VFD Output Ringing Mitigation 1

[LionelHutz]

Can anyone point me to any specific references concerning the elimination of ringing on the output of a VFD. I'm looking for some more detailed help here. Just the advise of installing an output reactor or a motor filter is no help.

I will give you an example. We had a call to a failed VFD mounted on the machine with a short run of Tech like cable, maybe 1', going to an output reactor and then another 8' of this Tech like cable going from the reactor to the motor. After installing the new unit, the voltage at the output of the VFD was measured as ringing to about 2600V peak. We removed the reactor and the ringing almost disappeared entirely so we left the installation like that.

So, reactors and filters by themselves don't solve the problem. I'm hoping a few people who have more experience with this can point to some ideas such as specific wiring techniques or grounding techniques that are known to help.

 

[DickDV]

Something is terribly wrong with the data you just reported, LionelHutz. That's not how ringing works.

Ringing does not appear at the drive end of the motor leads and therefore has no effect on drive components. The ringing gradually develops and grows as you progress down the motor leads toward the motor. The further you get away from the drive, the larger the overshoot and undershoot gets until, in cases that require some extra conditioning, the overshoot and undershoot are high enough to damage the motor insulation.

Putting a simple reactor in the motor leads at the drive end strips some of the higher frequency components out of the drive output pulses and increases the distance down the motor leads you can go before damaging levels of overshoot and undershoot occur again.

Adding a plain reactor to the motor leads should have no effect whatever on the signal between the drive and the reactor.

It may be that your method of measuring "ringing" is faulty. There are few meters (including true RMS meters) that will report accurate data on the drive output. I would use a 20MHz or higher scope to visually analyze the signals. Be sure the scope is rated for 1600VDC minimum.

The other possibility is that you are adding something to the motor leads that is causing oscillation. That would require some capacitance somewhere in the circuit. That also would be very visible on a scope.

But a plain reactor, no. I don't think it will do what you are reporting. Something else is going on.

 

[Skogsgurra]

There are two things to look for. Or two phenomena, rather, that often get confused. One is the reflected wave and the other one is the LC ringing caused by reactor inductance and cable+motor winding capacitance.

The reflected wave doesn't show until your cable length is comparable to the wavelength of the PWM voltage front in the cable. An 8' cable doesn't compare, too short. So, you can exclude the reflected wave.

The LC ringing also usually needs a rather long cable in order to get a capacitance high enough to create oscillations in a frequency band (usually tens of kHz) where damping isn't high enough for the oscillations to die after a few periods. Your cable does not have that capacitance.

There must be something else - as Dick says. I would look for any extra capacitance added to the motor. There might also, for some reason, be an extra length (a stub) of cable that has been forgotten or installed for some purpose. That cable's length will, even if it doesn't feed the motor, be included in the LC and make ringing possible. Or, there may be capacitors added to the motor for surge protection.

What frequency is your ringing? Does it show also when measuring between phases or only phase-gnd? What is your system voltage? And motor ratings?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[LionelHutz]

At the site of my first example, I was there and there was nothing extra in the wiring I missed. It was very straightforward. However, it was late in the day and the site personal wanted to leave so when we jumped the reactor and everything looked good that was as far as it went.

I'm looking into yet another site but I was not personally there. The installation is a 200hp, 600VAC VFD. There is approximately 100' of Anixter VFD cable to the motor. The tech captures a L-L waveform at the VFD that shows a ~13MHz 1800V peak ringing when a transistor turns off. The tech also captured a 550kHz, 1500V peak waveform at the motor when a transistor turned on. He tried some different carrier frequencies from 0.6kHz to 3kHz while monitoring the VFD end but nothing changed. To me, the two don't seem related. I don't understand the cause of the ringing on the VFD end.

We are using a 100MHz Tek scope.

 

[Skogsgurra]

Was a differential probe used for the L-L measurement? Or did your technician use a battery supplied scope?

The latter usually produces lots of artefacts that are difficult to explain. A differential probe or a good isolation amplifier is necessary to get a reliable reading.

Neither 13 MHz nor 550 kHz suits the 100' cable. Not for reflected wave, anyhow. It would typically be around 3 MHz.

It is a mystery. Does the VFD fail very often?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[LionelHutz]

Well, it's a TH720 scope. We've had decent luck with it but maybe I'll have to look into a differential probe. Actually, it might be getting time to just find a new more powerful unit.