No Z Motion On Probe or Probe Test - DRO Indicates Motion

plasma001.ini (tried various values for those w/o values)
[AXIS_Z]
MAX_VELOCITY
MAX_ACCELERATION

[JOINT_3]
MIN_LIMIT
MAX_LIMIT
MAX_VELOCITY
MAX_ACCELERATION

Tried various values for these on Z and X axis (for comparison) with scope or LA monitoring 7i96 outputs.

DIRSETUP = 5000
DIRHOLD = 5000
STEPLEN = 1000
STEPSPACE = 5000

Added placeholders for ohmic probe and ohmic enable in 7i96 configuration tool and latter commented them out in plasma001_connections.hal and plasmac.tcl (attached) after Z wouldn't run. (Also the ohmic input activations were indicated on plasmac UI as breakaway sensor - didn't know if that was normal - assumed it was since Ohmic probing not activated).

Removed and then replaced kinstype=BOTH in plasma001.ini

Imported someone's SheetCAM materials to plasma001_materials.cfg

Grabbed PCW bit file (zip) from email again, extracts without error, matches file from which I flashed.

Flashing the 7i96 again now - only because I didn't think about it until now - but not expecting anything as there were no errors indicated during the initial flash.

Table configured for 20x20in to test in the middle and have a couple feet on either side of X. Step scales are round numbers (from memory). Metric belts now but 3/8in XL, 1/2in ACME Z lead, and ABN ordered. 'course non of that matters - but explains odd numbers...

Reinstalled and updated LinuxCNC on current PC 3-4 times according to plasmac manual directions. Used new and/or reused 7i96 configuration tool settings.

Z motion while testing with parallel breakout board worked, as did Z switch input, but couldn't figure out settings for Z switch to work properly (and didn't care then because I wasn't going to cut with BOB and without THC).

Haven't powered up plasma power supply with 7i96 attached so noise isn't an issue. Cables all 100% shield Beldon with single ended/common/star termination.

Thanks!

STEPLEN = 1000

is a guaranteed no-go on a G540
I would leave it at 5000

Changed STEPLEN to 5000 (don't know why it was 1000 when all the others were 5000) and still no z motion.

Flashed again (only thing I hadn't tried) with supplied 7i96_g540dpl.bit and now motion on all four joints! Changed config and see output on stepgen 4 now, so all known issues disappeared.

Thank you to everyone who has helped.

The bad thing about having one parameter like step length too short is that
any needed changes that you tried would not work so you cannot make any
progress diagnosing the issue.

Also when you program the firmware, you must either power cycle the 7I96 or use
the --reload mesaflash command or you will still be running the previous firmware.

Changed and tested with STEPLEN = 5000 and no Z output from LinuxCNC configured for stepgen 3 or 4.

Flashed board the second time got output when configured for stepgen 3 and 4.

Changed STEPLEN back to 1000, still have Z motion on stepgen 3, changed config and see output on stepgen 4.

Cycled power after both flashes, per instructions, and probably 50 to 100 times during week between the two flashes.

Not judging Mesa boards by this one issue, and nobody else should. Still have machines (mine and others) from which ongoing SmoothStepper and/or Mach3 must be eliminated. 7i96 with the G540/P1 firmware is only interface I'd consider at this point in time.

Not sure what happened but mesaflash verifies the flash content after writing
so re-writing the same firmware (that was successfully written by mesaflash) really cannot
make a difference, so something else is involved.

Previous issues reappeared as soon as base to which 7i96, G540, PS rails, etc. were mounted was installed in cabinet. Soon discovered that the slightest amount of finger pressure on Mesa's FPGA would cause axis issues to appear and disappear. Couldn't see any defects under microscope. Went to friend's cellphone shop with better optics and we couldn't see anything. She re-flowed FPGA, adjacent components, P1 pins, and it hasn't failed since. Didn't take long to dial in THC and successfully cut many small nests.

Ethernet SmoothStepper/TMC3in1 will be reinstalled due to it's manufacturer being able to provide support for their hardware.

This experience was an extraordinary waste of time, money, and effort (including a 5 hour road trip in the middle of a pandemic) to get a basic I/O card to work. The sole benefits were finding that LinuxCNC has greatly matured over the past couple years, plasmac is pretty awesome, and providing a baseline that makes Windows/ESS/THC/Mach combo that much less of a nightmare...

Umm, you may not have known but you did receive support from the manufacturer right here on this thread.
PCW is the owner of Mesa.

Given the opportunity, I'm sure he would have replaced the hardware under warranty....