Problem with failing lighting ballasts

[eeprom]

Hello,
In our plant, we have 5 buildings with maybe 200 similar fluorescent fixtures, all with GE ballasts. In one building in particular we have a very high incidence of failure. 5 ballasts (out of 16) are being replaced this week.

This building is the sand blasting building. The equipment in the building are one 200 Hp compressor, lights, some fans, small heaters, etc. The compressor is the only significant load. No VFDs, welders or otherwise noisy loads. The lights are the only single phase loads on the 480V bus. There are two 277V circuits (different phases) dedicated to lights. The ballasts that failed are all on the same circuit. The building is on it's own 300 KVA transformer.

GE says the failures are due to the voltage swings caused by the compressor cycling on and off. They told me that the voltage could spike up to 360V when the compressor shut off. I have no idea where they came up with that. The transformer is fairly small, and I estimate that the voltage would drop 20% on starting. But I don't know that voltage sags would hurt the ballasts. I don't know how often the compressor cycles, but it is a compressor so it probably cycles quite a bit.

I had considered that the sand blasting building has a tremendous amount of very fine silica sand, All of the failed ballasts are in fixtures near the main blasting area. All the fixtures that failed are in the same area and on the same phase. GE said that the sand couldn't/wouldn't cause the failures.

I'm out of ideas, and I don't have time to turn this into a research project. Does anyone have any ideas what could be causing the ballast failures?

 

[crthompson]

You may want to use an O-scope to capture the peak voltage spike when the compressor shuts off. I would then try to get an idea of how many times the compressor cycles per day. Then compare the results to the absolute maximum ratings on the ballast spec sheet. If it really is the compressor that is causing the problem, it may be cheaper to replace the old ballasts with the next size up as opposed to installing dv/dt filters or something of the like...

 

[eeprom]

Please explain this voltage spike concept to me. Are you suggesting that the spike is being caused by Ldi/dt of the transformer winding?

EE

 

[MikeHalloran]

I wonder if the compressor control has a means to unload the compressor (dump the compressor output to atmosphere instead of making it push air into the tank) before the motor contactor opens.

I wonder if it used to have such a feature, and it got repaired out.



Mike Halloran
Pembroke Pines, FL, USA

 

[crthompson]

Whenever you energize a transformer, you have a current inrush and the voltage will dip. Whenever you de-energize your compressor load, the current drastically drops and as a result the voltage will spike for a short period of time. I think that those repeated spikes are degrading the electronics/insulation in the ballasts over time. If GE was close, then you are applying almost 130% of rated voltage in short bursts, several times a day. I don't know if you have access to an Oscope or not, but if you could capture the max peak voltage at the moment when the compressor is shut down, it should give you a pretty good indication of what is happening. You may be able to accomplish this with a multimeter that has a min/max hold function.

 

[burnt2x]

I'm with Mike here.

Please check if the compressor cycles on and off. My experience tells me compressors with that type of loading needs only to unload (vent off when air banks are full and load up when air bank pressures drop to a set level), not put off. That way you do away with erratic voltages.

 

[eeprom]

It sounds like the voltage spike issue is a reasonable answer. I didn't believe GE, but now it does sound like a possibility. I don't have a good scope, and I don't have the time to build a data collection circuit. For the time being, I will replace the ballasts with new ballasts with more appropriate ratings.

I might also look into having the electricians install caps on the ballasts. A 0.1uF capacitor with proper voltage rating might do the trick.

Thanks for your help.

 

[gattie]

In the area that the affected ballasts are mounted, is the ambient temperature much higher than elsewhere?
Also how are the ballasts mounted?

 

[eeprom]

The ballasts are all fairly equal in temperature. And the temperature ranges between 40F and 100F. The ballasts are mounted inside of the fixtures. I don't know how they are mounted on the inside, but I assume they have 4 machine screws holding them to the top of the fixture.

 

[John2025]

I would think a 200HP compressor for sandblasting would be a screw compressor. Ours don't shut down the motor, just modulate the intake. I suppose it could be a big ol' recip though.

 

[jcaden]

Wouldn't the compressor affect all 3 phases, not just this one with the failures? Are the phases balanced, ie voltage not too high?

 

[eeprom]

I don't really understand the problem fully. But I do see how an inductive circuit could cause a voltage spike if the current were to change drastically. The current in all three phases will not be equal at the time of the motor stoppage. One phase may be at a current of zero, while another would be at about 80%. The phase with the zero could would not cause a transient.

Another thing I don't know is whether all of these ballasts failed all at once, or over some time.