Observations on a UPS - follow up to a previous post

[John Williamson]

OK. Well in that case, I don't think that I was over-simplifying, because if
you have read the whole thread, you will see that it was I who questioned
the validity of attaching an RMS value to a non-sinusoidal waveform.
However, several posters then came back to me with considerable levels of
mathematical proof, to say that RMS was a valid notion for any waveshape or
symmetry factor, the only qualifiers being DC content or variable cycle
periodicity. Although it might not be too clear, that second paragraph was
more of a musing based on that. My original contention was that a power
meter (or whatever) designed to derive and display an RMS value from a sine
wave, would not give a meaningful reading from non-sinusoidal or
non-symmetrical drawing loads, such as a SMPS may be, for instance. The
replies suggested that the waveshape was immaterial, and that the chipset
could very easily still calculate a meaningful result. I was a little
sceptical about this, as it seemed to fly in the face of what I was taught
many years ago in college, but I bowed to what seemed to be superior
knowledge in the field.

Now, you seem to be saying something quite different ? Comments ?

Arfa

Sorry to butt in here, but when I was studying such things, the RMS
value of a current or voltage waveform was calculated by working out the
area inside the curve plotted over a full cycle, which then allowed you
to calculate an equivalent DC value. This involved counting squares on
graph paper of the plotted waveform or similarly counting squares on a
calibrated oscilloscope tube face. The earlier & most of the current
cheap meters that give an alleged RMS reading take a peak reading &
apply a correction factor of 0.707 to it (1 divided by the (near enough)
square root of 2), as that gives the right answer with a clean sine
wave, which is what most of these meters are used to measure. (Mains
power round here is near enough a pure sine wave that you can ignore the
error, as it's less than the accuracy of the meter)
The RMS value of a (theoretical) pure square wave is exactly the same as
the average of the absolute values of the positive & negative peaks, as
the value is either fully positive or fully negative, with, in theory,
no other value being present.

The most (theoretically) accurate way to measure RMS values is to use a
hot wire meter, which doesn't care what the waveform is, it just
measures the heating effect which is more or less frequency independent
& includes any DC offset automatically.


Tciao for Now!

John.

 

[kony]

I actually think that at the moment, digital display technology - without
wishing to open up *that* can of worms again - lags behind CRT display
technology, by a significant amount. Next time you go to the cinema, look up
at the booth window and see if you can see film looping around the ceiling.
If you can't, then it uses one of those new-fangled DLP video projectors.
Sit back comfortably with your popcorn, and wonder what's happened to your
eyes, when the first car drives across the screen ... d;~}

Many people in their daily use cannot see any lag or
ghosting from 19" and smaller LCD computer monitors.

If you can't actually see it, does it matter if it exists?
I can play 50 FPS video or games running at over 50 FPS on a
19" LCD computer monitor and not see any problems except the
obvious lack of contrast (but with CRT I am spoiled in this
respect, having bought Diamondtron tube based monitors for
the last few I used myself before switching to primarily LCD
usage).

 

[Arfa Daily]

Sorry to butt in here, but when I was studying such things, the RMS value
of a current or voltage waveform was calculated by working out the area
inside the curve plotted over a full cycle, which then allowed you to
calculate an equivalent DC value. This involved counting squares on graph
paper of the plotted waveform or similarly counting squares on a
calibrated oscilloscope tube face. The earlier & most of the current cheap
meters that give an alleged RMS reading take a peak reading & apply a
correction factor of 0.707 to it (1 divided by the (near enough) square
root of 2), as that gives the right answer with a clean sine wave, which
is what most of these meters are used to measure. (Mains power round here
is near enough a pure sine wave that you can ignore the error, as it's
less than the accuracy of the meter)
The RMS value of a (theoretical) pure square wave is exactly the same as
the average of the absolute values of the positive & negative peaks, as
the value is either fully positive or fully negative, with, in theory, no
other value being present.

The most (theoretically) accurate way to measure RMS values is to use a
hot wire meter, which doesn't care what the waveform is, it just measures
the heating effect which is more or less frequency independent & includes
any DC offset automatically.


Tciao for Now!

John.

