Dual Y gantry CNC router slowly getting out of square

[ryanjg11]

I'm finishing up my CNC router build and I'm air-cutting now in preparation for making the spoilboard. I'm using the AVID PRO4896 hardware kit, but with my own custom electronics box utilizing Acorn and Clearpath SDSK servos. This machine utilizes dual servos on the Y.

In CNC12, I have properly setup "Auto Homing and Auto Squaring with Software Pairing Procedure" and have confirmed the machine is very square after this routine is completed. However, once I start air-cutting a project (in this case, a simple but fairly long spoilboard file which bores holes for mounting hardware to attach the spoilboard to Avid's aluminum extrusions), I notice it progressively gets out of square, especially on the Y Master side. After running this program for about 10 minutes, I "Cycle Stopped" the program and commanded the spindle back to WCS Y0 X0, and noticed Y was now about 1/2" off the spoilboard. Also, measuring the gantry for squareness shows its now out of square a similar ~1/2".

I went into the CNC12 wizard just to confirm my turns ratios were the same between both Y axis (though I'm not sure if this matters because I think CNC12 just maps Y1 values to Y2 once you set it to pair and mirror).

For the Clearpath servos, I believe I tuned the X servo and simply applied that file to both of my Y servos, since there's no easy way to tune them while they're both running. The settings for each Y servo are identical; both are configured for 6400 steps/revolution.

Both Y axis are using AVID's Pro Rack & Pinion Drive, NEMA 34. Tensioning appears identical on each side, and manual movement of the gantry doesn't indicate any issues with the drive system. I attached the motor pulley to the servo using the key and Loctite 648, so I highly doubt there is any slippage at this point.

Here's the weird part: when I square the machine and repeatedly command it to go back and forth across the entire Y length, it comes back to the exact Y0 position. Which got me to thinking if it might be an issue with my Gcode or post processor. I'm using Fusion 360 free (here's the Fusion 360 spoilboard file I was using. For the post-processor, I'm using "CENTROID Milling swissi-005."

Any ideas what might be causing this progressive out-of-square issue?

P.S. Latest report.zip attached.

[RogDC]

Just a couple of things to check -
Make sure you use loctite on the homing screws and flag stops on the bolts
Calibrate the turns per inch ratio on each motor and adjust for backlash
Do you have the Clear Paths Position Recovery enabled ?

If you haven't already, post your latest report

[ryanjg11]

Make sure you use loctite on the homing screws and flag stops on the bolts
Calibrate the turns per inch ratio on each motor and adjust for backlash
Do you have the Clear Paths Position Recovery enabled ?

I just posted the latest REPORT.ZIP to the original message, thanks.

The homing screws have been loctited and also are secured with a nut. They aren't going anywhere.

I'm pretty sure my Turns Per Inch ratio is correct, but I haven't corrected for backlash. I know that's common for ball screws, but is this something I still need to correct for on a rack and pinion Y drive?

I do not have Position Recovery enabled on the Clearpaths. My understanding is this feature is only utilized when the servos lose their enable (bus power). They don't appear to lose power through the entire program. Though I am only running one IPC-5 power supply, and during certain parts of the cutting process, all four servos are in use. Could this make one of my Y servos starved for power? Wouldn't that cause the servo to shut down and result in a 9031 DRIVE FAULT?

[ShawnM]

I have a ShopSabre 5x10 plasma table with 4 Clearpath motors and only one IPC-5 power supply and it can rapid all four motors to the “park” position at over 1000 IPM on the X and Y axis with no power issues and no drive faults. The gantry alone has to weigh a few hundred pounds.

For the record, you have it backwards when it comes to backlash. Backlash is far more common in a rack and pinion system than a ball screw system for several reasons. Helical is better but still a rack and pinion.

[ryanjg11]

I have a ShopSabre 5x10 plasma table with 4 Clearpath motors and only one IPC-5 power supply and it can rapid all four motors to the “park” position at over 1000 IPM on the X and Y axis with no power issues and no drive faults. The gantry alone has to weight a few hundred pounds.

Good to know I can likely rule that out. I haven't done anything special to compensate for backlash but will give the tooth meshing a closer look today.

I'll do some more full-traverse Y movements and measure any drift with a dial indicator tonight as well.

What's really stumping me here is how I don't seem to have any drift issues when I'm moving it back and forth via CNC12 MDI commands.

