Problem with Z Axis - Suddenly goes too deep mid-project

Hey Guys,

Thanks in advancee for any help. Today I was running a project and conducting a pocket toolpath operation. Mid-project, my CNC started digging into the stock. It appears my Z axis dropped about 1/4". The g-code specified a depth of .0625". Fortunately I was right at the machine and stopped it, whereby preventing it from going any further.

I’ve had this happen before a few months ago on a v-carve when suddenly my letters I was carving on a sign started going way too deep.

I have a 48x48 Pro with Nema 34s and an 8.7hp spindle.

Any ideas on what to look for and inspect?

My project was created in vCarve Pro and is solid, as I have cut this very file in the past. I reviewed the g-code lines in Mach4 where it went sour, and Mach showed it should be correct at .0625" for a pocket depth. My bit is in the collet tight. I rehomed the machine and checked the project xyz location. xy was spot on and the z was too deep by almost .25 inches (.2042)

Again, thanks for your help!

Russ

Last time this happened to me, it was the bit pulling out of the collet. Are you sure the bit hasn’t moved? Can you measure it and/or the spindle’s height?

I’ve become a bit paranoid about cleaning new bits to remove shipping oil, and really torquing the collets pretty tight.

Milling bit nice and tight - not the issue. Thanks

Is your collet clean? Tight means nothing if the collet can’t close properly.

Not sure I follow you. My bit is in the same location it was when I put it in the collet, torqued it down and used my touch plate to establish Z. The collet is tight as well. After the damage, the Z height was inspected and the new “zero location” was .2042" deeper than before I began the operation. Again, nothing with the bit or the collet changed. Its almost like the Z slipped during the cut.

Even though the gcode in Mach did not show any changes, the pocket got progressively deeper as it went, before I stopped it.

I think the reason folks are zeroing in on the bit slipping is that is the most likely possibility, either because it wasn’t tight enough to start with, or the collet is dirty or scored, etc. If you’re sure the bit hasn’t moved, then it might be an issue with your PC.

Here is my experience with that: For quite some time I ran my CNC with a PC that met spec, but barely. Everything ran fine as long as I didn’t go to fast. When I did, I had issues with the z-axis ‘slipping’ and getting more and more off as the system missed steps. Since all the compute happens on the PC side if anything slows down the computer too much, then the G-code doesn’t get to the controller board quickly enough, and weird things (like this) start to happen. In my case I replaced the PC and all the issues vanished. Now I’m going to upgrade to the EX Controller, and things will be even better.

Since your problem is intermittent, perhaps try rebooting the PC and run it again and see if the problem goes away. Also, you might recheck the PC tuning utility that Avid provides to make sure the computer isn’t doing anything extra.

I may, of course, not be right about this, but I hope it helps.
Good luck,
Garth

Thanks Garth. I’m inclined to think it is a PC issue over a bit/collet problem. I hardly ever shut the laptop down but rather put it in sleep mode.

I ran this exact project file without problems on the first run. Second time was the problem. I didn’t attach pics but you can literally see how the 1/4” end mill gradually ramped down into the project surface almost 1/4 to 1/2 inch from zero. The tool path was only supposed to be .0625 deep!

When I first built my Avid, using custom electronics, I had a step/dir interface between the controller and the driver boards. Despite the wires only being about 6 inches long, noise was just bad enough that I could watch the X axis crawl across the gantry when it was idle as the step line got triggered.

I was able to reprogram the step/dir interface to be a quadrature interface, which is far less likely to do this. I know my old generic stepper drivers had a jumper to do this; I wonder if the Avid systems can.

For what it is worth I saw this happen a couple of times as well, also on an underpowered computer. Unfortunately I never figured it out, and chalked it up to mach4 being weird. I really wanted to blame missed steps, but I think that would have had it be too high and not too low. Too bad it wasn’t a loose collet, that would have fit the symptoms perfectly.

That’s a good question. My entire electronics are plug and play, from AVID so I would think it’s solid and all bugs were worked out.

I didn’t mean to offend, but it’s just the troubleshooting process. We have no idea what your skill level is or what you have tried and verified. That is why I started with the physical root causes before moving into the realm of magical pixies. Once you get past the physical there are so many variables all we can do is shoot from the hip.

I’ve never had an issue with “missteps” that didn’t turn out to be something physical like a loose belt or grody collet. However I built my PC from individual hand selected components, to eliminate that variable as much as possible. I also run a completely gutted image of the Windows 11 OS, so that helps as well. I even wrote PowerShell scripts to stop everything from the endpoint protection to Wifi when I’m running Mach4. Probably overkill…but that is how I do.

In your case it’s adding steps, which is even less logical. Computers do what you tell them too, so adding Z steps didn’t make sense. I guess it’s all relative to where you zero from, so maybe it is logical. It could be missing steps on the up or down and just losing its place. A missed step could be anything from a physical cable, hardware in the control cabinet, hardware on the PC, firmware on the components, the OS, or any number of ancillary software running.

The only thing that doesn’t make sense is why would it only lose steps on the Z axis if it were a computer\software issue? I’d wager your G-code is doing a heck of a lot more in X and Y. Statistically you would be more likely to see missteps on those.

Nevertheless, I have been a Computer Systems Engineer for nearly 20-years now and I have seen quite a few unexplained gremlins in both software and hardware. The variables are overwhelming, but I have always found it’s best to start simple and work up from there.

Good luck!

I learned the hard way that Mach4 needs to have frequent PC restarts or shutdowns to be reliable. After about 6 months of following the same Sleep protocol as you I started to have all kinds of unexplainable events. At first I thought it was the ESS or some kind of electronics issue but it was simply Mach4 becoming corrupted even though I shut the application down after each session. Without a PC restart once in a while it must retain some things in memory and eventually becomes corrupted. I have had no more problems since I have started to shutdown the PC after every session.

Couple of things to look at:

https://www.avidcnc.com/support/troubleshooting/mach4ControlsProCnc/zAxis/#z-axis-height-lowers-unexpectedly-during-a-program

There have also been reports of Z axis “drift”

A solution that has helped some users is flipping z direction pin:

Yes sir. AVID troubleshooting flow charts indicate a loose collet problem. Since that wasn’t it, it would have to be the Z axis itself or the computer is having issues. Even though the PC more than meets specs, I’m thinking that’s it.