Are there any other large pieces of equipment that are connected to the same circuit?  If you have a large motor trying to turn on, it can pull down the power line and cause a current spike when the motor turns off.  Is the heating/AC connected on this circuit?  Try to look around the building and see if there is anything that is turning on every 30 seconds.  If so, then that might be what is causing the spikes!

Assuming that this is the problem, the solution would be to move your sensitive computer equipment to its own dedicated power line.  BTW, keep the UPS in series with the power.  It is a lot cheaper to replace a bad UPS than the whole network server!

Keep us posted.

it looks like that there is a load connected to the UPS that you are not aware of aside from the network server. it is possible that a motor trying to start up in cycle like what malone said. one way to check is to shutdown your server and measure the UPS output current...if you measured something then for sure there is a load connected which you don't know....is there a recent construction in your site? UPS power outlet should be identified and only CRITICAL load should be connected to it..

do you have static bypass on UPS?

dydt

Possibly a laser printer?

For a temporary fix (until you can track down the mystery load), you can reconfigure the server to begin the shutdown process when the UPS batteries reach 20% State Of Charge or so rather than shutdown immediately after the glitch.  It usually takes 5 minutes or less to fully and safely shut down most servers.

Some servers are configured to begin the shutdown process shortly after the UPS transitions to battery power, while others are configured to 'ride through' the power anomaly on UPS power, then transition back to mains when it is available and stable.  The particular configuration is typically set within the UPS client application on the server.

Some probable causes in addition to what has been posted above are:
1. A leakage impedance/resistance at some location that heats up and cools off periodically.
2. A unique type of resonance.
3. Some loose connection(s).
4. A defective component, e.g. an inductor, a capacitor, IC, etc.
5. Etc.

Thanks for the suggestions guys. I still have not found the phantom component that is causing this sag / swell in this circuit. I do know for a fact that there are no other loads on this circuit. The hot, neutral and ground run un-interuppted all the way back to the panel. When I unplug the Server, this circuit is absolutely clean with no sags or swells. This is why I tend to believe that some component in the Server is causing this problem. However, Busbar did suggest a Laser Printer maybe the cause. We do have 2 Color Lasers that are on the network that I will have to check out. So far the Server is still running fine without the UPS. Our company's network guru suggested that maybe we should change the power supply in the server as a stab in the dark but my feeling is that power supplies are either good or bad, not intermittant. Am I wrong? Is this a worthy attempt or a waste of time. I am sure that I could put the scope on the power supply to check it but the company will not give me the time to shut the server off to do this. They prefer to let the server run until it finally pukes.

I'm not sure if I followed your discussion of what configuration the equipment was in when the problem went away. UPS completely unplugged?  UPS still plugged in but disconnected from server?  When you saw the problem did you see it on input to ups or input to server? I might throw in two farther-out ideas:

#1 - Is it possible that the controls on your ups cause it to charge internal batteries only intermittently (once every 30 seconds).  Maybe only causes a problem only if there is a load on the ups

#2 - Is it possible that system voltage changes are causing your current surges, rather than the other way around? (not entirely consistent with your discussion)

Perhaps you can monitor the current at several locations to try to narrow down where you can see a current surge and where you can't.

hi shot gunner..

a fully loaded UPS should be able to charge the batteries at the same time. what is the nominal bus voltage that supplies the inverter and the batteries? it is possible that you are draining your mains supply when a motor or printer is trying to start..is there a transformer tap adjustment that you can bring your voltage up...normally a a standard ups has a +/- 10% of the nominal voltage input..you can adjust the voltage ( if there is transformer ) to maintain a good supply of dc voltage to te battery and the inverter...

when you said you are running without ups..did you totally disconnect the ups and connect the servers to the same mains supply or used a static/mechanical bypass of same ups..

the first thing i will do is monitor the voltage that supplies the inverter and charges the battery and compared it with the ups specs...if the rectifier voltage drops below the battery voltage then you have to start checking your charging/rectifier circuit, transformers or mains supply..

good luck
dydt

Thanks for the reply electricpete and dydt,
Here is the current situation: As of this moment the only component that is plugged into this designated circuit is the Server. The UPS has been totally eliminated from the mains.The incoming power still has the inrush / voltage drop every 35 seconds. Voltage is steady at 119.5 volts / current draw at 8.5 amps and then every 35 seconds, current spikes to 35 + amps and voltage drops to 102 volts. Duration must be a nanosecond as no circuit breaker trip occurs. When this problem first happened and the Server was shutdown, I removed the UPS and the Server from the outlet and monitored the voltage for several minutes and the line was absolutely clean with a steady 119.5 volts + or - a half a volt. The only time I see this spike is when the Server is plugged into this circuit. I have determined that the UPS can not handle this large of a spike probably due to the inability of poor batteries that cannot be fully charged and / or improper set-up of the UPS ( Ihave no manual for this UPS ). But is the power spike a common occurance for network servers or is this server about to die???