[RogDC]

I've attached a test surfacing file from V-Carve Pro which should work on your machine, X-Y Zero will be left front corner of the machine. If you try this g-code file to see if you still have the issue. It may help you isolate if there is a problem in your other file, but not likely if the motors are slaved on the axis.

How do you have your motors wired for power? Are you using the Teknic DC Power Distribution Hub with 24VDC logic backup?

Are you using a CNC4PC board for the motors or straight to the acorn?

Did you go into the MSP software and verify if you have the position recovery enabled to recover any lost steps if you lose main power to the motor?

How do you have the slaved motor reversed? thru Acorn or thru the motor settings in MSP?

I have an AVID Pro 60x72 with 4 Nema 34 motors, one IPC-5, and the Power Distribution Hub. No issues with lost steps unless I don't have position recovery enabled in the motors.

[ryanjg11]

I've attached a test surfacing file from V-Carve Pro which should work on your machine, X-Y Zero will be left front corner of the machine. If you try this g-code file to see if you still have the issue. It may help you isolate if there is a problem in your other file, but not likely if the motors are slaved on the axis.

Thanks, will load the file and test tonight. Definitely haven't ruled out an issue with the file or post processor. Speaking of which, what post processor are you using?

How do you have your motors wired for power? Are you using the Teknic DC Power Distribution Hub with 24VDC logic backup?

Not using the Teknic power hub, so no logic backup. Line power comes into a rotary disconnect, then to a 30a breaker, then to terminal blocks and then to the IPC-5, going to another bank of terminal blocks and finally to the individual servos. (Attached schematic page at bottom, there was a provision here for a second IPC-5 but I'm hoping to get by with just one for now.)

Are you using a CNC4PC board for the motors or straight to the acorn?

No CNC4PC board. Step and direction directly from the Acorn.

Did you go into the MSP software and verify if you have the position recovery enabled to recover any lost steps if you lose main power to the motor?

I checked earlier today, and I am NOT using position recovery. I can enable it on all servos, but as I understand it would only help if the servos lost position without power... but I don't think they're losing power at any point between when I square the machine perfectly, and when it drifts out of square. All four servos are Enabled (green LED) through the entire process. If you think it's worth trying, happy to flip it on.

How do you have the slaved motor reversed? thru Acorn or thru the motor settings in MSP?

Pairing was setup in the CNC12 Wizard (and mirrored on one). I am not using Clearpath MSP to do this, though I could try to see if it solves the problem.

I have an AVID Pro 60x72 with 4 Nema 34 motors, one IPC-5, and the Power Distribution Hub. No issues with lost steps unless I don't have position recovery enabled in the motors.

Did you ever experience any out of square issues like this before you enabled position recovery? Would love to learn more about how this solved it for you. Thanks!

[RogDC]

When I experienced lost steps was typically on the Z Axis without position recovery enabled before I installed a gas spring to help counter balance the weight of the spindle and dust shoe, or when i would hit something which fell into the path of the gantry. Once I enabled position recovery the Z axis would move to the correct height if it dropped between jobs.

If you aren't using the Teknic Power Distribution Hub and you are losing steps due to intermittent power from the IPC-5 you may not notice any benefit from position recovery being enabled. The Hub keeps a small voltage on the buss to prevent loss of data.

You could also swap the Y master and slave step/dir and see if the problem follows the motor ( stays on same side ) or follows the Y-Master data cable.

Did you install the recommended resistors on the Step, Direction, and Enable near the Acorn ?

When I initially set up the 4 motors on 3 axis (reserving one axis for rotary) with Y slaved by cable I would experience squaring issues of the gantry.

Once I set up auto squaring and slaved the second motor in the Acorn/Wizard I didn't have any issues with auto squaring other than idle hands in the shop messing with the homing flags causing an out of square position, but the Acorn and motors did what they were instructed to do.

The file I sent was created in Aspire, using the Vectric Centroid inch PP. I rarely use Fusion 360, and when I do I import the model to Aspire or V-Carve Pro to generate the G-code files. I do most of my work in Vectric Software so it is just more familiar to me. The Vectric SW works better for my needs than Fusion 360.

[ShawnM]

Do you have the 100 ohm pull up resistors on the step, direction and enable lines for each motor/drive?

[ryanjg11]

Do you have the 100 ohm pull up resistors on the step, direction and enable lines for each motor/drive?

Using 1000 ohm resistors on servo lines.