208 volt UPS

[[email protected]]

A UPS is rated to normally operate on 208V, but can be configured to operate
also on 200V, 220V, 230V, and 240V. It comes with NEMA 6-30P plug and has
NEMA 6-20R x2 and NEMA 6-30R x2 outlets. These are standard for USA 208/240
volt circuits that have no neutral connection.

What system type would you expect such a UPS to output when it is configured
for each of the voltages (200, 208, 220, 230, 240)? Would you expect each
hot wire, when measured relative to ground (not neutral since there is no
neutral) to be 50% of the L-L voltage and a phase angle of 180 degrees, or
57.7% of the L-L voltage and a phase angle of 120 degrees? Would you expect
the phase angle (and hence L-L : L-G ratio) to be different with different
voltage settings?

If it ONLY had voltage choices of 208 and 240 I could believe it might do
this by keeping the L-G voltage at 120 and varying the phase angle. But
with the other voltages being an option, I'm not so sure. While any of
those voltage could be derived from selected phase angles of 120 volts, I
would expect that to actually be an odd thing to do.

I'm curious if such a unit could be run on 208V input (two phases in a
208Y/120 system) and 240V output reasonably (it can apparently be set to
do that so I would think the vetted the design for it). I'm just not sure
what kind of system I'd be getting (120/240 at 180 degrees or 139/240 at
120 degrees). Given that it has no neutral, clearly they are not expecting
any L-N loads. But some devices could, in theory, monitor a L-G voltage
and treat 139 volts as overvoltage.

I am currently looking at the APC model SURT5000RMXLT.

 

[Charles Perry]

A UPS is rated to normally operate on 208V, but can be configured to
operate
also on 200V, 220V, 230V, and 240V. It comes with NEMA 6-30P plug and has
NEMA 6-20R x2 and NEMA 6-30R x2 outlets. These are standard for USA
208/240
volt circuits that have no neutral connection.

What system type would you expect such a UPS to output when it is
configured
for each of the voltages (200, 208, 220, 230, 240)? Would you expect each
hot wire, when measured relative to ground (not neutral since there is no
neutral) to be 50% of the L-L voltage and a phase angle of 180 degrees, or
57.7% of the L-L voltage and a phase angle of 120 degrees? Would you
expect
the phase angle (and hence L-L : L-G ratio) to be different with different
voltage settings?

Given the info provided, and our experience testing UPS, I think they UPS
will produce neither the 120/208 or 240 split phase. It will produce the
Line to Line voltage that is selected. The relationship to ground will NOT
be part of that output. There is a ground, and it will be somewhere in
relation to the output voltage but there will be NO attempt by the UPS to
maintain a certain phase angle relationship. The output inverter will NOT
even use the ground. So, the voltage measured from each Line to ground
could be anywhere from full rated voltage to zero, and may not be fixed.

Charles Perry P.E.

 

[[email protected]]

|
| |>A UPS is rated to normally operate on 208V, but can be configured to
|>operate
|> also on 200V, 220V, 230V, and 240V. It comes with NEMA 6-30P plug and has
|> NEMA 6-20R x2 and NEMA 6-30R x2 outlets. These are standard for USA
|> 208/240
|> volt circuits that have no neutral connection.
|>
|> What system type would you expect such a UPS to output when it is
|> configured
|> for each of the voltages (200, 208, 220, 230, 240)? Would you expect each
|> hot wire, when measured relative to ground (not neutral since there is no
|> neutral) to be 50% of the L-L voltage and a phase angle of 180 degrees, or
|> 57.7% of the L-L voltage and a phase angle of 120 degrees? Would you
|> expect
|> the phase angle (and hence L-L : L-G ratio) to be different with different
|> voltage settings?
|>
|
| Given the info provided, and our experience testing UPS, I think they UPS
| will produce neither the 120/208 or 240 split phase. It will produce the
| Line to Line voltage that is selected. The relationship to ground will NOT
| be part of that output. There is a ground, and it will be somewhere in
| relation to the output voltage but there will be NO attempt by the UPS to
| maintain a certain phase angle relationship. The output inverter will NOT
| even use the ground. So, the voltage measured from each Line to ground
| could be anywhere from full rated voltage to zero, and may not be fixed.

