Cross post from Labview forum; Noise issue with prox sensor/counters on PCI-6602/Lenze AC motor control

[Ralph@NES]

http://forums.ni.com/ni/board/message?board.id=170&message.id=265421&jump=true

The short story:  I'm using a Cherry Corp prox sensor to record RPM on a shaft on a primarily steel rig.  The rig is powered by a large AC motor which is connected to a Lenze AC Tech VFD/speed controller.  My sensor is connected to a PCI-6602 counter on PFI 30 and I have the correct 470 ohm resistor from +5v to signal, and my ground is connected to DGND and RGND.  I've been able to get successful RPM readings on my desk with this configuration and a handheld drill.

The sensor which is on the rig goes haywire when the VFD/motor control is switched on.  We've wired the motor direct and bypassed the VFD and had no issues with interference, it's only going haywire when the VFD control is on.  Random huge numbers in my frequency readings, few correct ones, and it seems to go in a slow sine wave.

Any ideas for things to consider?  Extra grounding?  Shielding?  Does anyone out there have any ideas for negating this condition?  We're chasing building grounds and trying to eliminate steps but I'd be very, very glad to hear any ideas.  I'm really up against a wall here time-wise.

[Henrik_Volkers]

What type of sensor are you using?

Cherry make magnetic field sensors, and near a motor you probably con have strong magnetic fields flying around, however since it's gone without the UF-converter/controller it might be something else...

Do you have a good, short direkt grounding from the controller to the motor?  Maybe a shielded cable?

The noise currents are generated in the controller and (normally) marching out through the output cable , so you should spend some nice copper with as less inductivity as possible to give them a chance to easely go back to the controller

My last jokers are 5B Moduls to create an additional galvanic isolation and break up ground loops 

[Ralph@NES]

The sensor is magnetic in nature but acts with any metal object, a true prox sensor and not one that senses a magnet, so to speak.

It's definitely ok with the motor running only.  The only time it occurs is when the VFD is controlling the motor speed.

The total cabling, grounding included, from VFD to motor, is about 3' long and sufficient gauge.

So shielding the controller-motor cable would be the best bet?  The thing that confuses me is that there are other VFD's and motors nearby that are running with no effect at all, and their cables are unshielded.  It seems to be this particular one.

[Henrik_Volkers]

What mean 3' long for someone how nows only metric? 3 foot (bigfoots?) 3inch ??

VFDs come in two tastes, some with some without output filter.... Have you tried another one, maybe the filter doesn't work ?

My guess about the grounding cable was just because people are laisy. If you put in and out the VFD, why not only change the cabling for the three phases and keep the ground where it was ?!..   I'm pretty shure that there is a big nice grounding screw right beside the VFD output, and it makes sense to use it

How is the reading if you (can) place (hold) the sensor near the VFD driven motor without contact to the rigg?  

According the sensor: 90-95% grounding problems ... BUT just an idea: If the sensor acts with metal objects, then by a change in the electric or magnetic field (field weakening by eddy currents...)  the sensor produce. What if the one of the switching frequences of this VFD correlates with the frequence of the sensor?    Just like some TVs goes mad if your new poversaving electronic lamp hits the IR main frequence

[Ralph@NES]

Yes, we used the grounding screw.  3' ~ 1 meter.

This unit doesn't have an output filter to my knowledge, but neither would the other ones nearby so that's probably not the issue.

Now switching frequencies, that's an interesting idea.  This does have the option to chance the operating frequency.  Maybe I'll look into that.

In the meantime we're rerouting the motor wire in conduit and shielding it just to eliminate that variable.

Thanks for your input!

[AnalogKid2DigitalMan]

Hi Ralph:

I posted this in the original thread, probably a bit late to the party:

Hi Ralph:

If possible, try mounting the prox to the rig so it is insulated from ground. And if you have an oscilloscope around that may prove helpful at looking at the signal while you try different things.

Also try to route the prox cable away from motor leads and drive. Do you know if the prox cable is shielded?

Where does the prox get it's power from? You may want to scope the power out too. Motor drive noise can be a bugger to deal with sometimes.

[Ralph@NES]

I've shielded the prox the complete run right to the end.  Tried isolating the ground as well.  I am powering the prox with my NI system, it's referenced to my PCI-6602 5V out.  The prox cable isn't anywhere near the VFD.

Per a conversation with the VFD manufacturer, apparently the problem is likely to be in the output frequency of the unit and is eminating from the VFD-motor cable.  We're going to try routing that cable through conduit and also checking with an oscilloscope.  Not sure how to use one, but here goes!

Thanks for the input.

[AnalogKid2DigitalMan]

Ralph:

Make sure your conduit is steel, not aluminum; it has better shielding properties.

Belden also recommends using motor leads that are shielded, see:

http://www.belden.com/pdfs/Techpprs/Evaluating%20VFD.pdf

EDIT:

PS, suggest that if you are not familiar with how to use a scope, PLEASE DO NOT attempt to learn looking while at VFD drive signals. High voltage/current can be dangerouse/lethal.

-AK2DM

Message Edited by AnalogKid2DigitalMan on 08-17-2007 11:14 AM

[Ralph@NES]

One of the other guys is pretty handy with the scope, I'll have him attack it.  Thanks again.