Nut upsuser - Nov 2024 - Interpreting data from 220V input APC UPS

John Ackermann N8UR via Nut-upsuser
<nut-upsuser at alioth-lists.debian.net> writes:
> I am monitoring via the SNMP driver an APC SmartUPS that has split
> phase (2L + neutral)  240V input and 120/120 volt outputs.  The data
> for the voltages is not what I'm expecting, and I am wondering how I
> should interpret it.
Wow, that sounds kind of industrial.  Model?

Is the input an L14-20P?  Or equivalent non-twistlock?
> Here is an example from upsc:
>
> input.L1-L2.voltage: 121
> input.L2-L3.voltage: 120
> input.voltage: 121.20
> output.current: 5.90
> output.L1-L2.voltage: 119
> output.L1.current: 5.90
> output.L2-L3.voltage: 119
> output.L2.current: 1
>
> The "input.voltage" value doesn't reflect the 240 volts that
is
> actually being applied.  To get that, would I combine the L1-L2 and
> L2-L3 voltages?
I would address figuring this out as two steps.  One is to see what the
device actually sends and what makes sense there.  And then to see if
NUT is mapping or making synthetic.

I find it odd for the neutral to be labeled L2, as it seems to be.  I
would think there would be L1 and L2, both reported as phase-to-neutral.
You might see if there is confusion.

odd for output.L2.current, vs L3, given that voltage seems to be L1/L3
vs L2.

Probably the first voltage is mapped to just 'voltage' and probably it
would be better to use the L1/L3 voltage.  Similar for output.voltage.

output.current should probably be average if output.voltage is 240ish.
Or better yet 
> Note that the "output.current" value matches the L1 output
current,
> ignoring the L2 current.  Perhaps that is a clue that the overall
> "input" and "output" values are not meaningful in this
case?
yes, and they are perhaps mapped from something else?
> How should I interpret these values to know the actual input and
> output voltages and currents?
You're going to need to put a meter on it and compare to the reports, I
suspect.

First, see if you can get specs, and look at the snmp mib directly and
see what you can figure out.  Turn up debugging/verbose in nut.

Sounds pretty normal from what I see. It appears that this is *NOT* a true 240v
UPS, but rather one that provides two legs of 120v output from two legs of 120
input, and what is seen as "L2" is actually neutral, and the line
names are reporting incorrectly. With that in mind, each leg in should be 120v,
as well as the outputs, which is exactly what you are seeing.(The fact tht L1 to
L3 is 240v really isn't relevant).

- Tim

On November 11, 2024 6:56:51 PM EST, Greg Troxel via Nut-upsuser <nut-upsuser
at alioth-lists.debian.net> wrote:
>John Ackermann N8UR via Nut-upsuser
><nut-upsuser at alioth-lists.debian.net> writes:
>
>> I am monitoring via the SNMP driver an APC SmartUPS that has split
>> phase (2L + neutral)  240V input and 120/120 volt outputs.  The data
>> for the voltages is not what I'm expecting, and I am wondering how
I
>> should interpret it.
>
>Wow, that sounds kind of industrial.  Model?
>
>Is the input an L14-20P?  Or equivalent non-twistlock?
>
>> Here is an example from upsc:
>>
>> input.L1-L2.voltage: 121
>> input.L2-L3.voltage: 120
>> input.voltage: 121.20
>> output.current: 5.90
>> output.L1-L2.voltage: 119
>> output.L1.current: 5.90
>> output.L2-L3.voltage: 119
>> output.L2.current: 1
>>
>> The "input.voltage" value doesn't reflect the 240 volts
that is
>> actually being applied.  To get that, would I combine the L1-L2 and
>> L2-L3 voltages?
>
>I would address figuring this out as two steps.  One is to see what the
>device actually sends and what makes sense there.  And then to see if
>NUT is mapping or making synthetic.
>
>I find it odd for the neutral to be labeled L2, as it seems to be.  I
>would think there would be L1 and L2, both reported as phase-to-neutral.
>You might see if there is confusion.
>
>odd for output.L2.current, vs L3, given that voltage seems to be L1/L3
>vs L2.
>
>Probably the first voltage is mapped to just 'voltage' and probably
it
>would be better to use the L1/L3 voltage.  Similar for output.voltage.
>
>output.current should probably be average if output.voltage is 240ish.
>Or better yet 
>
>> Note that the "output.current" value matches the L1 output
current,
>> ignoring the L2 current.  Perhaps that is a clue that the overall
>> "input" and "output" values are not meaningful in
this case?
>
>yes, and they are perhaps mapped from something else?
>
>> How should I interpret these values to know the actual input and
>> output voltages and currents?
>
>You're going to need to put a meter on it and compare to the reports, I
>suspect.
>
>First, see if you can get specs, and look at the snmp mib directly and
>see what you can figure out.  Turn up debugging/verbose in nut.
>
>_______________________________________________
>Nut-upsuser mailing list
>Nut-upsuser at alioth-lists.debian.net
>https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://alioth-lists.debian.net/pipermail/nut-upsuser/attachments/20241111/abc9bde3/attachment.htm>

Hi Greg --

My, I run into you in the strangest places. :-)

This is an APC RT5000-RM-XL.  The input and output are L14-30P and R, so 
L1/L2/N/G.  The "phase" value reports as "3" so there is
some
misinterpretation going on.  I've started digging into the APC mib tree 
but they don't give a lot of helpful description for the values.