Yes, all agreed, but the shape of the mains waveform is immaterial, unless
you are talking a purely resistive load like a light bulb. The point that I
have been trying to make all along is that when you are trying to measure
power, it's a function of both voltage and current draw, and in the case of
a SMPS, especially one that's in standby mode, the current draw from the
mains supply voltage, is very likely to be anything *but* sinusoidal.

Arfa

 

[Arfa Daily]

Many people in their daily use cannot see any lag or
ghosting from 19" and smaller LCD computer monitors.

If you can't actually see it, does it matter if it exists?

Well no, of course not. But I would be surprised if anyone actually couldn't
see it. I prefer to believe that it's a little bit of 'King's New Clothes'
syndrome, and people don't really *want* to see it, having just shelled out
a bunch of their hard-earned, on what they believed was going to be better
than they already had. Even my wife can see it, without any prodding from
me, and she's about as technical as a pound of oranges ...

I can play 50 FPS video or games running at over 50 FPS on a
19" LCD computer monitor and not see any problems except the
obvious lack of contrast (but with CRT I am spoiled in this
respect, having bought Diamondtron tube based monitors for
the last few I used myself before switching to primarily LCD
usage).

Yes, there is the lack of contrast issue, which is not insignificant in
itself. My son plays video games on his PC at high frame rates also. He also
has an expensive HP 4:3 LCD, and whilst it's pretty good at displaying fast
motion, there is, never-the-less, motion blur that wasn't there when he used
CRT monitors. When a pixel represents decimals of a uS, and the time to
switch that pixel is around a mS at best, there must be motion blur created.

Arfa

 

[John Williamson]

Yes, all agreed, but the shape of the mains waveform is immaterial, unless
you are talking a purely resistive load like a light bulb. The point that I
have been trying to make all along is that when you are trying to measure
power, it's a function of both voltage and current draw, and in the case of
a SMPS, especially one that's in standby mode, the current draw from the
mains supply voltage, is very likely to be anything *but* sinusoidal.

Arfa

It's also likely not to be a simple product of the RMS current drawn &
the RMS voltage of the supply, as the input circuitry contains reactive
and regulatory elements that alter the phase relationships between the
current & the voltage in a manner that can't necessarily be predicted
easily.
Most of the ones I've looked at have a rectifier across the mains feed,
with inductive & capacitive elements in the circuit before the
rectifier, with a SMPS pulling power from the rectifier after a
smoothing filter. This gives a power factor that varies with load,
possibly cyclically even at steady load if the oscillator of the SMPS
isn't locked to the incoming mains frequency.
Then the invertor on the output just runs off the DC from the battery
pack/ SMPS combination.

As you say, though, the shape of the mains waveform is immaterial, apart
from harmonics altering the power factor by altering the relative
impedances of the inductors & capacitors.

Tciao for Now!

John.

 

[Arny Krueger]

That's one way, subject to the problems with accurately measuring the
heating of a resistive element.

Another way to obtain a true RMS reading without complex electronics is to
use a certain kind of meter movement that mechanically integrates the
product of the current and the voltage. There are two sets of windings in
the meter, one for current and one for voltage. Their attraction or
repulsion that drives the pointer is based on the product of the current in
the windings. I have one that was made by RCA, and a very common tool during
the 50s, 60s, 70s, and 80s.

Simply not true. Even light bulbs have some degree of sensitivity to the
waveform, unless they have filaments with very long thermal time constants.

Historically rectifier-based power supplies have been very sensitive to wave
form shape, because their output voltage is strongly influenced by the peak
value of the power line wave.

In the old days some magnetic power line voltage regulators put out a fairly
pure square wave. This did a pretty fair job of heating tube filaments, but
did not provide full B+ voltage from the power supply. The problem was the
low peak voltage. If you jacked up the line voltage to get full B+, the tube
filaments ran hot and tube life suffered.

The point that I

Agreed. However there is a newer kid on the block, and that's the power
factor corrected SMPS. This technology has been reduced to an IC, and it
shows up in items as humble as compact flourescent light bulbs. If the
power factor is 1.00 or approaches it, then the current and voltage are
largely in-phase.

 

[Andrew Erickson]

That's one way, subject to the problems with accurately measuring the
heating of a resistive element.

