how do I measure power 'cleanliness'?

[groovin]

One of our cages in a remote location has servers constantly complaining about small power outages. We suspect the power in there is not clean or flaps. Can we use a multimeter of some kind to log any power anomalies? I am not sure how this would have to be set up... would the multimeter simply plug into any open outlet on suspect PDU strips, or would it actually have to sit between a server and the PDU?

of course the faciility will check it out on their end, but i was wondering if there was a safe, scientific way of collecting some metrics on my end.

thanks!

 

[pm]

In our lab we used a power line distrurbance analyzer to monitor a fairly rare power spike event that we were seeing. It turned out to be the power locks on the lab doors switching over (like when someone would activate their badge) at the same time as two or more of the compressors were running.

We used something like one of these:
http://www.testequip.com/sale/categories/01F.html


I was having a problem with my home computer causing a brown out when I powered up my 22" CRT... I solved it by putting the monitor input on $50 Belkin UPS from Office Depot. Before that, if I turned on the monitor while my computer was running, sometimes the computer would reboot. Afterwards, everything was fine.

 

[KMurphy]

Google "Power Quality Meter". You could rent one out, or hire an electrical testing and consulting company to supply the equipment and set it up. If the sags are coming from the utility, they will have to solve the issue. If the sags are caused by customer owned equiment, then the cheapest way to resolve the issue would be to install Sola constant voltage transformers (CVS or MCR series) to feed the distrubtion panels for the servers.

A simple multimeter with "MIN/MAX" recording functions may not be fast enough to capture the transient sag condition. You need a high resolution data acquisition device with configurable triggering. An oscilloscope would also work for very short duration transients.

If a large motor starting on the utility's system is the cause, they (the utility) will track the owner down and force them to address this issue. You would also need the oscillogram to present to the utility as proof of what's occurring.

 

[MrDudeMan]

Originally posted by: pm
In our lab we used a power line distrurbance analyzer to monitor a fairly rare power spike event that we were seeing. It turned out to be the power locks on the lab doors switching over (like when someone would activate their badge) at the same time as two or more of the compressors were running.

We used something like one of these:
http://www.testequip.com/sale/categories/01F.html


I was having a problem with my home computer causing a brown out when I powered up my 22" CRT... I solved it by putting the monitor input on $50 Belkin UPS from Office Depot. Before that, if I turned on the monitor while my computer was running, sometimes the computer would reboot. Afterwards, everything was fine.

I had a similar problem when the washer or dryer would turn on at my parent's house. I used the same fix (buying a UPS) and it wasn't an issue after that.

 

[Erik in sac]

An oscilliscope that has a peak/min hold or record feature could be left on in real time to show any spikes or large DC ripple

 

[cheesehead]

I'd second the scope.

IMHO, you're better off just getting a great big isolation transformer, and put some very tiny capacitors (a few pf) on both sides. It's sort of a "brute-force" solution, but isolation transformers are cheap if you get 'em used.

 

[KMurphy]

Originally posted by: Cheesehead
I'd second the scope.

IMHO, you're better off just getting a great big isolation transformer, and put some very tiny capacitors (a few pf) on both sides. It's sort of a "brute-force" solution, but isolation transformers are cheap if you get 'em used.

That's almost what the Sola constant voltage transformer is. It utilizes an LC circuit to create ferroresonance and can maintain 100% voltage output with incredibly low input voltages. It performs even better when load is much less than the transformer's rating. It is also very reliable; much more reliable than a UPS as there are no batteries to change or worry about open circuiting when needed. It's only flaw is that it is horribly inefficient as it gives off a lot of heat from the eddy current losses.