AP performance (again): txpower regulation

[Lev Serebryakov]

Hello, Adrian.
You wrote 8 сентября 2011 г., 8:02:28:

> These look fine, but I don't have as much experience with the ar5212
> series stuff as I do with the AR5416+ series stuff.
>>  It seems, registers are Ok. But clients don't see any difference!
>> Maybe, it is because my Senao card is "powerful" one, with additional
>> schematic (and power supply) to boost signal?

> Hm. Well, how far away are they? You can do a bit of math to see what
  About 7 meters and 2 concrete (internal) walls :)

    I've repeated experiment when client and server were within 3 meters
 without walls -- txpower seems to work. 30 gives about -55dB
(accroding to InSSIDer -- I'm not sure, that this numbers could be
used as absolute ones),txpower 15 gives about -65 and txpower 1 about
-75.

  But in any case iperf throughput is about 10-11Mbit/s with pikes up to 18Mbit/s.

  It looks pessimistic: 18Mbit/s  in 2m direct sight. I know, that
 54Mbit/s is half-duplex, but 18Mbit/s is less than 27Mbit/s and
 "ifconfig wlan0 list sta" shows data like this:

 00:18:de:08:e8:1d    1    9  36M 21.5    0  28696  48048 EPS  AQE     RSN WME

  Why 36M in almost-ideal conditions!? Why -70dB in room next to my AP,
when neighborhood AP (simple d-link, according to all data) is -65,
and this room is furthest from other apartments)!?

  Also, graph of signal strength is much smoother on small distances and
didn't look like mountain range.

  Another observation: when distance is about 5 meters txpower 30, 15 and
10 are almost indistinguishable (and grpahs are smooth, -75dB for 10,
-70dB for 15, -67dB for 30), but txpower 1 gives crazy graph: it
osscilate between -80 and -10 (!!!). Sometimes it gold on -77, but
after 10-15 seconds raise to the sky, then drop down again, etc.

> changes the TX power on the sender side would have on the receiver
> signal level.
> Have you ever looked at doing up a free space loss model to predict
> what the signal level(s) should be?
  It is too complex for me :-) And I'm not sure, that I have tool to
measure real signal to compare with calculations.

-- 
// Black Lion AKA Lev Serebryakov <lev at freebsd.org>