Just commited my latest changes on ath_info (madwifi-trace branch).

It took some time and we still have much to do (read per rate target
power levels + ctls) but we are one step closer on implementing
txpower setup. What we have is per channel power calibration data as
seen on eeprom info dumps here ->
http://www.oswave.com/mediawiki/index.php/Eeprom_output

this output helped a lot, you can see there eg. that pwr4 = max_pwr
and also pcdac values for pwr*_x3 are standard.

Easy part was to get frequencies (we already had some code to read
cal_piers, so i ran bin2freq on remaining offsets, doing the same
shifts and i got 8 5Ghz continuous channels (from lower to higher) at
offset 0x150. Then calibration data was just after 5Ghz channel list
(which is 10 channels long), starting as usual with 802.11a data, i
read pwr_1-3 using the same mask /shifts used for the channels and
after some more search i got the trhee pwr*_x3 (were zero in my case
anyway). Then i realized how those pcdac values were stored, after
trying various masks (i was looking for values 1-63 since that's the
range of pcdac values), first i got pcd 2-4 and then i got pcd 1. At
first i thought that i was reading pcds 1-3 but then i realized that i
was doing something wrong since values on dump are increasing while i
got decreasing values. I located pcd 1 just after pwr4_x3 and it was
also bigger than pcd 2. Then i noticed that values on the dump seem to
increase first with big steps  (1->2) and then smoothing. So these
were pcdac steps actually stored on eeprom, adding them together we
get the pcdac table for the channel (that's what we see on the dump)
;-)

Another thing is power units, normaly we have 0.5dbm units but on the
dump we can see that we have greater accuracy, it turns out that
values in eeprom are in 0.25dbm steps and scaled up * 4.


-- 
GPG ID: 0xD21DB2DB
As you read this post global entropy rises. Have Fun ;-)
Nick
_______________________________________________
ath5k-devel mailing list
[email protected]
https://lists.ath5k.org/mailman/listinfo/ath5k-devel

Reply via email to