Damaged Capacitors in VFD 2

[rockman7892]

I have a 30hp 480V Siemens MM440 VFD that recently failed. When pulling the VFD out of service to investigate, it was obvious that something has been smoked for there was black soot all over the enclosure and the control boards.

Upon taking all the boards apart and looking at them one by one we found that it appeared that only two of the capacitors were all that were damaged. These capacitors appeared to be burnt up and was obvious that they were the cause of the failure. Everything else seemed to be fine, and it appeared that these capicitors smoking were the cause of the soot elsewhere.

What would cause capacitors to fail suddenly like this? Is it possible that they were subject to overvoltage? What is the common cause of capacitor damage in VFD's?

I should mention that this VFD is on a HRG system limited to 5A.

We have suspected possible voltage transients in this area for a while. In some cases overnight when the VFD was not in use the drive would trip out on a DC bus overvoltage. This led us to belive that there were some transenits occuring in the system subjuecting the DC bus to overvoltage.

I'm going to put a fluke 1735 power meter down in the area to monitor voltge. The meter will only pick up events lasting for 1/2 cycle or more. Do you think the meter would pick up a voltage transient or are typical voltage transients shorter than 1/2 cycle?

 

[rockman7892]

Davidbeach

Like I mentioned the L1 readings may be misleading because I belive that the probe is damaged. I have seen this 12V reading on another solidly grounded system for L1. Also when measuring L-G on L1 with a regular meter we see 277V (measured on different section of MCC).

But I do agree that 371V would indicate a possible ground fault. Why would the L-G readings only be 371V for a ground fault condition? With the system neutral tied to zero (grounded) wouldn't we still see 277V L-G even during a ground fault. I thought that only on an ungrounded system you would see L-L voltaegs for L-G readings on ungrounded phases? Why would we be seeing 371V as opposed to 480V if the system was ungrounded?

It is strange to me that the voltge swells and then returns to normal periodicaly. Could this be a re-striking ground fault? What could we do to find this ground fault if it is intermitent?

 

[davidbeach]

Fault resistance and location on the circuit relative to point of measurement can have the unfaulted phase to ground voltages anywhere from nominal phase to ground up to phase to phase. Your grounding resistor will drop 277V for a line to ground fault adjacent to the transformer; that's how it limits the current.

 

[rockman7892]

davidbeach

You are saying that the grounding resistor will drop 277V for a L-G falut, so the other two unfaulted L-G readings should still be 277V? Is this correct?

 

[davidbeach]

The resistor drops 277V, so neutral is now 277V to ground. Other two phases remain 277V to neutral, and are therefore 480V to ground.

 

[LionelHutz]

It does appear there is a ground fault on L1 and that would be the most likely problem that would cause the other 2 phases to have a voltage rise like that.

However, without a proper measurement of that phase no-one can't say for sure. I recommend you get your equipment fixed as soon as possible. You write you've seen this probe problem on another system so obviously it didn't just come up broken when you went to use it this time. If you had a proper log of all 3 phases we'd know if it was a ground fault problem on line 1 without there being a "maybe" thrown in.

You could try putting a voltmeter with a min/max feature on the L1 and try turning different equipment on and off and see if the voltage varies. You almost need a helper and radio communications to do this though. You could also do it with the logger if you track the times you turn each thing on and off then go through the log and see if switching something matches to the voltage changing.

 

[ozmosis]

Rockman7892
Sorry for the delay in responding; my world this week has been shaken, stirred and turned upside down.

You asked earlier "what could happen if you didn't break the Y cap link". Well, if you get a ground fault on a HRG network, then the problem you first posted could be one.
I'm slightly alarmed that you think there might be a problem with your measurement equipment. If you think this is the case, then don't use it. Get comething else that is calibrated and works. Apart from the errors you might get in measuring, think of safety when measuring.

 

[rockman7892]

OK I went through this system today and heres what we found:

As I showed in the attached plots there was something occuring on this system causing the L-G voltage to read as high a 371V and be sustained during certain intervals.

After walking through the system piece by piece we found that there was not direct connection between the HRG resistor and the neutral of the transformer. This was because the neutral brought out from the transformer was landed on a terminal block on the HRG package that did not have continuity through to the Resistors. We went ahead and made sure that this connection. So it appears that this whole time were were indeed operating on an ungrounded system.

After connecting this neutral properly we stareted motors one by one and watched L-G voltages with our meter to look for a rise. We finally got to the last motor and after starting it, wha-la our L-G voltages went up to 371V and our ground fault indication package (now working properly) indicated a ground fault. Sure enough we wanted and meggered this motor and found that indeed the motor was grounded. We see the ground fault current returing through the EGC's back thrgouth the HRG resistors and through to the neutral properly system seems to be working properly.

We still get however strange readings when measuring L-G voltages during this fault event. Before the motor is turned on and the fault induced all voltages L-G read 277V. When motor is turned on and fault is present all three phases L-G voltages jump to 305V with a regular hand held meter. I would suspect that if one of the phases is grounded that phase should read 0V L-G. What should the other 2 phases read? Should they stay at L-G potential or should they read L-L voltaegs for L-G readings? Why are all three phases L-G readings being driven up when ground fault is present? Does this seem right? All phases measure 480V L-L during fault. Something still seems wrong.

We measured 305V with a handheld meter as mentioned above, however the fluke 1735 meter we were using read 371V L-G. Why would this be different. Something to do with meter input impedence.

The other thing that I am questioning is the fact that there is anotehr larger 75hp drive connected to this same MCC. Why wouldn't this other drive have suffered the same consequence as the one that failed. The other dirve i am referencing is what was feeding the faulted motor.

Thanks for all the help!

 

[rockman7892]

Took some more readings today and found that when measuring current on the ground bus of the MCC we appeard to have about 2A of current even when the faulted motor was not running. This 2A does not appear back at the neutral of the transformer however. It appears that there is maybe a circuilating current on the MCC? This current does not change much when the faulted motor is running which I would suspect it would increase due to the fact there would not be fault current flowing through our HRG system and should be pulsing from our kit to try and locate fault. We can see the pulsing current back throgh the HRG package and at the neutral of the transformer but do not see it on the MCC ground bus. This leads me to believe that this fault current is returning to the transformer via another path. Maybe earth ground?

Also when taking readings I put the meter in Harmonics mode and measured a large amount of harmonics on the system when the faulted motor and its associated VFD were running. We measured 1st, 3rd, 5th, and 9th harmonics, and the total harmonic distortion was about 20% which seemed very high.

 

[ozmosis]

Total "Current" harmonic distortion? 20% seems low if that's the case. If THVD, then yes, it seems very high.
The VFD should not produce triplen harmonics (3,9th). I would say other 1ph devices of a non-linear load are producing those.
Current Harmonic distortion will be high with the MM440 as there is no DC link choke facility and unless you have an AC line reactor, the typical THID will be in the region of 55-60% at full load.

 

[rockman7892]

Ozmosis

The 20% was the THD for the voltage. We saw these harmonics when the motor was in a fault condition and running. We replaced the motor and ran the new motor and saw all of these harmonics disapear on the system.

The harmonics we did see on the system with this drive and motor running were indeed 3rd, 7th and 9th. The 3rd harmonic was the dominant one at about 20% with the 7th and 9th only being a couple percent each.

Would a faulted motor maybe cause these harmonics?

 

[ozmosis]

I wouldn't expect a faulted motor to cause the harmonic distortion but maybe an unbalanced supply would introduce triplen harmonics. I don't have too much experience on HRG when faults occur but maybe davidbeach/others can.