Another way to obtain a true RMS reading without complex electronics is to
use a certain kind of meter movement that mechanically integrates the
product of the current and the voltage. There are two sets of windings in
the meter, one for current and one for voltage. Their attraction or
repulsion that drives the pointer is based on the product of the current in
the windings. I have one that was made by RCA, and a very common tool during
the 50s, 60s, 70s, and 80s.

And a third practical way, which I suspect devices like the kill-a-watt
meter and "true RMS" digital multimeters use, is to sample the
voltage/current waveform and then use a microcontroller to perform the
appropriate integration. Microcontrollers are amazingly cheap and
powerful these days. As long as the waveforms you're measuring are
relatively slow (compared to the sampling frequency), it should be quite
accurate. For mains work, this isn't a hard thing to achieve--a few
kilohertz sampling rate is probably overkill.

 

[David]

Yes, all agreed, but the shape of the mains waveform is

It's also likely not to be a simple product of the RMS
current drawn & the RMS voltage of the supply, as the
input circuitry contains reactive and regulatory elements
that alter the phase relationships between the current &
the voltage in a manner that can't necessarily be
predicted easily. (snip)
Tciao for Now!

John.

This all started with a discussion of the Kill-A-Watt meter
which measures RMS voltage, RMS current, and Watts. Maybe
the following will clear some of the confusion.
The computation of RMS was described in an earlier post. The
product of RMS voltage and RMS current is Volt-Amps. Power,
and therefore power consumption (Watts=power/second), is
computed by the integration of the product of instantaneous
volts and amps over time. The power factor then becomes
Watts/Volt-Amps. The accuracy of the wattage and RMS
calculations with voltage or current waveforms that change
rapidly is related to the sampling rate used in the
integration. Any wave shape for either voltage or current
will produce mathematically meaningful RMS and power
measurements. I am sure someone will point out any mistakes
I have made here.

David

 

[CBFalconer]

.... snip ...


Yes, there is the lack of contrast issue, which is not
insignificant in itself. My son plays video games on his PC at high
frame rates also. He also has an expensive HP 4:3 LCD, and whilst
it's pretty good at displaying fast motion, there is,
never-the-less, motion blur that wasn't there when he used CRT
monitors. When a pixel represents decimals of a uS, and the time to
switch that pixel is around a mS at best, there must be motion blur
created.

I suspect that the real effect is caused by the pixel decay time.
CRTs operate with a refresh rate between 25 and about 100 hz,
depending on interlace, resolution, etc. This means a pixel will
be refreshed no sooner than 10 mS (up to about 40) from the earlier
energization. If, at that refresh time, the pixel has a
substantial carry-over from the previous level, there will be
blurring. If the carry-over is too small, there will be flickering
and other evil effects. I believe the LCDs have, effectively, zero
carry-over, and compensate by having an instantaneous reset of any
previously set level; i.e. they don't require interlace, refresh,
etc. except to show motion.

 

[William Sommerwerck]

Power, and therefore power consumption

(watts=power/second)...

Power is energy/second. Power is not consumed, energy is consumed.

 

[William Sommerwerck]

Agreed. However there is a newer kid on the block, and that's the power

factor corrected SMPS. This technology has been reduced to an IC, and it
shows up in items as humble as compact flourescent light bulbs. If the
power factor is 1.00 or approaches it, then the current and voltage are
largely in-phase.

I checked some excellent CFLs from Home Despot (instant turn-on, excellent
color balance) with the Kil-a-Watt. It showed a power factor around 65%,
which struck me as rather low. Such a low PF also partly offsets the
money-saving advantages of fluorescent lamps.

Opinions, anyone?

 

[CBFalconer]

.... snip ...

This all started with a discussion of the Kill-A-Watt meter which
measures RMS voltage, RMS current, and Watts. Maybe the following
will clear some of the confusion.
The computation of RMS was described in an earlier post. The
product of RMS voltage and RMS current is Volt-Amps. Power, and
therefore power consumption (Watts=power/second), is computed by
the integration of the product of instantaneous volts and amps
over time. The power factor then becomes Watts/Volt-Amps. The
accuracy of the wattage and RMS calculations with voltage or
current waveforms that change rapidly is related to the sampling
rate used in the integration. Any wave shape for either voltage
or current will produce mathematically meaningful RMS and power
measurements. I am sure someone will point out any mistakes I have
made here.

