posted on Apr, 6 2017 @ 07:11 AM
originally posted by: Pilgrum
I was constructing probes and collecting data in the lab for quite a while but we found no real evidence particularly on analog meters (disk types)
and engineers' worries were alleviated at that time.
Analog meters are almost as good as calorimetric types such as the one you were mentioning. Essentially, their effective sample rate is infinite. So
they get the energy right, despite the waveforms being naff.
The problem comes with digitizing meters and how they decide to deal with non-sinusoidal waveforms. The generating side voltage source is very low
impedance and generally dominates anything on the house side. However, the house current can be quite non-sinusoidal. Especially as you mention with
switching power supplies. The generic non-PFC ones just hose the lines into a capacitor with a diode bridge. As such, they don't draw current at all
until you get near the top of the voltage sinusoid, then abruptly toss on a capacitative load. And most of the ones in the house will switch on at
about that point in the phase, so they aggregate. That's also true with loads that are being phase controlled, which we still do over here but which I
believe you guys forbid.
This provides the meter with a very non-sinusoidal load, when you add in all the bigger appliances that now do this. I have a washer and a dishwasher
that have stepper motors in, and they have switchers in the front end, as does the microwave et al, none of which are PFC controlled.
In a switcher with PFC, not only is the power factor set right, the abrupt turn on you get with a bridge driving a cap is eliminated. Basically, in
order to fix the PFC, the bridge is eliminated and the front end cap is powered with a boost switcher or Cuk that only stops drawing current within a
few volts of the zero cross for voltage. This gets the current much more sinusoidal and nearly but not quite aligned with the voltage.
A good many of the RMS algorithms are totally fuddled by the current draw of a bridge -> capacitor up front in a switching front end. It's a nearly
uncalculable crest factor. So they take a guess and try to correct the actual metering by fudging in a power factor correction, but that is also
really difficult to determine with any accuracy if the power is pulsed like non-PFC switchers do. The more appliances that do this, the worse accuracy
you end up with.
An analog meter just sort of sails serenely over all this and gets it right nearly every time by integrating the pulsed current with the inertia of
the drive train.