Need Help to Improve Surface Finish

Ok, code looks good in simulation mode using UCCNC (seeing if Mach was lying to you). As you said it's a series of arc moves so surface finish isn't in the code.

Not the spindle, even though you recently replaced the bearings and might have possibly gotten that wrong. If that were the case you'd be breaking tools or seeing the same problem on straight axis cuts too.

As you say, it's probably not backlash, as I think that causes out of round at the direction changes and not just a general surface finish problem on arcs. On the other hand, backlash-induced surface problems may not show up in a straight axis cut if the axis with slop is snug due to cutter pressure.

So... If not mechanical, and not the code, then what's left is Mach and your motion control hardware. If you want to rule out the VFD making noise and causing the stepper drivers to freak out, stick a fine point pen in the spindle and have it trace on paper with the spindle off. Just do a series of big arc moves and some direction changes. You might be able to see if there's any jittering in the lines.

The pen might be too coarse to see problems, but a scribe on a flat piece of aluminum going down 0.001" might be fine enough if you put a thin layer of dyekem on it first. Another thing to consider would be to see if your Z is oscillating up and down when the VFD is on. Using another flat piece of aluminum or MDF (or something else really flat), run back and forth with the spindle off and a DTI attached to the head and indicating the surface. Then do it again with the spindle on (no tool needed) and see if the needle is bouncing more. I would think that VFD noise would show up on that axis as well as the X/Y.

I'm also running a 126 and 107 board, so I don't think there's an issue there with steps making it through your BOB to the drivers. If the 107 board were freaking out, you'd hear the spindle throbbing as the RPM changed. Besides the VFD, motor, 126 & 107, what else have you changed? My guess is most of the wiring is new or different, but you'd obviously know better than I. Have you followed best practices and isolated all of your signal wires from the VFD? Is the VFD in a separate enclosure from the BOB & ESS?

I guess another thing to try would be to make a fairly deep cut in two (or three) depths. Then check the surface with a magnifying glass - if the surface lines line up everywhere, the machine is being commanded (by something) to generate that surface you can completely rule out lost motion or mechanical issues. There's no way noise or mechanical problems would duplicate a surface like that in the exact same location over two or three passes.

Seriously not an expert here, but maybe something will help.

-S
 
Ok, code looks good in simulation mode using UCCNC (seeing if Mach was lying to you). As you said it's a series of arc moves so surface finish isn't in the code.

Not the spindle, even though you recently replaced the bearings and might have possibly gotten that wrong. If that were the case you'd be breaking tools or seeing the same problem on straight axis cuts too.

As you say, it's probably not backlash, as I think that causes out of round at the direction changes and not just a general surface finish problem on arcs. On the other hand, backlash-induced surface problems may not show up in a straight axis cut if the axis with slop is snug due to cutter pressure.

So... If not mechanical, and not the code, then what's left is Mach and your motion control hardware. If you want to rule out the VFD making noise and causing the stepper drivers to freak out, stick a fine point pen in the spindle and have it trace on paper with the spindle off. Just do a series of big arc moves and some direction changes. You might be able to see if there's any jittering in the lines.

The pen might be too coarse to see problems, but a scribe on a flat piece of aluminum going down 0.001" might be fine enough if you put a thin layer of dyekem on it first. Another thing to consider would be to see if your Z is oscillating up and down when the VFD is on. Using another flat piece of aluminum or MDF (or something else really flat), run back and forth with the spindle off and a DTI attached to the head and indicating the surface. Then do it again with the spindle on (no tool needed) and see if the needle is bouncing more. I would think that VFD noise would show up on that axis as well as the X/Y.

I'm also running a 126 and 107 board, so I don't think there's an issue there with steps making it through your BOB to the drivers. If the 107 board were freaking out, you'd hear the spindle throbbing as the RPM changed. Besides the VFD, motor, 126 & 107, what else have you changed? My guess is most of the wiring is new or different, but you'd obviously know better than I. Have you followed best practices and isolated all of your signal wires from the VFD? Is the VFD in a separate enclosure from the BOB & ESS?