You omitted that a periodicity is required. An infinite wavelength
is also allowable.

BTW, please do not remove attributions for material you quote.

 

[Scott Dorsey]

I checked some excellent CFLs from Home Despot (instant turn-on, excellent
color balance) with the Kil-a-Watt. It showed a power factor around 65%,
which struck me as rather low. Such a low PF also partly offsets the
money-saving advantages of fluorescent lamps.

Opinions, anyone?

Well, your computer switcher probably has a power factor of about 1.2 to
1.5... so run a couple computers and the lead and lag will cancel one
another out...
--scott

 

[Arny Krueger]

I checked some excellent CFLs from Home Despot (instant turn-on, excellent
color balance) with the Kil-a-Watt. It showed a power factor around 65%,
which struck me as rather low. Such a low PF also partly offsets the
money-saving advantages of fluorescent lamps.

Opinions, anyone?

The dimmable CFLs I get from eBay (love those California lawmakers!) have PF
speced > 0.90. I haven't got around to testing one, though.

 

[Gene E. Bloch]

The integral is peak voltage times current. Simple. Not 0.7 *
peak voltage.

No, the integral is instantaneous voltage squared. Simple.

Current is also constant for resistive loads, not
proportional to voltage. RMS doesn't work.

You are contradicting Ohm's law, e = ir, which can be rearranged to
read i = e/r. Current is precisely proportional to voltage for
resistive loads. Simple.

I've used small letters to follow an old convention (is it still used?)
that lower-case letters represent varying values and upper-case letters
represent constant values (such as in analyzing DC circuits).

 

[Gene E. Bloch]

I checked some excellent CFLs from Home Despot (instant turn-on, excellent
color balance) with the Kil-a-Watt. It showed a power factor around 65%,
which struck me as rather low. Such a low PF also partly offsets the
money-saving advantages of fluorescent lamps.

Opinions, anyone?

Great minds: I also call it Home Despot...

 

[William Sommerwerck]

The dimmable CFLs I get from eBay (love those California lawmakers!)

have PF speced > 0.90. I haven't got around to testing one, though.

What brand?

I bought a dimmable GE two years ago, and it worked. You get only about 10
steps, at the top of the X-10's 256-step range, but it works.

 

[Arny Krueger]

What brand?

ULA, made you-know-where.

My understanding is that these bulbs are being sold in California for about
$1.00 each in Wal-Marts, with subsidy from the local power company.

I bought a dimmable GE two years ago, and it worked. You get only about 10
steps, at the top of the X-10's 256-step range, but it works.

The dimmable CFs I've been using have a standard Edison base and fit in
standard light bulb sockets. They are infinitely variable over a range that
goes down to pretty dark and then nothing, to full bright.

I use a few of them around the house, driven by standard wall-plate
residential dimmers.

I'm using 48 of them in 6 chandeliers driven by standard DMX-controlled
quad dimmer packs at church. Their brightness/drive curve is nonlinear, but
useable.

As you say, the color temperature is very constant over a usable range of
intensities compared to incadescent PAR bulbs. While they start pretty much
on the dime, they do get about 50% brighter the first minute or two of
operation.

If you've got the fixtures and ballasts that are designed for them,
4-terminal dimmable flourescents are marvelous. They dim over about the same
range of brightness and as linearly as an incadescent, but with constant
color temperature.

 

[William Sommerwerck]

I use a few of them around the house, driven by standard

wall-plate residential dimmers.

Rheostats? Triacs? I'm not familiar with the current technology. (X-10 is
triac-controlled, I believe.)

As you say, the color temperature is very constant over a
usable range of intensities compared to incadescent PAR bulbs.

Actually, I didn't say that, but you'd expect it to be so, given that a
fluorescent lamp is a quantum device.

While they start pretty much on the dime, they do get about
50% brighter the first minute or two of operation.

That's what I noted with the Home Despot lamps. It was startling at first to
see a fluorescent lamp come on faster than an incandescent.

 

[Arno Wagner]

Rheostats? Triacs? I'm not familiar with the current technology. (X-10 is
triac-controlled, I believe.)

Actually, I didn't say that, but you'd expect it to be so, given that a
fluorescent lamp is a quantum device.

Very much so, but mostly in the same sense that a glowing
pice of metal is a quantum device....

Arno