Suggestion: Please, would you post the server:
1. Manufacturer
2. Type or Model
3. Catalog number
4. Serial number
5. Year of manufacture
6. And other information that you may consider helpful.

hi again..

you mentioned that even if the servers are connected straight to the mains supply you are still getting a voltage drop as low as 102 volts...i suggest you investigate where your feed is coming from..

do you think that the laser printer is causing this?
is it possible to connect the printers to the main line( or another line) and connect the servers to the UPS..

what is the brand of the ups..is it the ffg..

merlin gerin
exide
emerson
sola
or deltec?

what size of ups you have...and if you have your calculated load on it..


dydt

Hi!

Bottom line, shotgunner, there is a capacitor and/or motor involved in the aberrent activity somehow. You have already stated that the line is pure when nothing is plugged in. Then, when you attach the server, the activity begins. QED, something is wrong with your server.

I have seen this exact behavior before in a locked-rotor situation with a small ventilating fan. It is trying to start, the rotor is locked (won't turn), and the thermistor, sensing the rapid overheating of the motor, cuts out. The thermistor cools down, the motor tries to start again, etc., etc. The cycle is very uniform, just as you describe. Some larger servers have cooling fans that run straight 110v instead of loading the power supply. Does yours? Is it locked? What happens if you unplug it from the computer?

I have also seen this with capacitors in power supplies - but usually along with catastrophic failure...does yours have two supplies, perhaps, that cross over?

Anyway, if it is not a cooling fan of some sort, it is your power supply on the verge of a magnificent death. Replace it (them) QUICKLY.

Hope this helps.

Hello everybody.  I have followed this thread and it is very interesting, since the behavior is constant.
My suggestions:

1.- Try connecting the server to another branch and check
if you have the same spikes and dropping voltage, of course try that the only load is the server (don't want to damage other equipment)

2.- If you have the same variations, monitor the voltages of the power supply of the server (+5, +12, -5, -12 Vdc) to check if they change considerably.  With these you can make sure that the problem is the power supply, and thus replace it with justification.  I would almost assure it is the power supply, since it is cyclic and most probably be a problem with the power supply, which I assume is of the switching type.

My 2 cents.

Regards.

There are at least two different categories of UPS, online and standby.  The cheap individual UPS is likely a standby UPS that switches to battery and inverter upon the loss of main power.  An online UPS rectifies the power, stores to battery and the inverter runs from the battery all the time, often while synchronized to an alternate source.  The loads run off the UPS that automatically switches to the main power when the UPS fails.  Usually there is a maintenance bypass switch to manually select mains when needed.

John

This thread is getting rather old, but if you haven't solved your problem, it would appear to me that 18 volts is a pretty serious drop across a supply line that is connected to a quasi infinnit buss.  This leads me to believe that you have some resistance in the line, probably due to a loose connection at one of your electrical outlets or in the pannel at the breaker.  As previously suggested, try your server on another branch cct.  Tighten the connections in the main panel... perhaps replace the breaker.  
Also, try checking the voltage at the pannel.  You shouldn't see any significant voltage drop drop between the panel rail voltages and those at the UPS.   

I may say you have two faults to locate. first is that you have a thermally controlled cooling fan on the server unit that could be dried out of lubrication on its bearing or shaft bushing.this is causing a heavy draw of current when its called to run. second is that you have a loose connection on the power input circuit. probably on the plug or outlet where you tap the server. If you made the voltage measurement on the outlet or rececptacle where you connect the server, then you have a loose connection at the power feeder or on the distribution panel where you get the power.Circuit breaker internal contacts develope resistance due to heat.do investigate.tighten all termination physically not only look at it with eyes.human eye cannot identify most electrical loose connections.try to measure your voltage fault at the distribution panel board to verify your fault location.