Semi-offtopic question. I’m using a R11-5HacD card in Linux and noticed that either the ath10k firmware is acting weird or there is some issue with the calibration on the card in relation to power. The card is supposed to top off at 27 dBm and indeed it correctly allows setting power up to 54 (internally atheros uses 0.5 dBm increments), but the output power measured by an analyser doesn’t like up with what I set, whereas the driver firmware doesn’t really disagree with what was being set (columns 2 and 3):
There is a huge linear section of that data (before it caps off at 54 as expected) so I assume there is some kind of calibrated gain value error or something that causes the card to pick a wrong setting for the PA:
I’ve read that the card’s PA is powered by the 3.3aux pins and some adaptors don’t supply those, but it doesn’t seem to be the case for me because it can reliably output up to 20 dBm no problem:
A general observation: all RF amplifiers, built into consumer electronics (which includes WiFi, LTE, 5G, etc.), are cheap ones and consequently lack the desired linearity and overall performance. Also they are not very consistent in performance over a larger batch of chips (even if produced in the production series). The profesional ones (e.g. used in cell towers) are better, but also cost "a fraction" more than the consumer ones. BTW, this largely includes also Rx performance which in consumer electronics is very often mediocre at best.
In WiFi world, APs actually use similar RF PAs as WiFi station devices, so APs don't really work much better than most stations (apart from powering where APs may count on higher available power as they mostly don't run off a tiny battery unlike majority of stations or at least they are not designed to operate in this way).
I used to work for an incumbent MNO in my country and we used to do thorough measurements on every new phone model which we were about to start to sell through our own sales channels. And quite often we found their radio performance underwhelming (to put it mildly). That included both power linearity, response to power control (in mobile networks, it's cell tower which (tightly) controlls Tx power of stations), also perfromance of band-pass filters and side-band spurious power emissions. And we did reject a few models due to poor performance (and some of them actually did sell as per marketing dept. insistance ... making some problems in the network later on and we made very sure that marketing dept. got served with "we told you so" at every opportunity).
