Unexplained SCR failures

[GGOSS]

Hello all,

 

I have encountered 2 problem sites in which the SCR's within a soft starter have failed. A post mortem of the failed devices strongly suggests overvoltage damage however one would expect other sensitive electronic equipment also connected to the same supply bus would also have suffered, but this has not the case.

 

As the soft starter is used with line and bypass contactors, the failures can only occur in the short period of time when the motor is starting. This is an approximate 8 second window.

 

Any guidance would be greatly appreciated.

 

Regards,

GGOSS

[marke]

Hello GGOSS

 

From your description, it would appear that the SCRs can only have failed during the start or during the stop process. A bridging contactor and line contactor will provide godd statistical protection at other times.

 

I would be looking for potential resonance problems, i.e. high impedance supply with power factor correction connected to the input of the starter, (not using a separate capacitor contactor) or a high impedance supply with bulk correction connected relatively close to the starter and no detuning reactors, high capacitance circuit on the output of the starter, i.e. a long length of neutral screened cable.

 

There could also be a problem with the starter itself, i.e. poor or faulty snubber components, bad connections, unstable firing of the SCRs etc. Insufficient gate current / pulse width can cause the SCR to switch ON then OFF within a conduction period.

 

Can you give us more details?

Best regards,

[GGOSS]

Hello Marke,

 

Many thanks for your considered response. Application details are as follows.

 

Motor: 375kW/605A/415VAC 3-phase TEFC

Load: Direct couple Screw Compressor (Unloaded start)

Connection: 6 Wire (Inside delta) with main and bypass contactors as previously advised. The line contactor is positioned in the line circuit.

Starters: Closed loop controlled current technology.

Duty cycle: The starters are being operated well within their designed/rated operational duty cycle.

Cables Cable run to the motor would be in the order of 15 - 20 meters. Cable type & run back to the transformer will be advised. Can advise however that the cables are not screened.

PFC Equipment Certainly none connected to the output of the starter. My understanding is none connected to the line side either but will re-check.

 

A total of 3 such failures have occured at 2 seperate sites. In all cases the starters/motors are supplied from their very own transformer with negligable volts drop detected during startings.

 

In 2 of the above cases, there have been no repeat failures since replacing the power assemblies ie more than 3 months ago. As no other components (PT's, CT's, PCB's) were replaced, one can only assume that there are no issues with the snubber circuits and that the SCR's are seeing sufficient gate current. We are still working through the third case.

 

Note also, there has been 1 motor failure. Motor manufacture advised transient over-voltage damage thus supporting our evaluations of the failed SCR's.

 

Any ideas?

 

Regards,

GGOSS

[marke]

Hello GGOSS

 

There is no doubt that the vulnerability of the SCRs is higher in 6 wire connection than in 3 wire connection. In 6 wire connection, the SCRs are subjected to the line voltage plus the voltage generated by the motor. whereas in 3 wire, the voltage is distributed over two SCRs in series.

Additionally, the motor will behave a little bit like a transformer with any disturbance on one phase transformed through to another.

The OFF time during start in 6 wire connection, is about twice that of 3 wire connection for equal current flow, also increasing the vulnerability to transients from outside.

 

One thing that I have seen occur, is the change in load, i.e. a motor being started can cause bulk correction to switch in other capacitor banks and this can cause transients on the line. Are there automatic correction capacitors on the primary side of the transformer?

 

Do you know if the problem occured at start, or at stop? That could also give a clue. If the problem occured at start, was it at the instant of start, or was it well into the start.

 

If, as you suggest, the starters are fed from their own transformer and there is no other connected load, there may be some transients appearing on the output of the transformer in sympathy with the phase control of the SCRs. It would be interesting to use a very fast recording scope or similar to look for transients as the motor is being started.

