Uh oh, the UPS is beeping again! is it harmonics?

Status
Not open for further replies.
I work in an office building and all the loads here are non-linear producing harmonics: fluorescent lights, computers, and small
split unit HVACs (heat pumps.) Almost every day, our UPS start beeping like they would if the power was off. But the power isnt off -
I measure it at 227 - 231 volts on a British system (nominal would be 230v 50hz) and I measure frequency at 50.3 hz. Neutral to ground I measure as 1.6v. Whatever is happening cant be measured on a regular Fluke and lasts for 5 minutes maximum usually. Not having a good RMS meter or anothing other than a regular Fluke meter, I think its a harmonics problem and im considering buying some sort of harmonic filter to try to fix it. Do this sound about right? :?
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
130312-0636 EDT USA

capybara:

An RMS vs an average reading meter is not your instrumentation problem.

A Fluke 27, an average reading meter, would work fine to measure MIN/MAX values. I can't quickly find the response time on my many years old 27. New specifications on the model 28 are:
Peak transient capture 250 microseconds (28 II)
Min/Max Avg Yes
Reading hold Yes

Rather than questioning harmonics I would rather phrase the question as transients.

You might just watch a 25 W incandescent bulb connected to the circuit when the beeping occurs. Does it flicker? I suspect you want to look for short duration voltage fluctuations as the starting point.
 
Location
NE (9.06 miles @5.9 Degrees from Winged Horses)
Occupation
EC - retired
Loads like bigger laser copiers/printers can cause quite a VD.

We watched a POCO reviewing a PQ recording they had done at a school a few years ago. Luck of the Dumb, I noticed a recurring problem that I had seen while playing with a borrowed Fluke 41B a few weeks earlier. POCO tech was still looking at the results while my helper & I had the found the guilty copier. The copier was causing excessive VD on woefully undersized feeders from one end of the school to the next. The low end undersized UPSs could not handle the swings in voltage.
 

brian john

Senior Member
Location
Leesburg, VA
I seriously doubt harmonics ever, harmonics like grounding are seldom the issue. As noted above if the UPS are beeping I would see if they beep on a sequence which as noted can be a printer, copier......

Get a Fluke RMS DMM and Clamp-on with a Min Max function, put the clamp-on at the source (panel branch circuit breaker) and the DMM at the load.
 

__dan

Senior Member
Lots of times the UPS will have a beep code, like Morse code, that you can look up in the factory documentation. Most common problem with small UPS's would be old, worn batteries. Check the battery age, and if older than three to five years, they could be gone. Change them. Check the basics, rating, running load, surge and transient load. Use the kW rating, not the kVA rating.

If you're on a big generator, generator power should be pretty clean except for switching transients. Does the UPS just beep or does it beep and go to battery.

Most good quality modern electronics are low harmonic loads. Cheap junk electronics can be a problem.

Also, lighting, refrigeration, fans, and the AC splits should not be loaded on the UPS. If so, your topology is wrong.

The neutral to ground voltage at 1.6v is a problem I would investigate further. Under 250 millivolts N to G would be max for me.
 
Last edited:

robbietan

Senior Member
Location
Antipolo City
I seriously doubt harmonics ever, harmonics like grounding are seldom the issue. As noted above if the UPS are beeping I would see if they beep on a sequence which as noted can be a printer, copier......

Get a Fluke RMS DMM and Clamp-on with a Min Max function, put the clamp-on at the source (panel branch circuit breaker) and the DMM at the load.

I would not dismiss harmonics outright. I had come across a case where the UPS have a harmonic input limit of 3% VTHD. as most of the loads in the building are electronic, we measured the VTHD going to the UPS at around 3.5% VTHD. enough to trip the UPS unit but well below the 5%VTHD limit the IEEE prescribes
 
Last edited:

__dan

Senior Member

Pretty much the same reasons pton listed. It indicates a problem that should not be there. Every time I've seen 1.5 v N to G like that, the system bonding jumper was missing at the service, usually a bonding bar that needed to be rotated 90 deg and bolted in.

