Observations on a UPS - follow up to a previous post

[Michael A. Terrell]

But that's my point. You can't have a 'true' sine-based RMS figure for power
consumers that draw a pulsed current. Just because it is a sine wave that's
*available*, it doesn't mean that the load will draw anything like a
sinusoidal current, from it. The only items that will are those that are
totally 'passive' in nature, such as light bulbs. Even power supplies that
are transformer based, are likely to draw current in pulses from the
available sinusoidal supply, and switch mode power supplies, on which most
modern consumer electronics are based, most certainly won't draw a
sinusoidal current from the supply.

Arfa

ICs to do the true RMS conversion are dirt cheap these days.


<http://www.analog.com/library/analogDialogue/archives/34-01/rmsarticle/index.html>

<http://www.analog.com/en/subCat/0,2879,773%5F866%5F0%5F%5F0%5F,00.html>
--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Arny Krueger]

I have a "Kill-A-Watt" and it appears to display true RMS
for voltage and current readings. The voltage reading for
a modified sine wave from a standard UPS displays the
actual RMS voltage.

It is speced to deliver true RMS readings. It will also display power
factor.

I do not know what crest factor it
can handle

Can't find it in the manufacturer's detailed specs.

but displays power factors of considerably
less than unity for most consumer electronics with DC
rectifiers off of the line voltage. It is a very nice
unit for the price.

Agreed.

 

[kony]

Lead-acid batteries (like is [sic!] in your car) have major issues
if they are completely drained, these are also the common battery
in cheap, well, just about everything.

The _only_ consumer products I've owned that used lead-acid batteries were
an early Sony Discman, and two APC SPSs. They are not common in consumer
products. They provide relatively high capacity at a low cost (which is why
they're used in SPSs and UPSs), but they are too-easily damaged by a full
discarge. I accidentally ruined a $45 batter for the Sony D-T10.

Practically all consumer grade UPS use lead-acid batteries.
Other common items using lead-acid batteries are some of the
handheld halogen spotlight (flashlights), and automotive
portable power station "thing-a-majigs" like those that can
jump a car or have a flashlight/pump/inverter. A few
transportation devices have them also like a kid's motorized
big-wheels or skateboard or motorized bike, etc.

Any decently designed UPS will cut off power before the lead
acide battery is drained to a critical level, although it is
still much harder on the battery to drain to any significant
extent, a large % of total capacity instead of the UPS
turning off as soon as possible.

 

[kony]

The price is all wrong. Nobody can build a decent product with
these specs for that price. You cannot even buy the components
needed in decent quality for that price.

Depending on your definition of "decent", this may be true,
but it's going to be roughly equivalent to what you'd buy
from APC as a 500VA for about $50 so if that's what the
budget allows, it's not an exceptionally low price (also
considering you can sometimes get the APC discounted or with
rebate putting it closer to $20-30 than to $50.

And with the el-cheapo battery in there, that figure will be down to
<1 minute in no time. And you would be well advised to test the
claim. Software can be made to lie to you, you know.

There's no reason to belive the battery is especially cheap,
it's going to be a standard lead-acid which is quite a
mature technology today, a commodity item essentially and
all that's really at question is whether the manufacturer
conservatively rated the unit or if it's unlikely to meet
the specs due to design budget constructions. Either way,
it's probably $70-100 worth of UPS, it will likely do as
well as anything else available for the same price.

If one were to pay >= 2X as much, naturally the expectation
would be it's either fancier or higher capacity or an online
type, etc, but there is no expectation the battery is
inherantly any higher quality per se, though probably higher
capacity due to being a larger size or a series of two
batteries.

 

[Richard Crowley]

"James Sweet" wrote ...

"Doc" wrote...

EE friend of mine compared one to a $2500 power analyzer at work, found
that the Kill A Watt performance is pretty much inline with the specs
printed for it. It's not as good as the professional equipment, but it's
really very impressive for what it is and certainly adequate for consumer
use. The wonders of modern microelectronics, it's amazing what they can do
with one inexpensive chip and a handfull of passive components. I still
remember when a pocket calculator was $300, then a few years later $50
would buy one just as effective, and not long after that they were under
$20 and those are all more capable than large machines costing many
thousands just a few decades earlier.

Agree completely. They are great when used with the regular
sine-wave mains power grid. However note that many have
been fried beyond repair when used with any kind of square-
wave source: inverter, UPS, etc. even "stepped sine" waveforms.

The problem appears to be the capacitive voltage divider used
to power the Kill-A-Watt electronics. The high frequency harmonics
deliver way too much power to the shunt regulator through the
capacitor and something fries.

 

[CBFalconer]

In general, I'd say that's my seat of the pants impression too,
though not always. I also notice they'll often come in clusters.

The transformer - if that's the correct term for it - big gray
basically cylindrical unit on top of a power pole - near my house
blew once. Powerful **BOOM** and a huge column of flame. Not sure
what the fuel for the flame was, do they have oil in them? Also a
bit disconcerting since anyone nearby surely would have been in
jeopardy from flaming debris.

Needless to say, power was out for a while on that one.