Let me do some more reading about the mibs, and then do some direct snmp 
fetches to see what the raw values look like.  I'll report back.

Thanks!
John

(This is a new (to me) unit that runs pretty much my whole basement 
lab/workshop.  It's replacing a much older unit that barely had any 
comms capability, so monitoring was by LED.  I've already learned that I 
don't have the loads balanced very well across the two output phases, so 
I need to move at least one circuit to the other phase.  But I'm running 
way below total capacity, so I can live with it that way for a little 
while.)
----

On 11/11/24 18:56, Greg Troxel via Nut-upsuser wrote:
> John Ackermann N8UR via Nut-upsuser
> <nut-upsuser at alioth-lists.debian.net> writes:
> 
>> I am monitoring via the SNMP driver an APC SmartUPS that has split
>> phase (2L + neutral)  240V input and 120/120 volt outputs.  The data
>> for the voltages is not what I'm expecting, and I am wondering how
I
>> should interpret it.
> 
> Wow, that sounds kind of industrial.  Model?
> 
> Is the input an L14-20P?  Or equivalent non-twistlock?
> 
>> Here is an example from upsc:
>>
>> input.L1-L2.voltage: 121
>> input.L2-L3.voltage: 120
>> input.voltage: 121.20
>> output.current: 5.90
>> output.L1-L2.voltage: 119
>> output.L1.current: 5.90
>> output.L2-L3.voltage: 119
>> output.L2.current: 1
>>
>> The "input.voltage" value doesn't reflect the 240 volts
that is
>> actually being applied.  To get that, would I combine the L1-L2 and
>> L2-L3 voltages?
> 
> I would address figuring this out as two steps.  One is to see what the
> device actually sends and what makes sense there.  And then to see if
> NUT is mapping or making synthetic.
> 
> I find it odd for the neutral to be labeled L2, as it seems to be.  I
> would think there would be L1 and L2, both reported as phase-to-neutral.
> You might see if there is confusion.
> 
> odd for output.L2.current, vs L3, given that voltage seems to be L1/L3
> vs L2.
> 
> Probably the first voltage is mapped to just 'voltage' and probably
it
> would be better to use the L1/L3 voltage.  Similar for output.voltage.
> 
> output.current should probably be average if output.voltage is 240ish.
> Or better yet
> 
>> Note that the "output.current" value matches the L1 output
current,
>> ignoring the L2 current.  Perhaps that is a clue that the overall
>> "input" and "output" values are not meaningful in
this case?
> 
> yes, and they are perhaps mapped from something else?
> 
>> How should I interpret these values to know the actual input and
>> output voltages and currents?
> 
> You're going to need to put a meter on it and compare to the reports, I
> suspect.
> 
> First, see if you can get specs, and look at the snmp mib directly and
> see what you can figure out.  Turn up debugging/verbose in nut.
> 
> _______________________________________________
> Nut-upsuser mailing list
> Nut-upsuser at alioth-lists.debian.net
> https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser

Following up on this, I've figured out what's going on.  What I'm
not
sure of is how best to address the situation.  I'd like to avoid going 
down to the MIB level because I just don't understand it.

"Split phase" output devices seem to fall into an undefined area. 
They're not three phase, but require additional MIB values so APC uses 
their 3-phase MIB structure for them, and NUT plays along.

I'm no electrician, but I think the proper terminology for a 
"split-phase" device is to consider each of the input and output
phases
reported by APC as a "leg".  If it were possible to flag a UPS in nut
as
"split-phase" it could do some conversions.  Is that feasible?  Would
it
make more sense to do it in upsc?  Or should I post-process in my own 
tooling?

The mapping I came up with follows.  Any suggestions on how/where best 
to deal with this?

John
===# Substitute "Leg1" for "L1-L2" or "L1"
input.L1-L2.voltage=121
input.L1-L2.voltage.maximum=123
input.L1-L2.voltage.minimum=118
output.L1-L2.voltage=119
output.L1.current=6.10
output.L1.current.maximum=13.20
output.L1.current.minimum=5.50
output.L1.power.maximum.percent=63
output.L1.power.minimum.percent=27
output.L1.power.percent=30

# Substitute "Leg2" for "L2-L3" or "L2"
input.L2-L3.voltage=120
input.L2-L3.voltage.maximum=122
input.L2-L3.voltage.minimum=118
output.L2-L3.voltage=119
output.L2.current=2.30
output.L2.current.maximum=6.30
output.L2.current.minimum=0.90
output.L2.power.maximum.percent=30
output.L2.power.minimum.percent=4
output.L2.power.percent=11

# these show L-N, apparently
# for either leg.  There is also
# a pair of L-L transfer voltages
# that I don't see an analog for
# in the uspc output.
input.transfer.high=132
input.transfer.low=98
input.voltage=121.30

# these show Leg 1; they
# should be the sum of the two legs
input.voltage.maximum=123
input.voltage.minimum=118

# These only show Leg 1;
# they should be the sum of the two legs
output.voltage=119.70
output.voltage.nominal=0
output.current=6
# This reflects only Leg1.  There
# should be separate "ups.leg1.load"
# and "ups.leg2.load values, as well
# the calculated total.
ups.load=32.10

# Show these as "split" rather than 3
# But does it have to be an integer???
input.phases=3
output.phases=3