Hmmm. So if multiple UPSes connect multiple pieces of equipment, and those
multiple pieces of equipment are interconnected to each other by ground and
signals (a room full of computers), this could produce strange voltages in
the network? ... like as much as 416 volts? ... or maybe more?

I think I'd rather have a fully grounded system, but with a 30ma GFI to
block attempts to use the grounding wire as a current conductor.

 

[[email protected]]

On 8 Dec 2009 12:08:28 GMT [email protected] wrote:
| |
| | | |>A UPS is rated to normally operate on 208V, but can be configured to
| |>operate
| |> also on 200V, 220V, 230V, and 240V. It comes with NEMA 6-30P plug and has
| |> NEMA 6-20R x2 and NEMA 6-30R x2 outlets. These are standard for USA
| |> 208/240
| |> volt circuits that have no neutral connection.
| |>
| |> What system type would you expect such a UPS to output when it is
| |> configured
| |> for each of the voltages (200, 208, 220, 230, 240)? Would you expect each
| |> hot wire, when measured relative to ground (not neutral since there is no
| |> neutral) to be 50% of the L-L voltage and a phase angle of 180 degrees, or
| |> 57.7% of the L-L voltage and a phase angle of 120 degrees? Would you
| |> expect
| |> the phase angle (and hence L-L : L-G ratio) to be different with different
| |> voltage settings?
| |>
| |
| | Given the info provided, and our experience testing UPS, I think they UPS
| | will produce neither the 120/208 or 240 split phase. It will produce the
| | Line to Line voltage that is selected. The relationship to ground will NOT
| | be part of that output. There is a ground, and it will be somewhere in
| | relation to the output voltage but there will be NO attempt by the UPS to
| | maintain a certain phase angle relationship. The output inverter will NOT
| | even use the ground. So, the voltage measured from each Line to ground
| | could be anywhere from full rated voltage to zero, and may not be fixed.
|
| Hmmm. So if multiple UPSes connect multiple pieces of equipment, and those
| multiple pieces of equipment are interconnected to each other by ground and
| signals (a room full of computers), this could produce strange voltages in
| the network? ... like as much as 416 volts? ... or maybe more?
|
| I think I'd rather have a fully grounded system, but with a 30ma GFI to
| block attempts to use the grounding wire as a current conductor.

I've started getting some answers from their tech support. After a couple
back and forth they finally asked an engineer, and passed on the statement
that it is possible for one leg to be 120V and the other to be 88V relative
to ground. Now that is odd.

If you were DESIGNING a UPS, part of a series of different models for many
different needs, with THIS model intended to serve cases where power comes
in on a NEMA-6 (at whatever amperage), at either 208V in cases where three
phase is the mains supply, or 240V in cases where single phase is the mains
supply, and it needs to be double conversion continuous on-line type, how
would you design it?

 

[Charles Perry]

On 8 Dec 2009 12:08:28 GMT [email protected] wrote:
| On Tue, 8 Dec 2009 06:33:54 -0500 Charles Perry
<[email protected]> wrote:

I've started getting some answers from their tech support. After a couple
back and forth they finally asked an engineer, and passed on the statement
that it is possible for one leg to be 120V and the other to be 88V
relative
to ground. Now that is odd.

If you were DESIGNING a UPS, part of a series of different models for many
different needs, with THIS model intended to serve cases where power comes
in on a NEMA-6 (at whatever amperage), at either 208V in cases where three
phase is the mains supply, or 240V in cases where single phase is the
mains
supply, and it needs to be double conversion continuous on-line type, how
would you design it?

Why do I care about the relationship between the ground and the two "hot"
conductors? As long at the line to ground voltage never exceeds the line to
line voltage why do I care? You should never connect a load from line to
ground. On this type of UPS the load has to be connected line to line. As
long as the line to line voltage stays in spec, what does it really matter?
Without an output isolation transformer the UPS does NOT qualify as a
separately derived source, so you would never bond the ground to either
Line.

Charles Perry P.E.