 

The final possibility that springs to my warped mind, is that the problem actualy stems from the motor itself. All insulated winding wire has pin holes in the insulation. This can cause a corona discharge or a complete breakdown of the insulation. If there was a light discharge, that could result in a fast transient voltage to occur and of course this is directly across one SCR pair. Had the motor been OFF for an extended period of time before starting? how about moisture in the motor (say condensation) causing an internal discharge, creating the transient.

 

Just a few rambling thoughts.

Best regards,

[GGOSS]

Hello marke,

 

These are not rambling thoughts at all.

 

Have some questions and comments for you.

 

First of all, could you please explain what you mean by "The OFF time during start in 6 wire connection, is about twice that of 3 wire connection for equal current flow, also increasing the vulnerability to transients from outside"

 

I am still trying to get answers on whether PFC equipment is installed and what type. Will advise in du course.

 

Impossible to state whether the fault occured during start, run or stop. These are connected to PLC controlled (un-manned) compressor units. In addition to that the very nature of the product is such that it is impossible to determine time of fault ie no date/time stamping and no differentiation through fault coding.

 

The installation of fast recording equipment has been suggested, particularly as all other information available to us has proved useless in determining fault cause.

 

An issue with the motor/s would not surprise me. At one of these sites a motor failure was experienced, report suggestion insulation failure due to voltage transients. It may well be that we experienced a corona discharge as you described, however it is very unlikely that this or any other motor manufacturer would accept responsibility. Is it possible that a corona discharge could result in a winding failure which resembles that of voltage transient damage?

 

The compressors on which these starters are installed run 24/7, however they are installed in a refrigeration plant room. Moisture ingress has been considered a possible cause of failure however I have (rightly or wrongly) ruled that one out simply becuase we have hundreds of such installations around the country, all of which (with the exception of these 3) have provided troublefree operation since installation and continue to do so.

 

As I see it there are 4 things we need to consider further.

 

1. Determine if PFC equipment is installed and how it is controlled.

 

2. Install high speed recording equipment with the aim of capturing events not obvious to the naked eye.

 

3. Further evaluate the corona discharge theory.

 

4. Get out of the game altogether. Start up a news agency or fish & chip shop. The fact that I am putting this in writing may suggest it is a serious consideration. Any thought of adding sections for news agencies and fish & chip shops on this forum? ;d;;c;

 

Thanks again for your support,

 

Regards,

GGOSS

[marke]

Hello GGOSS

 

Keep going and you may find the answer!!

 

First of all, could you please explain what you mean by "The OFF time during start in 6 wire connection, is about twice that of 3 wire connection for equal current flow, also increasing the vulnerability to transients from outside"  

When the starter is conected in 3 wire connection, the SCR conduction angle varies between 120 degrees and 180 degrees per half cycle, or the OFF time varies from 60 degrees down to 0. In six wire connection, the conduction angle varies from 0 degrees to 180 degrees per half cycle and the off time from 180 degrees to zero degrees. At half voltage, the OFF time of the SCRs is : in 3 wire, about 30 degrees and in 6 wire, about 90 degrees so it is more like 3 times than the 2 times that I mentioned. The SCR can be affected by transients at any stage when it is in the OFF state. In 6 wire it is in the OFF state for longer per cycle than in 3 wire.

 

1. agreed

 

2. agreed

 

3. difficult but??

 

4. Good luck!!

 

As there has been two cases at one site, I would put my money on the supply. Open stepping tap changers changing when the load comes on or similar.

Best regards,

[marke]

Hello GGOSS

 

Casting my mind back, I recall a problem at a site in SECV territory where there were intermitant problems with inverters that we traced back to HV power factor correction switching and poor earth-neutral bonding on the transformer.

The switching on the HT would cause a sudden change in the network potential relative to earth. This was capacitively coupled through the transformer, lifting the whole inverter above earth for a very short period of time. This was in effect, the same as applying a very high voltage, very fast transient into the output of the inverter. Bipolars are not to keen on this sort of thing! If there are transients to earth, there can be sufficient capacitance to earth in the motor to cause problems.

Just another thought.

Best regards,