Measuring real power consumption

[hlavac]

How do I go about measuring real current consumption of a circuit?

If I just sample the current shunt, I will miss some spikes... I'd probably need ridiculously high sample rates...
I was thinking, could I somehow run the current sensed thru an integrator and sample that at regular intervals?
Like, measure the real amount charge that has passed thru the circuit, sample that at regular intervals and get the real average current by dividing the charge by the time span instead of averaging the samples of immediate current?

Maybe two integrators so I can do double buffering and not miss part of the waveform.

Like, have two caps each charged with a current that is proportional to the current consumed by the measured device and tracks it very fast.
I could then sample one of the caps then discharge it, and next time sample then discharge the other cap... and so on
Similarly I'd have to integrate the voltage over the same period to be able to calculate power...

Could this work?

[jimmc]

If you are talking about a fixed voltage DC power source, you can average the current before sampling and then perform the integration digitally.
If you want the integrated power (=energy) then, since the voltage is fixed, multiplying the voltage by the integrated current will give the total energy in Joules (Watt seconds)

A simple RC lowpass filter should suffice for the averaging.


However if the voltage varies with time (as well as the current) then you would need to take the product of instantaneous voltage and current before averaging (and then integrating).
This would require an analogue multiplier before the lowpass filter.

Jim

[digsys]

What are you considering "fast" sampling?? 1Ksps, 100Ksps? Maybe you look at True RMS converter
chips, have them do the dirty work and read them at a "reasonable" read rate. Some of these
have other maths functions as well. IF you want fast and accurate, you won't do better and easier.