I guess another thing to try would be to make a fairly deep cut in two (or three) depths. Then check the surface with a magnifying glass - if the surface lines line up everywhere, the machine is being commanded (by something) to generate that surface you can completely rule out lost motion or mechanical issues. There's no way noise or mechanical problems would duplicate a surface like that in the exact same location over two or three passes.

Seriously not an expert here, but maybe something will help.

-S

Thanks for your input. Yes, my wiring is new with the electronics installation. All of the spindle control wiring has the drain wire grounded in the VFD enclosure. Limit switch drain wire is grounded in the BoB enclosure. Driver drain wires are grounded in the Driver enclosure. I make it a habit not to bundle power cables with control wiring and do my best to separate them as much as possible. Hope I did this right.

I don't think it's bearings either. The preload is in the ball park because I can run the spindle at 8600 on long runs (more than an hour) I get less than 30 deg. F temperature rise.

Tomorrow I'll run the tests you suggest and see what I get.

For everyone's reference I've attached a few pictures of my control cabinet.

Here's a shot of my BoB/ESS enclosure. 115vac enters at the top. Limit switch and 107 control wiring, ethernet cable, and driver to BoB wiring at the bottom.
20170729_142757.jpg 20170714_152848_resized.jpg

My VFD enclosure. 107 board wiring enters at the top. 230vac comes in on lower left back (black and white wires) and exits to the motor lower left front. PT cable enters above the high voltage wiring.
20170806_165915_resized.jpg

View of my control cabinet. Power panel and e-stop on the upper left. Power supplies and driver are in the lower rollaway drawer. VFD enclosure is lower left and BoB/ESS enclosure is lower right.
20170806_170834_resized.jpg
 
I created a 4-1/2" diameter circle and generated gcode so I could monitor Z axis movement as Spumco suggested. The DTI is contacting the top of the moveable jaw on my 5" Kurt clone vise that is held down to the table with four strap clamps. The attached video shows the DTI needle bouncing around quite a bit. Looks like I have some harmonics being fed back into the table. Source of my poor surface finish? Steppers maybe? BTW - this test is with the VFD off.
 

Attachments

  • Z Axis Test.mp4
    20.7 MB · Views: 23
Now we're getting somewhere. Great video, BTW.

Next thing I would check is to see if the PMDX-126 is outputting signals to the Z-driver while you run that test. Open the cabinet and see if the Z pin LED is flickering while the X/Y are doing the circle thing. If it is, then open the Mach diagnostics page to see if Mach is outputting on those pins even though your code isn't commanding any motion.

Or try it with the Z-driver unplugged and see if the Z is being commanded (again, by noise or by Mach or something else) to move or if you're getting mechanical resonance. You can also try it at a different feed rate - I find my steppers hum quite a bit at about 12-15IPM, and I've read that the typical 2-pole steppers are prone to this at about this speed. Can't remember where I read that, but there you go. Maybe it was one of Tormach's white papers, which are excellent by the way.

Run the test(s) again and then we'll start speculating on root cause. And pray it isn't a combination of mechanical AND electrical gremlins...

-S
 
Now we're getting somewhere. Great video, BTW.

Next thing I would check is to see if the PMDX-126 is outputting signals to the Z-driver while you run that test. Open the cabinet and see if the Z pin LED is flickering while the X/Y are doing the circle thing. If it is, then open the Mach diagnostics page to see if Mach is outputting on those pins even though your code isn't commanding any motion.

Or try it with the Z-driver unplugged and see if the Z is being commanded (again, by noise or by Mach or something else) to move or if you're getting mechanical resonance. You can also try it at a different feed rate - I find my steppers hum quite a bit at about 12-15IPM, and I've read that the typical 2-pole steppers are prone to this at about this speed. Can't remember where I read that, but there you go. Maybe it was one of Tormach's white papers, which are excellent by the way.

Run the test(s) again and then we'll start speculating on root cause. And pray it isn't a combination of mechanical AND electrical gremlins...