With the UPS, depending on the UPS type, the output could be separately derived and require a N to G system bonding jumper at the output. If it's not done properly, it will most likely find a haphazard higher resistance N to G path through connected equipment that also references N to G.

The N to G voltage indicates the N to G system bonding jumper needs to be checked to see if it's in place as required (most likely). It could also indicate other things, like the circuit passing from one building to another that do not have their steel structures bonded together, misconnected neutrals, bad neutral connection under load.
 

brian john

Senior Member
Location
Leesburg, VA
I would not dismiss harmonics outright. I had come across a case where the UPS have a harmonic input limit of 3% VTHD. as most of the loads in the building are electronic, we measured the VTHD going to the UPS at around 3.5% VTHD. enough to trip the UPS unit but well below the 5%VTHD limit the IEEE prescribes


On a plug and play UPS?

Can't say it does not exist but in the 100's of harmonic surveys we have conducted never seen this issue. Harmonics are seldom the cause, harmonics with grounding are catch phrases thrown an unknown issues. I never dismiss them but try to rule them out early so we can find the real cause.
 
Last edited:

brian john

Senior Member
Location
Leesburg, VA
Excessive N to EG voltage could indicate on overloaded neutral, poor connection or excessive length. The overload could be from harmonics or someone tapping onto any white wire they could find in a j box somewhere. Meaning more than one hot from the same phase using one neutral.

Some VD should be expected under normal operating conditions.

You llimit N-G Voltage to 250 millivolts about 5 amps at 30', for each branch circuit?
 

GoldDigger

Moderator
Staff member
Location
Placerville, CA, USA
Occupation
Retired PV System Designer
You llimit N-G Voltage to 250 millivolts about 5 amps at 30', for each branch circuit?

Possibly the suggestion was to limit to .25 volts N-G on an unloaded branch circuit, regardless of the loads on other circuits. That is certainly doable if the neutral and ground are at the same voltage back at the panel. :)
 

brian john

Senior Member
Location
Leesburg, VA
Possibly the suggestion was to limit to .25 volts N-G on an unloaded branch circuit, regardless of the loads on other circuits. That is certainly doable if the neutral and ground are at the same voltage back at the panel. :)

Yeah, If you have no load the only thing you would get if the neutral was grounded downstream from the N-G Bond would be a lower reading if the reading was .25 VAC at the transformer, which it should be lower I would think at the Transformer.


Studies were done were they had 15 VAC between neutral to ground with no asverse affects on sensitive loads.

I had a site that had an ungrounded MG Set, at some point an "A" phase branch circuit shorted so they had an "A" phase grounded system. The site ran that way for some time (no one knows how long) this was a SCIF (skiff) full of very sensitive electric equipmment, no adverse affects.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
130314-1017 EDT

brian john:

To some extent I have to disagree.

Sensitive equipment is probably not the correct description for devices that will have a problem with differences in EGC voltages. For the moment I do not have a suitable name to group devices with ground path voltage problems.

One device with a problem is an oscilloscope. The chassis of this instrument is connected to the power cord EGC, and to the common of the electronics in the scope. The common or "ground" lead of the probe also is essentially at chassis potential. Do measurements on something, like a computer, that also references its electronics to its EGC, and then there are problems from the difference in potential of the EGCs. You might argue that one should use a differential input. Not always practical, or workable. Thus, one opens the scope EGC and lets its chassis float.

Another example. A computer with direct RS232 connection to a CNC machine. Big noise problem and data errors. I have discussed this before.

Then there are ways to build some sensitive instruments, millionths of an inch displacement gages, that can operate with large ground voltages and provide accurate measurements.

.​
 

__dan

Senior Member
250 millivolt N to G is a simple go, no go, test after you've installed a new computer power circuit and are doing a first power on test with conventional equipment, like a Fluke 87 (we install wiring). If it's more than that, I would immediately start backtracking the power, all the way to the service if necessary. I would have to look at it before I would say it could be VD under load. You could check N to G resistance, which Fluke says the 87 will do even with an open voltage on the line. I would assume the voltage would obscure the resistance measurement and the open voltage is what I want to see. If I see more than 250 mV, I will check the installed path integrity first for pass, and check loading, VD second if the ground path looks good.

