Txpower calibration

[Jim Thompson]

G.W. Haywood wrote:

>Hi there,
>
>On Thu, 2 Feb 2006 19:18:15 +0530 Sukanto wrote:
>
>  
>
>>I am calibrating Netgate minipci card with hostap driver [...]
>>CR31(txpower control register) [...]
>>how this register is indicative of true txpower that is transmiited?
>>    
>>
>
>Short answer - it isn't.
>
>Transmitted power depends very strongly on many more factors than the
>power produced by the RF output device.  For example if you have poor
>impedance matches between the output device and the antenna and/or its
>feeder cable, most of your output power will never reach the ether.
>
>The registers in the card can only be broadly indicative of the output
>power, which for a given set of values will vary _very_ greatly with
>temperature, from one card to another, with degradation of components
>through the action of moisture and chemical processes, with time, with
>power supply characteristics (the output voltages will be affected by
>temperature and current drain, which in turn may change with data rate
>and amplifier gain - and temperature - to name but a few).
>
>See for example http://www.jubileegroup.co.uk/JOS/radio/diurnal.gif
>which shows the approximate signal strength seen by a receiver located
>750m from from an access point which is mounted in a black plastic box
>on a mast outdoors (in England).  The graph covers a period of about a
>month in summer last year.  Temperatures in the box in daytime were up
>to about 35C, down to about 10C at night.  The receiver was indoors at
>an approximately constant temperature.  Signal strengths often varied
>by more than 6dB during days when the sun was shining.
>  
>

Sure, and there are a number of factors here which are likely far more 
responsible for the 6dB
swing than anything you indicate.  These are, primarily:

1) There is likely a lot more "activity" in 2.4GHz during the day than 
during the night.  (People do eventually go to bed.)

2) The noise floor increases during the day both due to #1 above, and 
the simple fact that as the earth and atmosphere warm, the thermal noise 
seen by the receiver must also increase!

In answer to the poster's original question, you have to place the card 
in CTX mode with a Power Meter attached, or your results are essentially 
meaningless.   If you have the power meter, (or spectrum analyzer with 
the requisite attenuation (you really don't want to throw 23dBm at your 
spectrum analyzer... you won't like the results)) then you can measure 
the transmit power.

Its fairly easy to hack up iwpriv to send the right command to the right 
register to go into CTX mode.

(I couldn't load the gif, I got a server unavailable error.)

jim