Prev: Strategies for temperature-compensating differential pairs
Next: Flow Sensors revistied
From: Jan Panteltje on 11 May 2010 15:55
On a sunny day (Tue, 11 May 2010 11:39:55 -0500) it happened "Tim Williams"
<tmoranwms(a)charter.net> wrote in <hsc18r$h9p$1(a)news.eternal-september.org>:

>"Jan Panteltje" <pNaonStpealmtje(a)yahoo.com> wrote in message
>news:hsbn1p$k1a$1(a)news.albasani.net...
>>>> If they use a 10 bits ADC, then for a max of 2 kW the miniumum is
>>>> already 2 W.
>>>
>>>2mW*
>>
>> A 10 bit ADC has 1024 steps
>> 2000 / 1024 is just about 2
>> ???????????????
>>
>> Hello?
>
>Presumably, they're converting voltage and current seperately.
>1/1024 = -60dB, or 2kW/1e6 = 2mW.
>
>Tim

VERY clever!
But 1000 + or min 1 x 1000 + or - 1
makes worst case:
999 x 999 = 99800
1001 x 1001 = 100201
the diffence is now 4 W
hehe
:-)


From: Tim Williams on 11 May 2010 16:04
"John Larkin" <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote in message
news:6b8ju5tum627iakroik4mod7n4rah1f1be(a)4ax.com...
> Is the 2mW* thing some sort of joke? Please explain it.

"*" indicates a correction to an error (usually tpyographical).

*typo

Like that.

An error of 2W is obviously wrong, and was hopefully only a typo.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms


From: Martin Brown on 11 May 2010 17:45
Tim Williams wrote:
> "Jan Panteltje" <pNaonStpealmtje(a)yahoo.com> wrote in message
> news:hsbn1p$k1a$1(a)news.albasani.net...
>>>> If they use a 10 bits ADC, then for a max of 2 kW the miniumum is
>>>> already 2 W.
>>> 2mW*
>> A 10 bit ADC has 1024 steps
>> 2000 / 1024 is just about 2
>> ???????????????
>>
>> Hello?
>
> Presumably, they're converting voltage and current seperately.
> 1/1024 = -60dB, or 2kW/1e6 = 2mW.

Line voltage is usually 230 +/- 20 or around ~10% variation.

So the dynamic range is limited by the current measurement. This tends
to suggest that they actually use a 12 bit ADC in the common ones over
here since they will measure 1W and max out at ~4kW.

Clamp on ones are much cruder and good to about 10-20W but with a much
higher maximum to measure the entire domestic consumption.

Regards,
Martin Brown
From: John Larkin on 11 May 2010 18:21
On Tue, 11 May 2010 15:04:19 -0500, "Tim Williams"
<tmoranwms(a)charter.net> wrote:

>"John Larkin" <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote in message
>news:6b8ju5tum627iakroik4mod7n4rah1f1be(a)4ax.com...
>> Is the 2mW* thing some sort of joke? Please explain it.
>
>"*" indicates a correction to an error (usually tpyographical).
>
>*typo
>
>Like that.
>
>An error of 2W is obviously wrong, and was hopefully only a typo.
>
>Tim

He was apparently assuming that 10-bit quantization of the current
waveform limits power resolution to about 1 part in 1000. We are
talking about an AC line power meter, where the voltage doesn't change
much, and a 10-bit ADC is used to sample both the voltage and current
waveforms.

(Funny, I designed a 6-channel "power strip" meter recently, 10 bit
mux'd ADC, with ZigBee communications. I charged $1 for the design.)

A power meter needs extra headroom for the 1.414 peak thing, and more
for non-ideal waveforms, so the quantization thing is even worse than
it looks. On the other hand, if you dither the current signal before
you digitize it, the quantization disappears it's easy to resolve one
watt out of 20KW.

John


From: Paul Keinanen on 12 May 2010 01:38
On Mon, 10 May 2010 19:30:55 GMT, Jan Panteltje
<pNaonStpealmtje(a)yahoo.com> wrote:

>
>It is lonly logical.
>If they use a 10 bits ADC, then for a max of 2 kW the miniumum is already 2 W.
>Add any offset in the current sensing, or even one bit eror,
>and you are at a minium of 4 W etc etc.

What exactly are you trying to measure, real power or apparent power ?
Is it assumed that the current (and/or voltage) looks like a sine
wave?

While a constant amplitude clean sinusoid voltage connected to a pure
resistive load will produce an in-phase sinusoid current, measuring
the peak current (and assuming nominal voltage) could be used to give
some real power indication. Measuring the current with 10 bit ADC
could give a 1000:1 power range (such as 2 kW:2 W).

Any load with some reactive components would cause a phase shift
between voltage and current and the simplistic system would only
produce the apparent power. Any rectifiers would create a large peak
to average current ratio, so the 1.41:1 peak/rms ratio for a sinusoid
is no longer valid.

For real power, you must multiply the instantaneous voltage with the
instantaneous current and average out. This can be done by an analog
multiplier followed by analog averaging before the ADC or use separate
ADCs for current and voltages sampled at a high frequency and do the
multiplying and averaging in digital domain.

With rectifier loads, the peak current can be quite high, often much
higher than the fuse ratings and the harmonics on the voltage can
cause long duration peaks higher than the fundamental frequency. Thus
the signal conditioning, multiplication and averaging require a quite
large dynamic range, regardless if the most processing is done in
analog or digital domain.

First  |  Prev  | 
Pages: 1 2 3 4 5
Prev: Strategies for temperature-compensating differential pairs
Next: Flow Sensors revistied