The issue with the sampling rate comes in when you have an odd and fast current load. One thing I want to point out here is that for a battery, even though the voltage is mostly a constant 12 volt and the current is "direct" i.e., DC, the current can vary a lot and very fast with time. I think incandescent lights bulbs are about the only load on the 12 volt line that has a constant current. Just about everything has current transients, some vary faster than others.
Ill use the inverter example again.. say I’m using a 500 watt inverter to power some low power AC load
What exactly the AC load is doesn’t matter but let’s say its 20 watts. This is low power for the inverter so the inverter would tend to run at a low duty cycle. It might tend to pulse the 12 volt current for maybe 1 msec every 8.3 msec (assuming a 60 hz AC cycle). If I’m sampling this at 100 times per second, I’m taking a sample every 10 msec. Very often I would completely miss sampling the current pulse since it only last 1 msec and would underestimate the current by a large percent.
And 20 watts is not at all in the noise - that is 1.66 amps at 12 volts which for must of us is significant.
The A/D we recently used in a work project that is sampling at 62Khz would have sampled the above example 62 times during the 1 msec current pulse. This part costs $1.65 in a quantity of one, way less in volume. Even a part that samples 1/10 of that speed is still going to measure the 1 msec current pulse 6 times for each pulse. Note that on top of sampling, you also have to do processing and on my project, we are still doing a lot of processing in between A/D samples – not rocket science.
Because it’s not difficult or expensive to sample at a much higher rate than 100 hz even for a few year old design, I would expect much higher sample rates. But am guessing..