I'm having a heck of a time getting the last of my noise out of my plasma build, need some ideas.

First the specs;
scratch build table with ac servos and drives from DMM tech (Dyn4 drives)
Hypertherm Pmax85 with cpc and serial port
CandCnc Mp3500 DTHCiV EtherCut

Next the issues;
My arc volts on the Mach3 screen is floating between 1-10 volts with no arc fire.
My Z motor has a jerky downward drift to it.

Things I've found;
When I had full 3 phase power to the drives the arc volt reading was 2-18 or so volts, so taking the generated 3rd leg (Phase perfect digital converter for the shop) off cut the phantom volts in half.
If I only power up the Z motor and drive the motor is stable and quiet with no drift
Power up any other drive and the Z starts drifting.
Using an O-scope I can see noise on the Step/dir connections on the MTA board when the drives are powered up, drives off, no noise. Any drive on, all step and dir connections have noise.
If I unplug the MTA card from the Mp3500, Z drift stops.
I have tried switching Z motors, drives, cables, but the drift always stays with whatever is plugged into the Z step/dir connections.
I've tried putting a EMI noise filter ont he PC and Mp3500 supply lines, no change
I have made sure the ground plugs from the Pc and Mp3500 are not in continuity to the table ground. I have a PC ground that goes back to the electrical cabinet on the wall, and a table ground that goes to a ground rod next to machine. The PC ground currently has the PC, Mp3500, seros and drives connected to it, while the table ground has the pmax ground and gantry ground connected to it.
I even tried noise filtering line magnets on the db25 cables going to each drive with no change.

I really need to get this thing running and making parts, but it is kicking me. I've been emailing tech support at both Candcnc and Dmm tech, but have only managed to narrow down where the noise is coming from. I'm going to try moving the PC, Mp3500 and mta card outside of the electrical cabinet on the machine to see if that helps next.

Ok, so I moved the servo motor and drive ground connections over to the table ground. Still had the drift issue.

I pulled the step/dir breakout board and mach3 ubob board out of the cabinet as far as I could (only have short cables going to the drives from card), and plugged the controller, PC, and hub into a separate wall plug not from inside the cabinet and I think the servo drift issue is gone.

The noise is still there and is still causing an arc volt reading with torch off. I'll try upping the divider ratio to 20:1 next to see if that makes it less or stays the same. Maybe when the shielding sleeve comes tomorrow I can fix up a few of the signal wires and maybe help the arc volt issue some more as well. Plus I still have to hook the 3phase power back up to the drives to see if that changes anything but I just might be getting it working yet. I'll just have to forgo an attached control console and build a small separate cabinet on wheels or something for the comp and stuff to sit in and plug into the wall instead of the noisy cabinet.

When I hear about all these different grounds my first thought is ground loops. The NEC actually required ground loops for safety in power grounding.  For signal grounding you want just one ground point for everything so noise can only go one way, to the one ground point. If the driven ground helped you can be sure you don't have the grounding done right. You need to check all of your cable ground wires and shield grounding wires to make sure they only connect to ground at one end. Any ground wire that is connected to ground at both ends is a ground loop. Lift every ground wire from the ground point and measure between the wire you just lifted and the ground point. If you get a reading you have found a ground loop. Check every ground wire and shield this way.  sensor manufacturers often connect the cable shield to the case and when you mount them you have a ground loop. Pcs often carry a ground through Usb cables too.

What does your wiring job look like? Neat bundled wires with signal cables not running parallel to power conductors and as far apart as possible? Single conductors twisted together to provide cancellation?  All this stuff matters when you introduce a really noisy device like a plasma torch to a machine.

The problem seems to be coming from the Dyn4 servo drives.

I got the dirft issue fixed by switching the step/dir signal from +5v from the drives to +5v from the controller. and I shielded the wires from the controller to the drives with ground at one end and good seperation between each.

Current issue is faint arc volt reading with no arc fire. If I turn the drives off but leave the plasma powered up the arc reading is 0 like it should be, so its not coming from the plasma. As soon as I turn on any 1 drive the arc volt reading is back to around 20 volts.

I tried unhooking the torch cable, still 20 v reading, same with unhooking the ground clamp lead. Somewhow I need to bleed the noise off that the drives seem to be generating and backfeeding to the plasma. Servo motor cables are shielded and grounded at motor end. drives, motors, and plasma are all tied to table ground which is hooked to ground rod beside machine.

I think it has something to do with the Phase Perfect 3 phase converter that uses PWM to generate the 3rd leg of power and the drives are amplifying the noise or something.

YOU have a ground loop running.  Groundin at the motor  end is NOT a good idea as the is no easy path back to ground from teh motor for the moveing gantry . try Ground it from the drive end only. The way you have everything grounded it not always a good idea. SOME things need to stay isolated from creating a ground LOOP. The plasma  Work Grnd lead is NOT  a ground .  Safety grounds and sheilding  grounds are 2 different dogs .

Just a thought, (;-) TP

The motor cables are from DMM tech, they have a ground wire that runs inside of them out to the motor end where it then hooks to the shield, at the drive end that internal grounding wire is hooked to the drive chassis which is then hooked to the table ground, exactly as the book says and the DMM tech support says. The gantry also has a separate grounding wire back to the star connection for table ground, as does the X carraige.

And it sounds like you have a loop right there There are 2 ways to reach ground?? . It is OBIVIOUS you have a loop problem from the voltage readin you re getting. Take a decent voltmeter and find where it is coming from. The OEMs don't always see the WHOLE picture on grounding for noise. It can be a tricky critter to run down and eliminate. Sometimes you CAN"T fix it do to design of the machine so you live with it or redesign it. Sometimes you can filter it, drain it away , find teh source and eliminate it there.



(;-) TP

I'll try unhooking the extra ground and see if it helps.

Sometimes it just helps to argue online as you get forced to look at something from a dif perspective and see something you missed after staring at the problem for hours and hours.

No change.

And thinking about it, it wouldn't change as each motor has a ground wire, and basically if the carraige and gantry were electrically isolated they each would still have 2 motors on, and thus 2 ground wires, Y1 and Y2 on gantry and X and Z on carraige. And i thought ground loops that caused noise were more from being connected to 2 different earth ground points, like the "dirty" ground rod at the machine and the "clean" ground rod back at the electrical panel on the wall. I've checked for those types by disconnecting ground wires and checking back to the opposite grounding point they are supposed to be hooked to.

From what I can tell and have been told by tech supports the noise I am dealing with is more related to high frequency caused by PWM driven loads, which might by exagerated due to my Phase perfect  phase convert which makes the 3rd leg by PWM means as well, though nothing else in the shop has noise issues. But then both the machining center and turning center probably have transformers in that might isolate the noise....

Try some of these. Put one on each cable everywhere.

http://www.bgmicro.com/ferritesplitbeadwithsnapsleeves.aspx