xylotex experiencing "thermal shutdowns" due to overheating. It would tend to (occasionally) lose a step when it was lifting the Z. I can see you have a nice enclosure with plenty of fans, so this seems unlikely - but I thought I would mention it.

Are your driver chips warm to the touch? Is there a heat sink attached? In some cases, if the motor/driver is stressed on the Z, a counterbalance might help as well.

Last thought. The 800Mhz dual CPU PIII might be another issue. I thought I read that Mach had issues with dual CPU machines. Can you disable one CPU in the BIOS? That machine can probably sustain a 25KHz kernel speed, but I found that my own 800Mhz PIII had timing issues at high feedrates. It was sensitive because there was so few CPU cycles to spare.

Could also be a flakey pin on the parallel port. I've been down that road as well. So you might try swapping leads if you haven't already. Do you have a decent breakout board with opto-isolation and filtering?

Just something to consider (try another computer) if everything else leads to a dead end.

Anyway, great looking build!

Steve

[ger21]

Loss of Z position is frequently due to the acceleration being set to high. Cut it in half and try that.

Also, why is your machine so tall? It seems like it would be difficult to load at that height?