 

[[email protected]]

|
| |> On 8 Dec 2009 12:08:28 GMT [email protected] wrote:
|> | On Tue, 8 Dec 2009 06:33:54 -0500 Charles Perry
| <snip>
|> I've started getting some answers from their tech support. After a couple
|> back and forth they finally asked an engineer, and passed on the statement
|> that it is possible for one leg to be 120V and the other to be 88V
|> relative
|> to ground. Now that is odd.
|>
|> If you were DESIGNING a UPS, part of a series of different models for many
|> different needs, with THIS model intended to serve cases where power comes
|> in on a NEMA-6 (at whatever amperage), at either 208V in cases where three
|> phase is the mains supply, or 240V in cases where single phase is the
|> mains
|> supply, and it needs to be double conversion continuous on-line type, how
|> would you design it?
|>
|
| Why do I care about the relationship between the ground and the two "hot"
| conductors? As long at the line to ground voltage never exceeds the line to
| line voltage why do I care? You should never connect a load from line to
| ground. On this type of UPS the load has to be connected line to line. As
| long as the line to line voltage stays in spec, what does it really matter?
| Without an output isolation transformer the UPS does NOT qualify as a
| separately derived source, so you would never bond the ground to either
| Line.

It isn't the actual voltage across a line to ground connection that is the
issue. The issue is the accumulated phantom voltage between equipment that
is interconnected with metallic data lines. In an case with 120/240 mains
and equipment connected at opposing poles, the worst case is seeing 240 volts
across things. My concern was to make sure I wasn't getting an output that
was 208 volts L-G, which could give me a phantom 416 volts across connections
between equipment, and possibly outside the normal expectations of insulation
in various components.

I do not see how an inverter is not a separately derived system.

If the output is a pair of 120 volt inverters syncronized at either 180
degrees or 120 degrees apart (configurable) to get 240 volts or 208 volts,
and these two inverters ARE connected to ground (not neutral as that is
not even present), and the load on the inverter is connected line to line,
how is this creating a situation for operating current on the ground wire
at either the input or output?

 

[Charles Perry]

I do not see how an inverter is not a separately derived system.

The code says it is not. There is a bypass switch that can connect the
output directly to the input. If you want to qualify as a separately
derived source, you have to have a transformer on the output.

Even in normal operating mode, it does not qualify.

If the output is a pair of 120 volt inverters syncronized at either 180
degrees or 120 degrees apart (configurable) to get 240 volts or 208 volts,
and these two inverters ARE connected to ground (not neutral as that is
not even present), and the load on the inverter is connected line to line,
how is this creating a situation for operating current on the ground wire
at either the input or output?

It is not a pair of 120V inverters. It is a single 240V inverter that can
be programmed to output 208 volts instead (pulse width modulated and filter
usually).

Charles Perry P.E.

 

[[email protected]]

|
| |> On Sun, 13 Dec 2009 18:44:57 -0500 Charles Perry
| <snip>
|
|> I do not see how an inverter is not a separately derived system.
|
| The code says it is not. There is a bypass switch that can connect the
| output directly to the input. If you want to qualify as a separately
| derived source, you have to have a transformer on the output.

Where in the code is that?

Not all UPSes have a bypass switch.


| Even in normal operating mode, it does not qualify.

I never saw that in the code.


|> If the output is a pair of 120 volt inverters syncronized at either 180
|> degrees or 120 degrees apart (configurable) to get 240 volts or 208 volts,
|> and these two inverters ARE connected to ground (not neutral as that is
|> not even present), and the load on the inverter is connected line to line,
|> how is this creating a situation for operating current on the ground wire
|> at either the input or output?
|
| It is not a pair of 120V inverters. It is a single 240V inverter that can
| be programmed to output 208 volts instead (pulse width modulated and filter
| usually).

Then you can have problems with this if the UPS has a bypass switch, since
it will change ground reference when the switching happens.

Suppose you do have an isolation transformer on the output. Would you design
it so that the secondary winding has a ground tap in the middle, or not?

 

[Charles Perry]

On Mon, 14 Dec 2009 21:09:48 -0500 Charles Perry
|
| |> On Sun, 13 Dec 2009 18:44:57 -0500 Charles Perry
| <snip>
|
|> I do not see how an inverter is not a separately derived system.
|
| The code says it is not. There is a bypass switch that can connect the
| output directly to the input. If you want to qualify as a separately
| derived source, you have to have a transformer on the output.

Where in the code is that?

