I think I'd follow K7ITM's advice.

I'd split the signal right after the IF filters. Feed one of them into
the rest of the Rx chain, and the other one into one of the Analog devices
logarithmic detector chips.

This is assuming there is no AGC prior to the IF filters.

There are two major choices for the detector chip, AD8307, which is a
power detector, or a true RMS detector (AD8326??)

Feed the output into a A/D coinverter, and then digitally add correction
factors for the front end gain and possible different sensitivity on
different bands.

So, once you have the calibration factors, this wouls be a fairly accurate
iindicator of the input signal. Granted, it is assuming a matched 50 ohm
antenna, and you will get an error due to antenna reactances and so on.
However, it is as precise as an amateur radio system is likely to be able to
get and a heck of a lot more meaningful than most 'S' meters I've seen.


Jim
N6BIU