CMRR

On Wed, 29 Apr 2020 07:57:03 -0000 (UTC), Jasen Betts
<jasen@xnet.co.nz> wrote:

On 2020-04-28, Ricky C <gnuarm.deletethisbit@gmail.com> wrote:

and current through a resistor and measure the temperature rise.
(You'd have to have some means of getting rid of all the errors...
Like some sort of differential gizmo that heats and cools
(to environment) and known I*V as a calibration.. the details
are a bit fuzzy. :^)
But I was wondering if there are other 'silly' ways
to measure power/ energy*time?

The old disk meters actually use two coils, one for current and one
for voltage. Their effect on an aluminum disk is proportional to the
product including the phase angle, in other words the complex product
of voltage and current. The meter has fixed magnets that create drag
making the rotational speed of the rotor proportional to the power
being drawn.

So really the meter is measuring the voltage and the current
including the relative phase, then finding the product though the EM
effects on the disk.


Today's meter's do the same thing by measuring the instantaneous
voltage and current and finding the product. It is easy to find
various circuits to multiply two quantities. It's not hard to find
various ways to measure voltage and current. Async voltage to PWM
converters for both the voltage and the current taps, then use an XOR
to find the instantaneous product. Smooth the pulses and you have an
analog signal proportional to the instantaneous power.

seems kind of dodgy, you need to keep those two clocks decorellated.

Generate analog signals for the instantaneous voltage and current.
Generate the log of the signals and sum them giving power. This would
require the two inputs to be biased since the input to the log can't
go through zero. The bias would be subtracted out at the end
potentially creating a problem of a small difference of two large
signals.

bias precision could be a problem, you have the same issue with the
PWM scheme.

pass a current proportional to line voltage through a hall effect cell and
derive the magnetic field from the line current

output voltage will be proportional to istantaneous power.

Halls have been used, mostly in the ancient past, in expensive and not
very accurate AC power transducers, but not, as far as I know, in
utility-worthy meters.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Ricky C]

On Wednesday, April 29, 2020 at 10:25:04 AM UTC-4, jla...@highlandsniptechnology.com wrote:

On Wed, 29 Apr 2020 07:46:30 +0300, upsidedown@downunder.com wrote:

On Tue, 28 Apr 2020 16:20:22 -0700 (PDT), Ricky C
gnuarm.deletethisbit@gmail.com> wrote:

But I was wondering if there are other 'silly' ways
to measure power/ energy*time?

The old disk meters actually use two coils, one for current and one for voltage. Their effect on an aluminum disk is proportional to the product including the phase angle, in other words the complex product of voltage and current. The meter has fixed magnets that create drag making the rotational speed of the rotor proportional to the power being drawn.

So really the meter is measuring the voltage and the current including the relative phase, then finding the product though the EM effects on the disk.

Today's meter's do the same thing by measuring the instantaneous voltage and current and finding the product. It is easy to find various circuits to multiply two quantities. It's not hard to find various ways to measure voltage and current. Async voltage to PWM converters for both the voltage and the current taps, then use an XOR to find the instantaneous product. Smooth the pulses and you have an analog signal proportional to the instantaneous power.

Generate analog signals for the instantaneous voltage and current.
Generate the log of the signals and sum them giving power.

log(U) + log(I) = log(P)

If you need to sum (for averaging) individual power samples, you need
to use antilog( log(P) ) before summing different samples. The
exponential function required dynamic range is much larger than for
either log(U) or log(I).

This would require the two inputs to be biased since the input to the log can't go through zero. The bias would be subtracted out at the end potentially creating a problem of a small difference of two large signals.

Just use abs() (full wave rectification) before the log(),

The instantaneous power product goes negative when the power factor is
1. You've got to preserve the signs.

No one said to toss the sign. The point is before taking the log the absolute value can be used to make the log more tractable and the sign kept separately as in sign * exp(log(s)). Do the math by adding the log(V) to log(I) and separately multiplying the signs.

It's simple math. High school kids can do it. You should try to learn some of this stuff.


> Just multiply! Preferably in software.

I guess you missed the post about brain storming the approach. Not looking for perfect or even obviously practical. Just looking for novel.

A utility-class meter might handle 30KW and needs to be accurate to a
few watts. That's borderline impossible to do with analog electronics,
borderline trivial with a mediocre ADC and a cheap uP.

I'm glad you understand the utility of digital electronics. Now you just need to learn to use them.

--

Rick C.

-+ Get 1,000 miles of free Supercharging
-+ Tesla referral code - https://ts.la/richard11209

 

[whit3rd]

On Tuesday, April 28, 2020 at 9:46:36 PM UTC-7, upsid...@downunder.com wrote:

On Tue, 28 Apr 2020 16:20:22 -0700 (PDT), Ricky C
gnuarm.deletethisbit@gmail.com> wrote:

The old disk meters actually use two coils, one for current and one for voltage. Their effect on an aluminum disk is proportional to the product

log(U) + log(I) = log(P)

If you need to sum (for averaging) individual power samples, you need
to use antilog( log(P) ) before summing...

> Just use abs() (full wave rectification) before the log(),

You'll need to handle negative power, when V and I are opposite in sign. Grab
the signs before taking those absolute values!

A hybrid (keep the wheel) with some turns-counting in silicon is an elegant compromise.

 

[Ricky C]

On Thursday, April 30, 2020 at 4:59:28 AM UTC-4, whit3rd wrote:

On Tuesday, April 28, 2020 at 9:46:36 PM UTC-7, upsid...@downunder.com wrote:
On Tue, 28 Apr 2020 16:20:22 -0700 (PDT), Ricky C
gnuarm.deletethisbit@gmail.com> wrote:

The old disk meters actually use two coils, one for current and one for voltage. Their effect on an aluminum disk is proportional to the product

log(U) + log(I) = log(P)

If you need to sum (for averaging) individual power samples, you need
to use antilog( log(P) ) before summing...


Just use abs() (full wave rectification) before the log(),

You'll need to handle negative power, when V and I are opposite in sign. Grab
the signs before taking those absolute values!

A hybrid (keep the wheel) with some turns-counting in silicon is an elegant compromise.

The wheel is easy to tamper with. Just provide a magnetic field and it works like the brake slowing the meter. You can even use metered power to create the magnetic field and have the power company pay for it.

--

Rick C.

+- Get 1,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209