-S

Yeah, I think we've hit on something. I'm going to run the test again with the VFD on and spindle turning and see happens. Yes, I'll mount the DTI on the head before turning on the spindle. Then I'll run it and check the BoB Z axis lights.

I ran the DTI test at 5, 10 and 20 IPM. Didn't notice any significant difference in needle bounce.

And I'll take a look at the Tormach white papers.
 
Awesome stuff Tom and SpumCo... Nice collaboration there for sure. I am not into the smooth-stepper sides much as we moved to Centroid and a server based system for our upgrade. Sounds like you have the source about nailed down. I hope this fix is an easy one!
 
As much as I'd like to claim victory, I don't think anything is about nailed down at this point. We know it isn't X/Y backlash, and it isn't spindle bearings. And we know the code is good-to-go.

Depending on what he finds in the next round of testing we may find it's noise (from what?), mechanical vibration (from what?), the ESS flipping out, or Mach causing the issue.

Or some combination of the above, which would suck.

Tom may get lucky and find something silly like a loose drain wire, but if I'm involved my typical luck will rub off and he'll be chasing his tail for a while.

-S
 
*updated for clarity*

Add to when you are troubleshooting and you get into the ground wires try disconnecting all the drain wires except the control wires from the 107 to the VFD. Ground issues can be tricky to figure out. We went through ground issues with the router/mill we built for the school. Our issue was different but we found that even though we thought we had done the grounds correctly we created a condition that was worse than not having a ground at all.


Sent from my SM-G955U using Tapatalk
 
Last edited:
I created a 4-1/2" diameter circle and generated gcode so I could monitor Z axis movement as Spumco suggested. The DTI is contacting the top of the moveable jaw on my 5" Kurt clone vise that is held down to the table with four strap clamps. The attached video shows the DTI needle bouncing around quite a bit. Looks like I have some harmonics being fed back into the table. Source of my poor surface finish? Steppers maybe? BTW - this test is with the VFD off.

I'm not sure I would read too much into the needle movement running across the top surface of your vice. I would be more inclined to mount the TDI on the head or dovetail plate and measure against a fixed point on the column and take away any variables of the saddle, table and vice.
 
I see in your photos that the enable wires are not connected to the BOB, are they still hooked up to the drives? If so disconnect them at the drives. Also make sure the bare ends are not touching the metal back panel. On the BOB enclosure is the 110vac earth ground attached to anything else besides the back plate?

What we learned from the school ground issues was that terminated wires or improperly terminated wires acted like antennas, the grounding system was undersized and it was easy to create ground loops.

Where the motor drain entered the metal enclosure we had not terminated them at the enclosure wall but had instead ganged the drains inside the enclosure to a single ground lug. By not terminating the drains at the enclosure wall the drain wires inside the enclosure acted as antennas picking up interference from the inside.

Another issue was all the components of the system were grounded but the ground was not robust enough and we had daisy chained the grounds. The best strategy is to do a star ground system where each enclosure or device like the VDF and the machine are grounded to a single point separately and to not daisy chain the grounds. We moved to a 3/8" wide braided ground strap that was grounded to the machine structure, electronic enclosure case and VFD base as separate runs being careful not to inadvertently create loops. The VFD on this machine is mounted outside of the electronics enclosure. The electronics enclosure also contained the main power disconnect so it was also terminated to earth ground. The electronics enclosure was divided inside by full depth metal baffles to separate the high voltage from the low voltage and again for the signal wiring. Basically separate Faraday cages in one enclosure. In the course of troubleshooting the original build we tried separating out the the encoder termination boards in a separate metal enclosure but in the process mistakenly created a ground loop which made things worse.

On my PM-932 CNC conversion I removed any unnecessary wires (stepper enable) and any unused wires from the system. All the motor wires, limit switch wire, and spindle control wire drains terminate at the metal enclosure wall. There are no shielded wires or drain wires inside the electronics enclosure(s).
 
Back
Top