Weird UPS Situation

[George Stolz]

Exactly where did you check the phase rotation?

Phase rotation was checked at the output of the internal UPS transformer with one meter, and a second meter was applied to the bypass feeder input at the UPS. Essentially, the two sources of power (inputs) at the UPS were checked for rotation and matched.

I did not see where the neutrals between the two systems were common? If they are not (USP's often having floating output neutral, or where is it bonded?) then I vote with your statement above.

Both transformers have a system bonding jumper, the neutral of each is connected to the Equipment Grounding Conductors. Thus, all neutrals involved are connected together.

The maintenance bypass panel switches neutrals enough to keep neutrals segregated at the panel itself. With all three switches there open, there is no continuity between the neutral bus and the equipment grounding conductor bus.

APC does build some hokey equipment. I would be careful, it will not be *their* fault. The UPS document in the link above shows a unit that is 208 volt Y input with the bypass internal and your one line drawing shows a unit that is 480 volt (delta?) input.

I erred. The proper model (or closer to it) is here.

You are saying the bypass is external?

There's an internal static bypass, and an internal mechanical bypass, and a maintenance bypass panel with bypasses in it... Which one are you referring to?

The bypass document link above does not open.

Which?

Do you get a "System Not Synchronized to Bypass." error?

No. The only error I have seen on occasion is "UPS in Bypass Due to Fault," but it seems to go away on it's own.

Agreed, the inverter (UPS) output is typically sync'ed to the (static) bypass since it has to be able to switch over in <1 cycle.

So, I don't need to worry about this phase shift? The UPS output would always be tied to the bypass input cycle?

120915-2347 EDT

My reading of the original post was that the voltage measurements were made between the output of the internal transformer, before the rectifiers, and with the bypass feeder.

I did not read into the original statement that the UPS output was being compared with the bypass supply, but possibly that is what was implied.

I know I did include that in the measurements I took in the field, but I did not include them. I hadn't thought of checking which "phase group" the output was tied to until just now. When we reenergize the UPSes I will do just that to confirm Zbang and Ron's train of thought.

Often the UPS manufactures use transformers with a phase shift, your bypass transformer needs to match theirs and often is sold as part of the UPS sale.

What is the benefit of a phase shifting transformer?

The salesman mentioned that the transformer inside the UPS is an isolation transformer, which at one point led me to believe that this was the source of our voltage oddities. However, when I think of the term "isolation transformer" I immediately associate an ungrounded neutral conductor to achieve the isolation. Since there is a system bonding jumper that I can see and measure physically (it is roughly a 4 AWG conductor visible in the bottom of the UPS) then this transformer's isolating properties have been defeated by this system bonding jumper, have they not?

I guess the overriding question is, under normal conditions is the load supplied by the batteries or by the source charging the batteries? If the source supplying the batteries is not supplying the load, then there would be no occasion for the two systems to come into direct contact, right? Therefore no concern on my end is warranted...right?

I will also measure voltages at the maintenance bypass panel and see if the phase shift shows there when I get a chance.

I appreciate all the help guys, this is a first for me and it has a 30-year master electrician, the tech, myself and the sales guy confused.

 

[George Stolz]

120916-1030 EDT

zbang:

From the first post:


The one line drawing shows a 480 input to the UPS. From the above quote it says that the internal transformer takes in 480. This seems to imply that the transformer is at the input of the UPS. But I don't really know and your analysis is equally likely.

More detail is needed.

.

Here's a better one line, when taken with the images I attached previously.

View attachment ups oneline.pdf

 

[don_resqcapt19]

Phase rotation was checked at the output of the internal UPS transformer with one meter, and a second meter was applied to the bypass feeder input at the UPS. Essentially, the two sources of power (inputs) at the UPS were checked for rotation and matched. ...

Are the vector diagrams for both transformers the same? That is is one 30? phase shift one direction and the other in the opposite direction?
Also I think you can have the correct rotation and still have the voltages you see if "A" phase is not the same on both transformers. ABC = BCA = CAB a phase rotation meter cannot tell the difference. Try checking the A from one transformer to B or C on the other transformer. One combination should give you "zero" volts.

