In fact, in your 15 minute measurement frequency, 3.34 million cycles have passed.
Nope. 15 minutes = 900 seconds. 900 s x 60 Hz = 54,000 full cycles. You multiplied by 60 too many times.
26 channels, each with MHz frequency, 24 bit resolution
Again, incorrect use of multiplication.
You are clearly not measuring 26 input channels with one 24 bit ADC at 6ns intervals (167 MSPS). As mentioned above there is no 24-bit ADC that can handle that. You may be doing low Mhz rates per parameter, but that wouldn't be 6ns intervals. And you want to get the most of the 24-bits, so I wouldn't think you'd want to try multiplex the ADC inputs that fast. Even at 4096 samples per half cycle (<0.5 MSPS) it is most likely using a separate ADC per measured parameter.
Your own white paper states:
The technology is sampling/deriving 26 parameters using current and voltage samples in 24 bit resolution at MHz sampling rates on each phase, neutral and ground, through a precision, software controlled oversampling methodology.
So with 3 phases you are really only measuring current and voltage on 4 or 5 circuits, which is 8 or 10 total measured parameters. Most of your "26 channels" are actually
derived values (presumably all the standard things that can be derived from V, I, & t). Sure those extra derived parameters may be calculated and processed in real time, but that has nothing to do with your 24-bit sample rate.
In addition your white paper also states it is
oversampling. I hope you understand what that actually means, considering how much flack you are giving the power generation industry use of "guesses based on averages and integrals".
This kind of sales talk (obfuscation) you keep spouting is why the engineers here keep saying your statements and replies are nonsense.