Not all UPSes have a bypass switch.


| Even in normal operating mode, it does not qualify.

I never saw that in the code.

The UPS passes ground straight through. You cannot tie a "line" to ground,
it will create a fault. Therefore it is not separately derived. You can
try, but I would video tape it as it will be fun to watch after.

|> If the output is a pair of 120 volt inverters syncronized at either 180
|> degrees or 120 degrees apart (configurable) to get 240 volts or 208
volts,
|> and these two inverters ARE connected to ground (not neutral as that is
|> not even present), and the load on the inverter is connected line to
line,
|> how is this creating a situation for operating current on the ground
wire
|> at either the input or output?
|
| It is not a pair of 120V inverters. It is a single 240V inverter that
can
| be programmed to output 208 volts instead (pulse width modulated and
filter
| usually).

Then you can have problems with this if the UPS has a bypass switch, since
it will change ground reference when the switching happens.

The ground goes straight through the UPS.

Suppose you do have an isolation transformer on the output. Would you
design
it so that the secondary winding has a ground tap in the middle, or not?

No. I don't connect a load to ground so if I had a tranformer on the
output, I would tie one line to ground. This keeps any leakage current from
my load on the output side of the transformer.

--
-----------------------------------------------------------------------------
| Phil Howard KA9WGN | http://linuxhomepage.com/
http://ham.org/ |
| (first name) at ipal.net | http://phil.ipal.org/
http://ka9wgn.ham.org/ |
-----------------------------------------------------------------------------

I think part of your problem is you keep thinking of UPS in terms of
transformers (phase angles between 120 l-g connections to get 208 or 240).
Power electronic devices to not have to work like transformers. Once you
have a DC bus you can do any thing you want.

Charles Perry P.E.

 

[[email protected]]

|
| |> On Mon, 14 Dec 2009 21:09:48 -0500 Charles Perry
|> |
|> | |> |> On Sun, 13 Dec 2009 18:44:57 -0500 Charles Perry
|> | <snip>
|> |
|> |> I do not see how an inverter is not a separately derived system.
|> |
|> | The code says it is not. There is a bypass switch that can connect the
|> | output directly to the input. If you want to qualify as a separately
|> | derived source, you have to have a transformer on the output.
|>
|> Where in the code is that?
|>
|> Not all UPSes have a bypass switch.
|>
|>
|> | Even in normal operating mode, it does not qualify.
|>
|> I never saw that in the code.
|>
| The UPS passes ground straight through. You cannot tie a "line" to ground,
| it will create a fault. Therefore it is not separately derived. You can
| try, but I would video tape it as it will be fun to watch after.

There is no more of a line tied to ground in the UPS than there is in a
transformer producing the same system.


|> |> If the output is a pair of 120 volt inverters syncronized at either 180
|> |> degrees or 120 degrees apart (configurable) to get 240 volts or 208
|> volts,
|> |> and these two inverters ARE connected to ground (not neutral as that is
|> |> not even present), and the load on the inverter is connected line to
|> line,
|> |> how is this creating a situation for operating current on the ground
|> wire
|> |> at either the input or output?
|> |
|> | It is not a pair of 120V inverters. It is a single 240V inverter that
|> can
|> | be programmed to output 208 volts instead (pulse width modulated and
|> filter
|> | usually).
|>
|> Then you can have problems with this if the UPS has a bypass switch, since
|> it will change ground reference when the switching happens.
|
| The ground goes straight through the UPS.

As it should. So?


|> Suppose you do have an isolation transformer on the output. Would you
|> design
|> it so that the secondary winding has a ground tap in the middle, or not?
|
| No. I don't connect a load to ground so if I had a tranformer on the
| output, I would tie one line to ground. This keeps any leakage current from
| my load on the output side of the transformer.

And how is a UPS different?


| I think part of your problem is you keep thinking of UPS in terms of
| transformers (phase angles between 120 l-g connections to get 208 or 240).
| Power electronic devices to not have to work like transformers. Once you
| have a DC bus you can do any thing you want.

I'm talking about the "dual conversion continuous online" type of UPS, the
only type I ever consider using. It takes the AC input and converts it to
DC. That DC is paralleled with the battery with some circuit that manages
the usage in some way depending on how sophisticated that UPS design is.
Then the DC feeds one or more inverters.