About 30 years ago a friend of mine was in a neighborhood where
multiple houses were supplied from one undersized transformer, and
the power company just ignored all requests to replace it. The
residents got together, and all turned on air-conditioners, ovens,
whatever, and then went outside to watch the transformer blow. It
did, and the scheme produced a new proper sized transformer.

 

[Arfa Daily]

Arfa,
I know that the current draw on these things is anything but sinusoidal.
My point was that the "Kill-A-Watt" seems to actually compute the RMS
value for that complex current waveform as well as non-sinusoidal voltage
waveforms.

David

OK. I'm not sure that 'RMS' is the right term to attach to any value derived
from a ragged-arsed waveform, as it is a mathematical function normally
associated with symetrical waveforms, which the draw by a SMPS may very well
not be, but I see what you're saying.

What I am trying to say is that a chip which is designed to produce an RMS
reading from a sine wave, may well produce a meaningful figure from a
non-sinusoidal waveform also, but *only* if it is still symetrical.

Arfa

 

[Matt Ion]

In general, I'd say that's my seat of the pants impression too, though
not always. I also notice they'll often come in clusters.

The transformer - if that's the correct term for it - big gray
basically cylindrical unit on top of a power pole - near my house blew
once. Powerful **BOOM** and a huge column of flame. Not sure what the
fuel for the flame was, do they have oil in them? Also a bit
disconcerting since anyone nearby surely would have been in jeopardy
from flaming debris.

Needless to say, power was out for a while on that one.

TRMOAS (that reminds me of a story)...

Years ago, when Atari ST computers were common, friends of mine lived in
and old house with just such a transformer on the adjacent pole.
Buddy had been trying for ages to get his girlfriend to start learning
to use the computer, but she was always afraid she'd "break something".

Finally one day, he convinced her to sit down and give it a try. She
took a deep breath, put her hands on the keyboard... as the room was lit
up by a brilliant flash through the window, shaken by a magnificent
**BOOM!**, and the computer screen went black.

Poor girl was in a complete panic, she was SURE it was her fault...

As it turned out, a drunk speeding down the road outside in his van had
gone THROUGH two nearby power poles and finally stopped just shy of
hitting a third; the first impact brought the lines down and was
sufficient to explode the transformer just outside their window.

....took her years to ever touch a computer again

 

[CBFalconer]

.... snip ...

OK. I'm not sure that 'RMS' is the right term to attach to any
value derived from a ragged-arsed waveform, as it is a mathematical
function normally associated with symetrical waveforms, which the
draw by a SMPS may very well not be, but I see what you're saying.

What I am trying to say is that a chip which is designed to produce
an RMS reading from a sine wave, may well produce a meaningful
figure from a non-sinusoidal waveform also, but *only* if it is
still symetrical.

Oh? Try a square wave, for example. Nice and symetrical. You are
over-simplifying.

 

[Arny Krueger]

"James Sweet" wrote ...

Yup. I logged lots of hours on mechanical calculators, Wang shared-logic
desk calculators, and the HP35.

Agree completely. They are great when used with the
regular sine-wave mains power grid. However note that
many have been fried beyond repair when used with any kind of
square- wave source: inverter, UPS, etc. even "stepped sine"
waveforms.

News to me. I've used mine with UPSs, but the application was not long-term
use.

The problem appears to be the capacitive voltage divider
used to power the Kill-A-Watt electronics. The high frequency
harmonics deliver way too much power to the shunt
regulator through the capacitor and something fries.

You're aware that capacitive voltage dividers have flat frequency response,
right?

 

[Richard Crowley]

"Arny Krueger" wrote ...

You're aware that capacitive voltage dividers have flat frequency
response, right?

Perhaps you are not aware that many inexpensive consumer
mains power supplies use a capacitor rather than a resistor
as the series element of a shunt-regulated power supply.

The circuit is essentially a high-pass circuit and very much has
a rising response.

 

[David]

"Arny Krueger" wrote ...

Perhaps you are not aware that many inexpensive consumer
mains power supplies use a capacitor rather than a
resistor
as the series element of a shunt-regulated power supply.

The circuit is essentially a high-pass circuit and very
much has
a rising response.

Then it is not a capacitive voltage divider. The better
designs have a series and shunt capacitor and a small
resistor before the rectifier. This arrangement does have a
flat(er) frequency response. I may take my Kill-A-Watt apart
to see what they used in the design.

David

 

[William Sommerwerck]

You're aware that capacitive voltage dividers have flat

Perhaps you are not aware that many inexpensive consumer
mains power supplies use a capacitor rather than a resistor
as the series element of a shunt-regulated power supply.

The circuit is essentially a high-pass circuit and very much
has a rising response.

I feel obliged to side with Arny. In the context of the original
description, it appeared as if the capacitive voltage divider was used to
drop the line voltage (as opposed to using a transformer). Putting a
capacitor in a shunt regulator is hardly the same thing.

 

[Fred Kasner]

In general, I'd say that's my seat of the pants impression too, though
not always. I also notice they'll often come in clusters.