 

[__dan]

Note 12 in this link: http://www.apcmedia.com/salestools/JSHR-6T89LE_R0_EN.pdf

The UPS has a single input and the internal bypass is bridged internally (and synching the inverter) to that input.

It also shows how APC can get hokey, with the isolating transformer at the input instead of the output. Given the 480 v delta source, I would prefer the UPS input 480 delta, output 480 Y, the external bypass would be 480 delta, and the isolating step down transformer would be load side of the external bypass to 208/120 Y. What I do not like about the one liner above is the inverter pumping directly into the load (which varies and is an unknown). The inverter should pump directly into the primary of the step down transformer (which is a known and can be calibrated to suit the inverter.

The phase shift is a problem that has to be corrected before you become the owner of it.

As suggested earlier, the phase shift in this case could be caused by not matching the phases at the delta inputs to the UPS and external transformer. You could try rotating the phases at one of the 480 v delta inputs to get them to phase match at the secondary side. That is assuming the UPS internal input transformer can match the external transformer at its output.

Otherwise:

Throw it back to the factory as an RFI in writing and decline to close the bypass until ordered to do so in writing by the factory.

You may have a problem with the external transformer/bypass because the internal bypass must be synched to the inverter, the external bypass must be synched to the inverter, and the internal bypass input / external bypass input are not synched to each other.

Without a magic trick, you may be back at square one with an RFI, a change, and a new piece of equipment somewhere.

 

[zbang]

Here's a better one line, when taken with the images I attached previously.

That isn't a one-line of the UPS itself (doesn't show the internal transformer(s)). I don't see it in any of the PDFs I've found on line, so it's probably in the maintenance handbook (printed on the back of an NDA, I'm sure). I'm thinking of a dwg that shows input -> transformer -> battery charger -> DC bus...etc. Have you traced the wiring from output terminals to transformer/inverter to verify the connections?

Of course, the corrective action for "UPS in Bypass Due to Fault" is "Contact APC Customer Support (see rear cover)." Extremely helpful.


FWIW, almost anytime I spec/buy equipment like that these days, I specify in the bid docs that the customer will receive copies of all documentation and software relevant to the operation and maintenance of the product, including maintenance manuals and schematics. If they balk, I tell them that it's just part of the bid process and they can propose something else. Had to go back to the vendor once or twice to explain the meaning of "all".

 

[__dan]

Also check note 13 in: http://www.apcmedia.com/salestools/JSHR-6T89LE_R0_EN.pdf

The factory requires a GEC at the UPS input to the internal transformer secondary (also per 250.30 A 3).

Backwards and redundant at the same time.

 

[__dan]

Also check note 13 in: http://www.apcmedia.com/salestools/JSHR-6T89LE_R0_EN.pdf

The factory requires a GEC at the UPS input to the internal transformer secondary (also per 250.30 A 3).

Backwards and redundant at the same time.

Actually look at this again, the dashed line box that is the UPS. Right at the output is a tiny triangle symbol wth no note. That is the black box, the point where the magic trick happens.

Look at what is supposed to be there. The UPS is essentially two SDS's in series, the input transformer and the inverter output. The inverter output SDS is required to have the system neutral bonding jumper and the GEC atttach there at the same point, like the input transformer. If everything was in the same cabinet and both SDS's attach to the same common grounding busbar, I would say it meets code.

However, being a rack mounted unit, everything could be separate rack boxes, the input transformer a rack box, the rectifier a rack box, the inverter a rack box, the batteries are a rack box ... Then each box would have it own internal grounding busbar and the inverter busbar SBJ, GEC are required but not shown (right where that little triangle thingy is).

Also and relevant, that triangle symbol thingy is some kind of output filter between the inverter and the load. It could be air and a few wires (knowing APC) but it could or should be some line reactors and maybe something else (caps, transient supressors).

The question is if the phase shift is happening in the output filter line reactor configuration. If it is shifting the phase, that is the input to the external bypass, and you may never be able to synch the UPS output to the external transformer output. Not without a change.

If swapping the 480 v input phases cannot make the outputs phase match, I would let the APC explain it in writing. I am still waiting for one of my replies.

 

[T.M.Haja Sahib]

I don't need to worry about this phase shift? The UPS output would always be tied to the bypass input cycle?

