Powerful 7-8 kHz ringing. CSI inverters and high capacitance cables

[Skogsgurra]

I found this waveform (measured L1-PE) in a 400 V grid:

Fifteen years earlier, I found this:

The ringing could also be seen on the MV side of the transformers.

The utility guys said there was nothing to see. Their Fluke and Dranetz analyzer didn't show anything. That is understandable as they seldom show anything above the 51st harmonic and the ringing is between 140th and 150th harmonic.

The CSI inverter draws current in distinct steps, as shown in second picture, and these steps initiate the ringing.

The resulting high frequency harmonics heat the transformer cores so they are a lot hotter than the windings. Measured between 30 and 40 C higher than normal core temperature.

The high frequency also produced more capacitive stray currents in the resin insulation and that also caused high stress in the voids (yes, low quality - Gonella transformers, now out of business), which produced PD and ozone so that the transformers exploded.

I havent had this more than two times in my career, so it doesn't happen very often. But I guess that some of you guys have seen that, too.

Any comments?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

Back in university I spent quite a bit of time playing with a single-phase CSI driving an induction furnace via a matching transformer. I got to build most of it from scratch, which was enjoyable and educational: I doubt I would be allowed to build such a thing today - molten metal, high currents, water-cooled coils, 3-phase power... Your CSI looks like a multiphase type based on the number of steps in the current waveform - a motor application?

I recall that snubber networks fitted to the thyristors could be excited into a resonant condition with the DC link reactor, and that it was actually possible to start the CSI using leakage through the snubbers to establish current flow in the DC link. The gate frequency to cause this effect was much higher than normal operating frequency, which in my case was only 1kHz - 2kHz, but ringing due to the snubbers was present well below the frequency at which the CSI would self-start. I wonder if you're seeing a similar effect?

I'll have a look and see if my old notes are still around somewhere, it's been close to thirty years since they were written and I'm not sure where they are.

 

[Skogsgurra]

Thanks, Scotty.
The 2002 case was, and still is, a 1200 kW pump motor. CSI inverters are sturdy.

The other one (December 2017) is not known. Haven't been able to read the nameplate yet. But most CSI inverters are built for rather high power.

Hope that you find your notes.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

Hi Gunnar,

Well, I found my book of handwritten notes but I have none of the scope traces. They must be in the actual report, which I have so far failed to find. My recollection above seems fairly close, there was a definite interaction between the snubber networks and some of the circuit reactances which caused high frequency ringing.

My speculation above that the snubbers were interacting with the link reactor may be wrong, the resonant frequency would be too low: more likely it is the load commutating reactance and/or coupling transformer leakage reactance they are resonating with, at a frequency of a hundred kHz or so judging from my scribbled waveform. Each switching event in the bridge caused a decaying ringing waveform similar to that which you're seeing but in my case at a much higher frequency. I wish I could find the prints from the old Gould DSO, although I suspect the ink will have faded to the point of being invisible.

 

[Skogsgurra]

Makes sense. The cable capacitance (calculated) was 870 nF. A wide-spread distribution in a paper mill.

Hope that others will chime in when Monday dawns. There must be other cases than the two I have seen. Even if they are few and far between.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.