The transformer - if that's the correct term for it - big gray
basically cylindrical unit on top of a power pole - near my house blew
once. Powerful **BOOM** and a huge column of flame. Not sure what the
fuel for the flame was, do they have oil in them? Also a bit
disconcerting since anyone nearby surely would have been in jeopardy
from flaming debris.

Needless to say, power was out for a while on that one.

Yes that transformer on top of the pole is what blew. The fuel was the
oil that acted as a dielectric insulator and heat dissipator. Older
versions have the liquid filled with PCBs (polybrominated biphenyls).
That stuff is quite hazardous.
FK

 

[Fred Kasner]

Seems I've heard an occasional complete drain-down and recharge will
extend the life of a rechargeable battery, that being constantly
partially discharged and recharged is what shortens their life. Not
so? Does it depend what kind of rechargeable it is?

The ones with big "memory" problems were the generally no longer used
NiCad cells. They tended to lose capacity if not fully discharged now
and then.
FK

 

[Gene E. Bloch]

OK. I'm not sure that 'RMS' is the right term to attach to any value derived
from a ragged-arsed waveform, as it is a mathematical function normally
associated with symetrical waveforms, which the draw by a SMPS may very well
not be, but I see what you're saying.

What I am trying to say is that a chip which is designed to produce an RMS
reading from a sine wave, may well produce a meaningful figure from a
non-sinusoidal waveform also, but *only* if it is still symetrical.

Arfa

RMS is defined & well-behaved for *any* wave form.

It is, as its name implies, just the square root of the integral over a
time interval of V-squared, divided by the integration time.

In an attempt at ASCII math notation: sqrt [(1/T)*{integral from t0 to
t0 + T} V^2 dt]

It works best (esthetically) if T is very long, or at least a whole
number of cycles of the waveform.

What is *not* well-behaved for unusual waveforms is the shortcut
formula giving RMS as a factor times the magnitude of the peak voltage.
I forget: is it Vmax*[sqrt(2)]/2 ? Whatever the correct shortcut is, it
is only correct for a sine wave (or a rectified sinewave, as is
mathematically obvious) )

 

[Arfa Daily]

OK. I'm not sure that 'RMS' is the right term to attach to any value
derived from a ragged-arsed waveform, as it is a mathematical function
normally associated with symetrical waveforms, which the draw by a SMPS
may very well not be, but I see what you're saying.

What I am trying to say is that a chip which is designed to produce an
RMS reading from a sine wave, may well produce a meaningful figure from a
non-sinusoidal waveform also, but *only* if it is still symetrical.

Arfa

RMS is defined & well-behaved for *any* wave form.

It is, as its name implies, just the square root of the integral over a
time interval of V-squared, divided by the integration time.

In an attempt at ASCII math notation: sqrt [(1/T)*{integral from t0 to t0
+ T} V^2 dt]

It works best (esthetically) if T is very long, or at least a whole number
of cycles of the waveform.

What is *not* well-behaved for unusual waveforms is the shortcut formula
giving RMS as a factor times the magnitude of the peak voltage. I forget:
is it Vmax*[sqrt(2)]/2 ? Whatever the correct shortcut is, it is only
correct for a sine wave (or a rectified sinewave, as is mathematically
obvious) )

I bow to your knowledge on this one, as you clearly have a better grasp of
the math involved than I. So are you saying that a meter that measures
*true* RMS anything - power, current, volts - will do so correctly even if
the waveform is asymmetric about the zero point ? I always believed ( was
taught ? - college seems so long ago now ... ) that even distribution about
the zero point, was a requirement for an accurate representation of an RMS
value ??

Arfa

 

[William Sommerwerck]

So are you saying that a meter that measures *true* RMS

anything -- power, current, volts -- will do so correctly even
if the waveform is asymmetric about the zero point? I always
believed that an even distribution about the zero point was a
requirement for an accurate representation of an RMS value.

The RMS value of a waveform has nothing to do with the shape or symmetry of
the waveform.

Assuming the meter is correctly designed, yes. RMS has a clear, specific
definition, and if the measurement is correctly implemented, the reading
will be correct.

Two qualifications... If the waveform is non-periodic, the measured RMS
value will vary according to the sample period. Also, if the waveform
includes a DC component, and the meter blocks it with a capacitor, then the
RMS reading will not include the DC component.

 

[James Sweet]

It does not extend the life. Back in the seventies when NiCd packs had
memory effect, folks were recommending deep discharge. But today, you
are more likely to wreck a NiCd pack by reverse-charging a cell that way.

I read recently that the memory effect "myth" was created due to cheap
chargers overcharging the batteries unless they were first fully discharged.
A decent intelligent charger should prevent this, and batteries have in
theory improved as well.

 

[James Sweet]

Hmm. Interesting. I trust he did this right and tested non-ohmic
loads such as a PC PSU as well?

Yes, of course, pure resistive loads can be measured just fine with a
multimeter. We were interested primarily in using it to measure discharge
lamp systems in which the power factor and current waveforms can be all over
the place and vary greatly with the state and condition of the lamp. If the
meter wasn't able to handle odd waveforms, the power factor measurement
function would be useless, but it works pretty well, accuracy is within
about 2% on the sample tested.