Data closet monitoring results.

[itsmoked]

I have a "data room" that is filled with servers, routers, etc., in a multistory building. It runs with no problems for several months then takes a hit that causes most stuff in the closet to go down. This is a serious problem as stuff like phones and online stuff promptly fail.

It is a three phase closet. It has two lightly loaded phases 3-8amps and one unloaded phase <1A.

After several days of logging everything-under-the-sun there was one hit where one phase dropped to %80 percent V and the other dropped to like %90 but these didn't trigger a "closet event".(I haven't figured out the hit's duration but less than the many power supply carry throughs.)

I do have data coming later this week that shows a "closet event".

I looked at all the data +100 pages and saw nothing out of the ordinary except the ground... As seen below.

This is not the neutral! This isn't what I expected.. A discussion would be helpful. Thoughts?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[MikeHalloran]

Suggesting that there's a steady ground current of ~2.5A continuous, but it drops to ~300mA once a day? And spends several percent of a day above 10A?

Am I reading that right?



Mike Halloran
Pembroke Pines, FL, USA

 

[itsmoked]

That is how I'm reading it.. And why it disturbs me. I'm working on clarification.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[waross]

Hi Keith;
I am concerned with the 99.26 amp hit at 18:30:15, on Sept.22, '06.
A spike of that magnitude has the potential to trigger a "closet event" even if it did not this time.
I would suspect a faulty surge arrestor. They are normally the only components that you would expect to find deliberately connected from hot lines to ground. Do you show a corresponding current spike on either of the hot lines? Those electrons had to come from somewhere.
Can you shut down one piece of equipment at a time while monitoring the ground current If you can isolate one piece of equipment that is making a disproportionate contribution to the ground current, that may be the culprit. 5 AM isn't much fun but most of us have done it at one time or another. I suspect that you have already done more trouble shooting at inconvenient hours than most of us.
Respectfully

 

[MikeHalloran]

Can you borrow an infrared video camera?

There should be some anomalous heat associated with that current.

Maybe enough to detect with a handheld IR thermometer, manually scanned. Look for stuff that shouldn't be particularly warm, and nonuniform temperatures among similar components.



Mike Halloran
Pembroke Pines, FL, USA

 

[jghrist]

I'd check to see if the neutral might be grounded at some point other than the service entrance. This could put neutral current (including 3rd harmonics) on the ground. Have you done any harmonic measurements of the ground and neutral current?

What is the difference between the green and gray traces?

 

[rbulsara]

jghrist asked my question. May be some of the 'neutral' of the power supply transfromers are connected to ground in the equipment. Or it could be some capacitors (filters) in the circuit.

Would be interesting to see data of the "closet event". Sounds like some loose connection some where that goes to ground once in a while, long/large enough to bring voltage down to shut down equipment but not trip a breaker.

 

[alehman]

I agree with the previous two posts. In addition I would consider the possibility of an intermittent short in a power supply in one of the pieces of equipment. Some switch-mode power supplies are constructed very inexpensively.

It will necessarily be a process of elimination unless you can trace the ground current to the source.

 

[Snoball]

Take your fluke meter and clamp, prove the 2.5 ground amps are really there.
Then get a window to crash the room, unplug branch's until it's zero. Run it down.
But be sure your monitor is set correctly against the fluke meter's amp clamp.