This is not the same as the low resistance bonding path test mentioned in the other thread, where a constant 10 amp current is injected into the bonding path. They would be looking for the end point to end point voltage raise not exceeding some limit set by E=IR and the specified bond path resistance low limit. That is a more specialized dedicated test.

You walk into a raised floor room with rows of loaded racks. The room draws 100 kW from the UPS. You see push carts loaded with stacks of yanked boards and blades labeled 'bad board, bad port'. This is heavily validated IT equipment that is overbuilt and overtested to not fail in deployment in a production IT environment. I'm not saying bad grounding causes this. I'm saying the grounding will go on the list of possible causes and this list is processed by logic and the process of elimination, until a single logical cause remains, and liability for the cost of the failure is assigned to it. When that happens, it does not matter if the grounding is good or not. What matters is that it meet code (it does not).

In the OP's case, do you think he knows where the N to G system bonding jumper is required to be located? At the service? At the UPS? At some other separately derived source? And the integrity of the GEC and the bonding to the local IT steel? Once you look at and find your grounding path, you can clamp it looking for stray circulating current. It indicates a problem that requires further investigation.
 
Last edited:

brian john

Senior Member
Location
Leesburg, VA
250 millivolt N to G is a simple go, no go, test after you've installed a new computer power circuit and are doing a first power on test with conventional equipment, like a Fluke 87 (we install wiring). If it's more than that, I would immediately start backtracking the power, all the way to the service if necessary. I would have to look at it before I would say it could be VD under load. You could check N to G resistance, which Fluke says the 87 will do even with an open voltage on the line. I would assume the voltage would obscure the resistance measurement and the open voltage is what I want to see. If I see more than 250 mV, I will check the installed path integrity first for pass, and check loading, VD second if the ground path looks good.

T.

Why would you have neutral to ground voltage on an unloaded system? Unless the system was improperly grounded (neutral not grounded), or some one hacked into your circuit with a load?

VD in an unloaded branch circuit?
 

__dan

Senior Member
Why would you have neutral to ground voltage on an unloaded system? Unless the system was improperly grounded (neutral not grounded), or some one hacked into your circuit with a load?

VD in an unloaded branch circuit?

???

You're asking a question you already know the answer to. A sharp eye on the DMM could flag problems that would otherwise go unnoticed, as you state yourself in the post above, you found trouble. On a simple power up test you would check H to N, H to G, are you saying you would have no interest in checking N to G? I have found VD on a power up test. High resistance new breaker contacts were dropping 4 volts at no load, and I saw the slightly low voltage as a problem, checked it, and changed the breakers with no hesitation. That's why I always test with an 87 and not with a wiggy.

Or you are called to the site to investigate trouble. The networked computers just will not stay running more than ten minutes (for over a year) and the credit store manager is ready to move all the PC's out to the parking lot and run them over with her car. Everything else has been checked by the computer hardware maintenance vendor. So of course I checked the grounding, plus whatever else I could think of. The strip mall was several structures built separately but giving the look of one building. I wrote the report, gave it to the customer, and as requested called the landlord with the grounding deficiencies. The store circuit ran through several structures that were added over the years and not bonded together. Traced back to the panel, the feeder grounding conductor flew out of the drop ceiling in front of the main distribution and just lay on the floor in a coil unconnected to anything. So, there were technical issues to be fixed, and when the customer digested the report and asked me if I thought that was causing the network outages, I said "No". It was four hour travel and I had exhausted what I could do, I left.

Two days later I was in another state cabling another (same) credit office and I look at the PC, the guys got skotch tape on his thinnet lan connectors. The office manager sees me eyeing this and comes over, "do not touch that or breath on it, my network will go down". Bingo, lightbulb lights in the brain. I call my customer, "hey I'm in WV and this guy needs skotch tape to make his lan work". I say "that's your problem in Delaware, bad lan crimps (they had used lineman pliers for the crimps). The problem took several years and many people looking at it to fix. It's the process of elimination.

In the early days of computer equipment, I was told more than 2.5 volts N to G would start to fry the circuit boards.
 
Status
Not open for further replies.
Top