GHOST? Noise? Bad Luck? Could use some ideas..

[@dam]

At a loss here and could use some help. Have a highly detailed job and the Z axis is losing steps at almost the exact same place. It is not a slow loss, it is losing .05 - .08 at once in Z- direction.

Here's where things get odd... I have run this exact same job 5 times now. The second time I ran it, with no changes, it worked fine. The other 4 times, it loss steps within an 1/8" of the other runs. The painful thing is this doesn't happen until 5 hours into the job. This job has probably been run successfully 25+ times on this machine before the Acorn conversion.

My first reaction for this is electrical noise, but in my 15+ years of CNC work, I have never seen noise this "consistent". I have the job air running right now with the VFD off. Bad thing is, if it doesn't lose steps this time, is it truly noise, or like the one off above that did work?

After noise, I'm thinking a stepper or drive over heating. Kept a close eye on this and both are well within temp limits

Using the Acorn board with standard wiring for Leadshine DM series drives.

Greatly appreciate any help I can get here. Shop is hurting with our backup machine being down for a retrofit

Thanks!

Adam

[spindle Nerd]

Ok i will try and help
1.First check backlash on the Z axis i know you may have some but check it again ,use dial indicator run from highest point to lowest and see if it can repeat.
Make sure you test with machine cold and then warmed up , if you find it try to eliminate it as much as you can .

Make a z axis test program with just Z axis moves and put a indicator at the bottom and keep looking to see if it repeats to the same position every time ,if it does not then, swap driver first with other axis retest , if still does not work correctly swap motor's x to z and see what you get .

This should help you narrow it down and find it ,good luck and i hope you find it sooner than later.

[Gary Campbell]

A same position, multiple time position loss in the Z- direction could have a few causes.

Mechanically would happen at a sharp retract area or when there is a G0 Z call. This would happen after a number of hours of running due to heat and weakening of drive and motor. It may happen with the Acorn and not another because your acceleration is set higher.

I don't know of a environmental (noise) issue that would repeat.

Dust collection static can cause this. At a certain position on the table the dust collector hose will come in contact with the machine frame, discharge onto the machine resulting in added steps.

None are overly easy to find, but doable if you pay attention. Since the file has been run before, I would look at the accel settings first and make sure they are close to the same that were used on the other control.

[@dam]

Thanks for the thoughts Gary. Acceleration was the first thing I dug into and just forgot to mention in the original post. I come from Mach4 on this machine, so the acceleration is calculated differently. I do think I have it converted correctly to the time based setting on the Centroid side though. Right now I'm at .5 on the Z acceleration with max rate of 156. This was where I landed when I tuned things on the Mach4 side.

Now that you have me thinking about acceleration, I'm wondering if the smoothing settings could be a factor since they override acceleration. This is going to be a fun one to replicate. I'm going to try air running the job, but starting it a little before where this keeps happening with adjusted smoothing settings.

The dust collection hose is a long shot, but a good long shot to check. This is a raster style job that is moving slowly in the Y- direction. That happens to be the direction that would pull the hose closer to the frame like you mentioned.

Thanks!

Adam

A same position, multiple time position loss in the Z- direction could have a few causes.

Mechanically would happen at a sharp retract area or when there is a G0 Z call. This would happen after a number of hours of running due to heat and weakening of drive and motor. It may happen with the Acorn and not another because your acceleration is set higher.

I don't know of a environmental (noise) issue that would repeat.

Dust collection static can cause this. At a certain position on the table the dust collector hose will come in contact with the machine frame, discharge onto the machine resulting in added steps.

None are overly easy to find, but doable if you pay attention. Since the file has been run before, I would look at the accel settings first and make sure they are close to the same that were used on the other control.

[@dam]

Update: I have ruled out noise, heat, and dust collection static discharge. Did two limited runs, where I started about .5" away from the problem area. These were done in the air with the spindle and DC off. Same behavior with losing .09" each time. Also made sure the smoothing settings weren't multiplying the acceleration and/or feed rate.

This leaves me with acceleration. At least I think acceleration... I calculated for time from the Mach4 acceleration value and came up with .5xx. I forgot I had already changed the acceleration to .6s yesterday, so the last 4 runs have been at the .6s acceleration value. I just changed that to .7s, and started another test. We will see where this lands.

Are there any other specific parameters that are behind the scenes that could be part of this?

Thanks-

Adam

[martyscncgarage]

Are the motors and drives the same for each axis?
If so swap Z motor and drive with the pair from X or Y.
When you disconnect Z motor check the screw the full travel for any signs of binding before reinstalling the motor.
Retest

Also, do you happen the use a shielded cable between Acorn and the stepper drivers or individual wires?
Marty

[pulsenpal]

Bridgeport EZ Trak and EZ Path mill/lathe from factory 20+ years ago had a problem with lost position --dc servo-
due to compression locking mechanism on smooth output shafts of SEM branded dc axis motors

no key or pins or setscrews are involved in squeezing drive shaft sprocket--and under load slippage occured if
compression sleeve tension backed off--for the mill it typically was arc move

what hardware is control system coupled to?

[@dam]

Drives and motors are the same. I may have to try this if I don't get anywhere soon. Will just be a pain because there are no connectors on the motor side. For some reason they direct soldered the cables on the motor end.

I have already disconnected the Z motor shaft to check for binding.

Will have to double check if the wire is shielded going from the driver. I know it should be, but I do not think that could be the source here. Just too repeatable behavior.

Adam

Are the motors and drives the same for each axis?
If so swap Z motor and drive with the pair from X or Y.
When you disconnect Z motor check the screw the full travel for any signs of binding before reinstalling the motor.
Retest

Also, do you happen the use a shielded cable between Acorn and the stepper drivers or individual wires?
Marty

[@dam]

I will triple check for mechanical slippage here shortly. Did a quick torque check of all the bolts on the axis at the beginning of this mess.

Hardware is more or less a small footprint Chinese import.

Adam

Bridgeport EZ Trak and EZ Path mill/lathe from factory 20+ years ago had a problem with lost position --dc servo-
due to compression locking mechanism on smooth output shafts of SEM branded dc axis motors

no key or pins or setscrews are involved in squeezing drive shaft sprocket--and under load slippage occured if
compression sleeve tension backed off--for the mill it typically was arc move

what hardware is control system coupled to?

[@dam]

Update #2: I have performed 8 test runs with different acceleration values. Thought I had it on run #7 when I loss no steps.... but then ran one more time with the same settings for run #8 as a sanity check and loss steps

Below are the results from the tests:

1. Accel=.6s Steps lost= -.0943
2. Accel=.7s Steps lost= -.101
3. Accel=.8s Steps lost= -.0436
4. Accel=.9s Steps lost= -.0934
5. Accel=1.1s Steps lost= -.0464
6. Accel=1.5s Steps lost= +.0254
   
7. Accel=1.2s Steps lost= -0.0004
   
8. Accel=1.2s Steps lost= -.029

I'm really at a loss here. Acceleration appears to be a factor here, but I just can't get over why the issue happens in pretty much the same spot every time. The feed rate doesn't change in this area, and it is just as detailed as the rest of the piece.