Re: integrating true RMS wattmeter

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: Jim Lux <jimlux@xxxxxxxxxxxxx>

At 09:48 AM 3/25/2006, you wrote:
> Original poster: dest <dest@xxxxxxxxxxxxx>
>
> Hallo.
>
> i`ve found in archives (97) this thing:
>
> http://himplast.ru/father_dest/temp/killawatt.PNG
>
> the author is David E. Sharpe, TCBOR, this is revision 02, 6/26/97.
> i have a couple of question to you, Dave:
>
> a) have you upgraded the thing to 10-14KW range as was intended in the
> past?
>
> b) is there any more recent revisions here?
>
> c) why did this thing was so nonlinear: -12.66% error at the begin of
> scale and +2.69% error at the end? what was the reason of
> _nonlinearity_ at all - optocoupler or op amp?

Probably non ideal diodes and/or calibration.  These log/antilog 
circuits are tricky to make work over a wide range.  Sure, the 
junction itself behaves as expected, but there's also series 
resistance and offset voltages that vary with temperature, etc.


> d) http://www.pupman.com/listarchives/1997/march/msg00347.html
>
> "Variation in Yokogawa from calculated due to integration times, manual
> data entry, and TCR (temp coeff. of resistance) effects of heater
> element(s)."
>
> what? nippon meter was so SLOW??? what does it mean - "manual data
> entry"?


I think you're misunderstanding what was written.  I think he was 
using a heater as a load, and almost all resistance heaters change 
their resistance as the temperature changes.


> e) "The experimental design was basically use the controller to power
> a 120VAC,1.56kW floor heater.  Inductance is trival (shaded pole
> blower) and 1500W heater."
>
> so you have tested the thing only at 60Hz? and at LINEAR load???
> what`s the point in that


I think that a wise person might do the easy test first: resistive 
loads at low frequency sine wave to verify performance, before moving 
on to reactive loads and non-sinusoidal.


>  - anybody can go and buy $5 ferrodynamic
> wattmeter for example - it would be more accurate and linear in such
> case : ))) my $3 50Hz _mechanical_ meter accuracy rating is 1.5!

But your mechanical meter is exceedingly inconvenient to do things 
like feeding the data in real time to a data logger.

The original article in EDN was an example of a building block 
circuit, as in an application note.  I doubt that it's intended to be 
a finished product in and of itself, but more as a design that might 
provide a starting point for a purpose designed widget. Certainly, 
Dave's test data would be useful in evaluating whether it's 
appropriate for certain loads.


> f) "The differential voltages (Vshunt) and (Vload) are in fact
> MULTIPLIED to the form:
> 39 X Rshunt/Rmultipler * Vload * Iload"
>
> then why do you advice to reduce Rshunt (to 250 uOhm!) instead of
> changing oa gain for example? is it limitation of optocoupler, oa,
> something other? there are many other oa & optos here.

To reduce the power dissipated in the shunt perhaps?



> 10) "The only concern I have with this design is there is minimal circuit
> board clearance through the optocoupler (0.1 and 0.3 inches)."
>
> are you speaking about hv spikes in the mains? so strong, that they can
> cause breakdown of 2.5mm of air? don`t you think that in that case you
> better shut your coil down immediatelly instead of trying to measure
> it`s power consumption? % )

If you want to build something that would meet regulatory 
requirements for isolation (the usual reason for using an opto in the 
first place, as opposed to direct connect, as used in the 
"Kill-A-Watt") there are clearance distance requirements to address 
things like board contamination, etc.



> 11) this circuit is darned simple - maybe we can use a couple of
> additional elements to get real _electronic_ accuracy? : )
>
> -----
> I don't doubt that my logic is flawed, could somebody tell me where?
> 08-08-97 (c) Gary Lau


Certainly. That design is 10 years old, at least.