Re: AC power measurement


"Johnny B Good" <jcs.computers***@xxxxxxxxxxxxxxxx> wrote in message
news:313030303737303649432C1A09@xxxxxxxxxxxxxxxxxxx
The message <bfy0l.66547$Qf1.59314@xxxxxxxxxxxxx>
from "treblesix" <mike.666NOSPAM@xxxxxxxxxxxxxxxxx> contains these words:


"davesurrey" <dave@xxxxxxxxxxxxxxxx> wrote in message
news:494290f1$0$1347$fa0fcedb@xxxxxxxxxxxxxxxxx

"treblesix" <mike.666NOSPAM@xxxxxxxxxxxxxxxxx> wrote in message
news:bdv0l.31634$EL4.31207@xxxxxxxxxxxxxxxx

"Rob Morley" <nospam@xxxxxxxxxxxx> wrote in message
news:20081212132728.2789250c@xxxxxxxxxxx
On Fri, 12 Dec 2008 12:32:09 -0000
"treblesix" <mike.666NOSPAM@xxxxxxxxxxxxxxxxx> wrote:


"davesurrey" <dave@xxxxxxxxxxxxxxxx> wrote in message
news:4941786e$0$1349$fa0fcedb@xxxxxxxxxxxxxxxxx

It's not too clear what you are saying here.
The OT asked if taking an average of the product of instantaneous
voltage and current will give the power dissipated.
The simple answer is yes and it's irrelevant what the waveform is
or PF of the load.
However I do agree that it might be best if the OT begs/borrows a
suitable meter.
Cheers
dave


"measuring AC power consumption " was the original precept.
In that context, waveform, PF and any PF correction makes
considerable differences from your "simple answer". Resistive
loads are one thing ...complex inductive and/or capacitative
loads are somewhat different. Ask any electrical enginner or tutor
:0)


His "simple answer" is still correct. You take the mean of a bunch
of power figures, those figures are the product of instantaneous
voltage and current measurements over half a cycle (or two adjacent
half
cycles if there's a DC element). Look at the scope trace to decide
how
many readings it is reasonable to take.


A sampling oscilloscope with a take your best guess plugin is
somewhat removed from the usual op-amp sculling around in
odds and ends boxes. Some years ago, some folks used to fiddle
the electricity companies by the suitable attatchment of a huge
capacitor so as to produce substantial reduction of the Metered
power consumption by phase shifting. Power factors do matter,
especially to electricity companies and people who need to
know the actual power consumption.

A simple way for mains powered devices is to measure the
RMS current and multiply by the nominal/actual RMS voltage.
Alternatively, for an approximation of actual power consumption
of domestic/office devices... just buy one of the cheaper meters
made for the job...or just read the meter appropriately when
something is switched on for a while and then off again..

treblesix, the OT did say he couldn't afford a power meter.

Also I thought you understood PF in which case why do you suggest
he just takes the product of rms current and rms voltage
(not sure what you mean by nominal/actual rms...rms is just rms.)
This will give the apparent power (kVA)
and NOT the actual power dissipated. Sorry to say it but you really
need
to make clear what you are saying in such discusssions else you risk
confusing others.

Not trying to pick a fight but just want to ensure there is some
clarity.

Cheers
Dave

I originally mentioned PF (which I do understand) as a caveat
in the design of a simple op-amp measurement device. That is
a valid comment where any vaguely accurate measurement is to be
made. I did not drag in sampled measurements ...but have done
such things many times.

I do not need to do anything in colloquial newsgroup contexts.
Nominal UK AC supply is 240Vrms. The actual value is usually
different by some margin ... a fact which most would appreciate.

If you wish to consider power dissipation, I would suggest a
digital thermometer and doing the math on heating and cooling curves
in an insulated enclosure.

Not what the OP was after. He wants to fabricate the equivilent of my
Metrawatt analogue wattmeter using 'simple' analogue multiplier
techniques based on analogue integrated circuits such as op-amps and
FETs and such.

I reiterate that the measurement of electrical power consumption
is not a simple matter to do with any significant accuracy with a
few bits out of a spares box ..especially when significantly inductive
loads are to be found in many domestic items.

I think the OP fully appreciates this.

If you require further clarity RTFM ...power consumptions are
usually stated therein .... or if that is too much trouble then
the rating plate on the item should give some indication.

The rating plates on most domestic appliances are there to fulfil legal
obligations regarding safety issues rather than provide a precise
figure. The only time I've ever seen the rating plate figures match up
with 'reality' was (and still is) on my 3KW at 240volt / 2.8KW at 230
volt electric kettle.

In practice you can measure the current to reasonable accuracy
without regard to PF as the meter measures the small voltage
caused by the current flowing through a low value resistance.
If you then multiply that by the nominal 240 vac or the actual
rms supply voltage you will get a reasonable value for the actual
power used. In the absence of a laboratory full of equipment or
very expensive dedicated kit, this proves adequate for most purposes
in my own hands on R&D experience.