So I have a DC bus. I can do anything I want?

A bypass switch can be present, but some models omit this. I did see a
UPS somewhere online once (a rather large one) that had 480V three phase
in and 208Y/120V three phase out. I'd say a bypass on that would not be
a very good idea. And is it a separately derived system or not?

I see nothing in the NEC that says the inverter is not a separately derived
system. A UPS can create a different kind of system than its input. This
is easily done when using separate component. It could be done as a whole
UPS unit if there was a market demand for it (if it is done anywhere, it is
almost certainly a special case). Actually, I did see one once that input
three phase and output only single phase (but the loading was balanced on
the three input phases, presumably with very near unity power factor).

Looking at just single phase to keep it simple:

A UPS system that has two inverters with synchronized waveforms could be
connected as follows:

___Ground
/
*--[Inv1]--*--[Inv2]--*
| |
A B

There is no neutral, in or out. The source system may or may not have a
neutral. If it has one, it is not used and not connected to.

The bypass switch, if present, is an open transition two pole device.

Where is the short circuit fault? I don't see one. Is there any load
current on the ground wire? I don't see where.

You could replace Inv1 and Inv2 with two windings from a single phase
transformer. How is this different? Or are you one of those people
that always leaves transformer secondaries ungrounded?

But, just for fun, I'll add a transformer to the output:

Ground
|
*--[Inv1]--*--[Inv2]--* |
| \ | |
| -------- | ---*
A B |
| | |
*---/\/\/\/\/\/\/\/---* |
*---/\/\/\/\/\/\/\/---* |
=================== |
*--/\/\/\/--*--/\/\/\/--* |
| \ | |
| --------- | --*
| |
A B

That's a common 480/240 to 240/120 transformer wired with the 240 windings
in parallel and the 120 windings in series so it's a 240 to 240 transformer.

In what way are these two systems different while the bypass switch is in
the "not bypassed" position?

Of course if the load needs a neutral or any 120V loading, we can't use this
scheme for that load because we can't have load current on the ground wire,
on either side. But you could supply a neutral to the output system as a
separate wire. That's commonly done on the transformer. I see no reason
it can't be done on the UPS other than the fact that you cannot do a bypass
on such a UPS because there is no neutral coming in.

For 208V output, the UPS can "bend" the inverters and run them at 120 degree
phase angle difference. The inverters will need to handle the effective
power factor they will get from that. Add a third inverter and you can do
three phase delta in, wye-like out that can serve delta loads (a mix of
single phase line to line loads).

 

[[email protected]]

It definitely looks like all the APC branded 208V UPSes use a floating
ungrounded single inverter output. Every model that is designed to also
have 120V output uses a step-down transformer to obtain that. And I would
imagine that's just one ended 120V rather than 120/240 (which means 2 wire
at twice the current instead of 3 wire).

Do their 120V UPSes also work ungrounded? A 120V circuit would have a
wire that is designated to be the grounded conductor, so I would think
they would have to.

 

[[email protected]]

| The UPS passes ground straight through. You cannot tie a "line" to ground,
| it will create a fault. Therefore it is not separately derived. You can
| try, but I would video tape it as it will be fun to watch after.

[...]

| The ground goes straight through the UPS.

[...]

| No. I don't connect a load to ground so if I had a tranformer on the
| output, I would tie one line to ground. This keeps any leakage current from
| my load on the output side of the transformer.

[...]

| I think part of your problem is you keep thinking of UPS in terms of
| transformers (phase angles between 120 l-g connections to get 208 or 240).
| Power electronic devices to not have to work like transformers. Once you
| have a DC bus you can do any thing you want.

A sales engineer suggested the following unit to me:

http://www.stayonline.com/detail.aspx?ID=3546

Maybe you can tell me if this one has a 120 degree "bent" 120/208 output on
either 2W+G or 3W+G, as measured by a voltmeter using the EGC as a reference
(in case there is no output neutral, which I don't need, since I will be
using the 208 volts directly, but want voltage stability relative to ground
as part of that). I can take a 104/208 volt "straight" split single phase
system, too. Or I can take 120/240. It's all going into a computer power
supply that can take 100-240, but I want to limit the current (there will be
a lot of these).