Possible OT: power monitoring

[Ben Scott]

On Mon, Jun 16, 2008 at 4:19 PM, Curtis Sandoval
<curtis.sandoval at gmail.com> wrote:
> TrippLite 900s and a CyberPower 675 ...
[...]
> Is there a way, using these or other UPS units, to constantly monitor
> input line voltage and length of outages?

  Better UPSes can report detailed information like that (sometimes
called "smart" or "intelligent").  Entry-level models ("dumb") often
have either no monitoring at all, or just something that tells the PC
line power has dropped, and maybe, if you're lucky, low-battery.  The
dumb models generally use the moral equivalent of a contact-closure.
In the old days, they'd most often re-purpose the flow control lines
of RS-232 serial ports for this.  These days, they do the same thing,
but hook the USB up to a USB-to-RS232 serial adapter, since almost
nothing has RS-232 ports anymore.  The smart models will have an
actual data communication protocol, usually run over serial or USB.
It often allows for UPS control as well as monitoring.  The protocol
is usually simplistic and cryptic, but I remember you could hook a
VT-220 up to a Ferr-UPS and get a command prompt.  :)

  Based on the info on the manufacturer website, the CyberPower UP625
falls into the "dumb" category.  You won't be getting power quality
stats from it.

  Please clarify "TrippLite 900".  Checking their website, I find
several different products with "900" in their model name.

> ... a few hours of low input voltage (112v-114v) ...

  As far as the power company is concerned, 112 VAC is usually not
considered "low".  Anything within 110 to 120 VAC is usually
acceptable.  Not that that helps you much if your equipment *does*
care...

> Power quality monitoring would be nice, but my searches seemed to
> indicate that quality monitoring is much more expensive ...

  Yup.  "Smart" vs "dumb" is one of the key differentiators between
UPSes in different price bands.

> I could find nothing specifically saying that a UPS or quality monitor would report
> real-time to a computer (specifically SUSE 10.2 or Ubuntu 8.04).

  Like anything else, support and features vary with manufacturer and
model.  Not everything works well with Linux, either.  For some
models, third-party software is the best bet; for others, the OEM
provides good software.

  I've had experience with various models from APC, TrippLite, and
Eaton.  Quality and features vary.  Nothing was astoundingly bad,
nothing blew me away with price/performance.  APC is prolly the
best-known brand, and sometimes that's reflected in higher prices.
(Note that "best known" is neither necessary nor sufficient for "best
quality".)

  I know the APC Smart-UPS line can report information like input
voltage, output voltage, load (as a fraction of capacity), battery
voltage, runtime remaining, and unit temperature.  Their "PowerChute"
software has had various Linux releases of varying functionality and
stability.  (This information does not constitute an endorsement of
APC's products.  (The preceding statement does not constitute a
condemnation of APC's products.))

  TrippLite, Eaton (AKA PowerWare AKA Best Power), and Minute-Man also
have "smart" UPS products, and claim various levels of Linux support.

  Popular third-party UPS software for Linux includes "apcupsd" and
"NUT" ("Network UPS Tools").  Google the project names to learn more.
In particular, both projects list hardware known to work with their
stuff.

  YMMV.  Caveat emptor.  UPS manufacturers love to change things
without telling anyone, often selling a very different product under a
same or similar name.  Their software is sometimes astoundingly badly
written.  They sometimes regard their communication protocols as
proprietary trade secrets.  Etc.

> I have been looking at getting some new or larger UPS units, and I've
> noticed there is a nonlinear relationship between price and capacity ...

  Look carefully at the specs.  You may be comparing apples to
oranges.  Some parameters which affect price include:

- Load capacity (determines connected equipment)
- Battery capacity (determines maximum runtime)
- Expandability (external battery packs)
- Warranty and equipment coverage
- Monitoring capabilities
- Output waveform (square wave approximation vs "true" sine wave)
- Transfer time
- Operating principle (stand-by vs line interactive vs double
conversion vs ferro-resonant vs ...)
- Voltage regulation and surge suppression features (these are not
inherent in all UPS designs)

  There are non-technical factors, too.  If you're asking for a bigger
UPS, manufactures assume you want something better than the
bargain-basement crap you usually find at BestBuy, Circuit City, et.
al.  So the bar for minimum quality is implicitly raised.  At the same
time, if you're asking for a bigger UPS, manufactures assume you can
afford to pay more, and thus charge more simply because they can.

> makes me wonder why a person would not buy two or more smaller units and
> daisy-chain them to get higher capacity and redundancy.

  Daisy-chaining UPSes is sub-optimal at best.  The manfuactuers warn
you never to do it, but maybe they're just trying to sell more UPSes.
But there are technical reasons, too.  For one, the square-wave
approximation output by many cheaper UPSes will generally cause any
UPS downstream from it to trip over to battery, too.  So as soon as
wall power is lost, everything trips to battery, and all but the last
UPS is just generating heat.  More sophisticated UPSes can output
better power, but they also have higher draws for
supporting/conditioning, so you're adding inefficiencies.  And I've
heard horror stories of UPSes self-damaging in such configurations.
(Envision a UPS tripping on and off battery constantly...)

  I'm not saying daisy-chaining UPSes is the ultimate evil, but it
still seems like something to avoid.

-- Ben