The OP, quite clearly, wants to measure 'real power' not the VA
consumption and was (quite correctly) wondering whether it was possible
to create an analogue multiplier using op-amps which would give an
averaged output of the continuous product of volts and amps, thus
providing the basis for a 'true power' reading instrument regardless of
PF.

Sadly, this isn't possible due to the need to be able to multiply volts
and amps regardless of their instantaneous polarity so as to correctly
calculate both negative and positive power products required to produce
a correctly averaged true power reading. IOW, there is no electronic
equivilent to the good old fashioned analogue wattmeter which can
perform such a function with consumate simplicity.

The only reliable way to electronically generate the 'real power'
readings is to use suitably accurate[1] ADCs for both voltage and
current waveforms and mulitply each such pair of sample values obtained
during each sampling interval and _summing_ successive VA products over
a suitable integrating interval (at least one full cycle of mains supply
per update, but more likely several - most likely 50 or 60 cycles of the
mains supply for an updating interval of one a second).

In principle, it should be possible to make an interface box with a
couple of small 3VA low voltage mains transformers (one to act
conventionally to provide the voltage sample, the other, perhaps
modified[2], to act as the current transformer and provide the current
sample) which would let you use a PC's 'sound card' interface with
suitable software to emulate an otherwise horrendously expensive
'digital wattmeter' (certainly, "outclass" those cheap ten quid meters
you can buy in the likes of Aldi/Lidl/Netto and other good 'electronics
stores' such as Maplin ;-)

[1] Although the mains fundamental frequency is a mere 50 or 60 Hz
(depending on what part of the world you happen to be living in), you
still need a reasonably high sampling rate to maintain accuracy in the
ensuing calculations, probably by two orders of magnitude (5000 samples
per second or above).

You also need a good bitwise resolution, more so for the current
samples which have to cope with minimums that might range from as low as
half a milliamp rms (and lower) to transient peaks that might well
exceed 25 amps (the peak current for 13A rms being 18.382A). The voltage
waveform usually staying within +/-10% of the nominal supply voltage is
easier to 'fit' into a more modest number of bits of resolution, but
don't overskimp on this since innacuracies in either of the two
parameters will be multiplied up.

If you wish an accuracy over the 1 to 3000 Watt range on uk ring mains
outlets without range switching on the current transformer, you need at
least 16 bit signed integer samples for the current sensor readings and
11 or 12 bits signed integer resolution for the voltage samples.

[2] Most readily available small mains transformers in the 3 to 6 VA
sizes usually have output voltages not much lower than 6 volts. This
isn't a problem for the voltage sensing (you can use a simple resistor
divider to attenuate the voltage down to a peak of 1v required to avoid
overloading the line input buffer amp of the sound card) Only the older
ISA sound cards have sufficient buffer amp 'head room' to allow FSD
values to be achieved in the line input without clipping. Sadly, all the
PCI based 'reference designs' waste the two MSBs of resolution due to
analogue clipping in the stage that feeds the ADC.

Luckily, it's usually a relatively trivial exercise to replace the
hundred or so turns of the secondary with the half dozen or so turns of
much thicker wire required to carry the anticipated maximum of 13A rms
when used as a current transformer. The 240v primary now becomes the
secondary with a suitable shunt resistor which could be made up from two
resistors to form a tapped shunt to give you a nominal 2 and 20 amp
setting to improve the sub 100 watt accuracy of your soundcard based
wattmeter.

HTH

--
Regards, John.

Please remove the "ohggcyht" before replying.
The address has been munged to reject Spam-bots.

That is a pretty good concise trip through the technological
issues and alternatives John :0)
We used to make current transformers with toroid cores and
an apropriate number of secondary turns. For a year or two,
I was one of the local Analogue Devices sales rep's best customers
for DACs and ADCs. This is definitely not something that can be
easily and cheaply built out of odds and ends from a box of bits
if, repeat if, significant accuracy is required. Then of course the
problem is the availabilty of a standard to calibrate against. It is
refreshing to hear from someone else who understands the benefits
of analogue instruments over digital ones in certain circumstances.
This is why even in retirement, I possess an Avo8 and a Fluke
3.5 digit DMM.

I recall that there used to be several retailers of kits and addon
ADC interfaces for retro computers.. In any event, accuracy
requires costs money and technology beyond the budgets of most
people. A £10 wattmeter or a look at the rating plate is probably
way more than most people need or want. Yes, rating plate info can be
misleading in that it usually indicates the maximum (non surge) power
consumption which may only arise in certain conditions. Eg., a normal
washing machine only consumes a nominal 3Kw when the internal heater is
operating. This cuts out and changes to some 100W when the motor
enters a wash cycle ....or a few hundred watts when spinning. Worth
bearing in mind when using your Electricity Co's supplied wattmeter :0)





.