Yes. It seems to be so. With a static bypass which makes and breaks extremely fast , the voltage dip and over current would be insignificant. Similar phenomenon routinely happens, for example in a AC-DC rectifier with SCR's. During its operation, any two SCR's perodically shorts out corresponding phases for a fraction of second without any detrimental effects.

 

[Electric-Light]

FWIW, almost anytime I spec/buy equipment like that these days, I specify in the bid docs that the customer will receive copies of all documentation and software relevant to the operation and maintenance of the product, including maintenance manuals and schematics. If they balk, I tell them that it's just part of the bid process and they can propose something else. Had to go back to the vendor once or twice to explain the meaning of "all".

That's unreasonable. Computer manufacturers provide fairly extensive service manuals for customers with onsite techs, such as school districts and universities.

This doesn't extend to component level schematics or firmware source codes. Such things are NOT considered RELEVANT to operation and maintenance by customer. You'll get told the same thing if you were to demand schematic or internal algorithm for motor drive that's part of a manufacturing line you're bidding on.

Those things are considered confidential internal use only. When it comes to UPSs APC is especially uptight in regard to releasing technical information. They're modular and only thing they consider to be "maintenance" are battery and modules they mark as user replaceable. All aspects of operations are controlled by software (from internal voltmeter calibration to battery voltage to serial number).

Can't say I blame them either. Who wants China copying everything?

Many UPSs only use line as a frequency synchronization reference. These large UPSs aren't like your average home UPS with a switch over. "xx cycle switchover" is for off-line UPSs.

The power simply needs to switch over to secondary source before the battery is depleted.

The inverter is always regenerating power, but when power source is live, it gets frequency timing from mains to ensure mains dependent RTC (like alarm clocks) remain accurate. When it goes on generator that swings wildly in frequency, the UPS sometimes switch over to internal reference so the output is ~60.0Hz, but in this mode, RTC won't keep quite that well.


APCs in particular use "delta conversion" that makes thing even mre complicated. It's like 40+ page of reading just on that.
http://www.apc.com/prod_docs/results.cfm?DocType=App Note&Query_Type=3&Value=185

 

[brian john]

Yes. It seems to be so. With a static bypass which makes and breaks extremely fast , the voltage dip and over current would be insignificant. Similar phenomenon routinely happens, for example in a AC-DC rectifier with SCR's. During its operation, any two SCR's perodically shorts out corresponding phases for a fraction of second without any detrimental effects.

The UPS tracks bypass voltage and would always be in alarm, BYPASS OUT OF SYNC and would not go to Bypass when needed.. Additionally if this is aprt of a UPS external bypass KABOOM.

 

[zbang]

That's unreasonable. Computer manufacturers provide fairly extensive service manuals for customers with onsite techs, such as school districts and universities.

This doesn't extend to component level schematics or firmware source codes. Such things are NOT considered RELEVANT to operation and maintenance by customer. You'll get told the same thing if you were to demand schematic or internal algorithm for motor drive that's part of a manufacturing line you're bidding on.

It's very reasonable as your example demonstrates, and they are very relevant. Perhaps our definitions of "customer" differ.

I did not say that "relevant to the operation and maintenance" includes all schematics or source code, although it might include some. I want the same info and software that the factory tech's have, and have sometimes attended to the factory class to get it. I want to know how to replace that power module or download an event log, not how to call someone else to do it. I might call, anyway, but that doesn't help understand anything.

For example, I have all the maintenance manuals for my Thinkpad laptops. They don't contain source code but do list the error numbers and how to replace all the components. That's useful. I also have the manuals for a few mid-size phone switches, voice-mail systems, older UPSs, etc.

 

[Electric-Light]

It's very reasonable as your example demonstrates, and they are very relevant. Perhaps our definitions of "customer" differ.

I meant the end-use facilities and its support personnel. What did you mean?

I did not say that "relevant to the operation and maintenance" includes all schematics or source code

You were saying about going back to vendors over the meaning of "all". I included a link to apcmedia.com and though not model specific, they include general block diagram. As far as I know, that's all APC will give you. Vendor will tell you that you've got ALL you need for maintenance and use. The dissent is over what's relevant.

although it might include some. I want the same info and software that the factory tech's have,

Laptop manufacturers offer software to change board serial # and synchronize them to recovery partition to system. This is done to prevent user from moving HDD to another computer and installing Windows on another PC. Such software is available but usually only to factory certified techs with business registration.

