If uplink power is being tuned by MAC commands you should be able to get those, at least if you collect them historically.
But I’d question the basic assumption: for low power radios, transmit RF power isn’t all that drastic a component of power consumption. Likely what matters a lot more is radio on time - ie, transmitting at SF7 or SF10 even at the same RF power level is going to have a substantially different power cost, because SF10 will take 8 times as long. So SF and amount of traffic are probably a far more useful piece of information.
I’ve got a node on an integrating micro amp meter sitting on an Internet-reachable raspberry pi hiding behind enough urban obstacles that it counts as a distant but usefully workable range. Both the power meter and the node firmware output serial data streams that are combined into one log file you can ssh in and tail. It’s proved extremely informative to be able to watch what various firmware operations “cost” from wakeup to return to sleep, and to do so at a range where slower spreading factors are involved, vs. when the node is sitting a few meters from the gateway in the lab.
Edit: Variation is a little more than I’d expected, but providing you aren’t use the special +20dBm mode nor the very low one, I think transmit time is still the larger factor.
From SX1276 data sheet
- 20 dBm, on PA_BOOST RFOP = 120 mA
- 17 dBm, on PA_BOOST RFOP = 87 mA
- 13 dBm, on RFO_LF/HF pin RFOP = 29 mA
- 7 dBm, on RFO_LF/HF = 20 mA
(assuming decent matching)