APC battery and modular compartment replacements are covered in user manual. They simply consider everything else non user serviceable. They're most secretive when it comes to service information. You could ask apc-forums.com which is a factory sponsored discussion board, but don't count on much.

APC is just very secretive on releasing technical information.





and have sometimes attended to the factory class to get it. I want to know how to replace that power module or download an event log, not how to call someone else to do it. I might call, anyway, but that doesn't help understand anything.

For example, I have all the maintenance manuals for my Thinkpad laptops. They don't contain source code but do list the error numbers and how to replace all the components. That's useful. I also have the manuals for a few mid-size phone switches, voice-mail systems, older UPSs, etc.[/QUOTE]

 

[T.M.Haja Sahib]

The UPS tracks bypass voltage and would always be in alarm, BYPASS OUT OF SYNC and would not go to Bypass when needed.. Additionally if this is aprt of a UPS external bypass KABOOM.

I do not know whether the OP observed any BYPASS OUT OF SYNC alarm signal in the UPS and so got the confirmation that his observation of phase shift is indeed a serious condition.

 

[George Stolz]

I fired it up yesterday and measured again. 2.0V between bypass input and UPS output, so the output is tied to the bypass. I feel better about it.

Thanks all for the replies! :thumbsup:

 

[dave__asdf]

ups help way late

ups help way late

oops, i guess i'm too late for this one but i just figured i'd throw my thoughts in really quick.

the inverter always syncs to the bypass. the input phase rotation usually doesn't matter, it will simply give you an message. depending on where you take voltage inside a UPS can have strange effects on your readings. if you search for"20 year old Powerware UPS with no output filter" on youtube you will see what i mean. they couldn't figure out why they got 120v on all three phases to ground as well as 120v phase to phase on all three. after you put a scope on it you see it's because it had blown the fuses to the output filter. at teh output everything read fine, until you use the right meter. glad it's up and running fine.

 

[ATSman]

Consider the following (highly simplified) one line. There is a transformer inside the UPS, it takes in 480 and outputs 208/120V. There is a second feeder, by way of a second transformer outside the room, that serves the bypass inputs of the UPS.

View attachment 7413

I will call the two feeders 1 and 2. Feeder 1 is created inside the UPS, and Feeder 2 is from the outside transformer to the bypass of the UPS. All feeders are from solidly grounded wye systems.

Would these voltages be normal, or would you be concerned?

Feeder 1:
A - N = 120V
B - N = 120V
C - N = 120V
A - B = 208V
B - C = 208V
C - A = 208V

Feeder 2:
A - N = 120V
B - N = 120V
C - N = 120V
A - B = 208V
B - C = 208V
C - A = 208V

Then we measured between sources: "f1A" below means "Feeder 1, Phase A".

f1A - f2A = 120V
f1A - f2B = 120V
f1A - f2C = 240V
f1B - f1A = 120V
f1B - f2B = 240V
f1B - f2C = 120V
f1C - f2A = 240V
f1C - f2B = 120V
f1C - f2C = 120V

Now, these readings were unexpected to us, and we all had varying theories on why we were seeing what we saw.

We expected to see a zero voltage between phases, indicating that the two phases were the same phase, even if f1A were zero volts to f2C - we expected to find three zeroes in there somewhere.

My feeling is that, despite the two transformers being derived from the same 480V source, and having interconnections via grounding conductors, that they were two independent sources and that readings between the sources were irrelevant and meaningless.

Another electrician felt that there had to be a winding open somewhere, a motor load connected inside the UPS somehow, causing an oddly high voltage between phases of the two systems.

In theory, only one source will serve the load panel at a time, so we decided to roll with what we had.

Thoughts?

Maybe I am missing something but it is my understanding that a UPS "manufactures" it's output sine wave thorough PWM. It monitors the input voltage (phase angle, rotation) and syncs the output voltage to match it. The output voltage is changed depending on the trigger settings in the UPS. I think the error in your thinking is that you are expecting to see voltage values between 2 hardwired transformers when in fact you are comparing a hardwired 3 phase source to a manufactured 3 phase source. Maybe someone with more smarts than me can expound on this