[Madwifi-users] iwconfig txpower does not seem to change actual power output of 5212 based cards

[Robert J. B. <rb...@ib...>]

<context clarification>
I first want to say that I REALLY appreciate the effort that Sam Leffler an=
d
everyone else has done to make an incredible piece of software. I was able
to install MADWIFI from a snapshot onto a Fedora Core 3 and Mandrake 10.1
system with absolutely no problem and have it working with many different
vendor's equipment.

My quotes of Sam Leffler that follow are not any attempt to cast any
negative aspersions, but to just to show I've at least attempted to researc=
h
this problem and convey my own confusion and lack of knowledge.
</context clarification>

I have now tested 3 different Atheros 5212 based cards with MADWIFI under
Fedora Core 3 Linux.

Senao
Zcommax AG621S
Ubiquiti

I am measuring the output of the cards by having them directly connected to
very cool Agilent 89600 Vector Signal Analyzer that very accurately measure=
s
the average power output of the cards.

When I use the command:

iwconfig ath0 txpower <value>

And vary the value all over the range from 0 - 100 (and also "off") , there
is no change at all in the average power output, though when I check the
values via the iwconfig command, the readout shows the values that I
entered.

I can vary the output levels when using the Atheros "ART" tool, but can not
while using MADWIFI.

I am using MADWIFI:
ath_hal: 0.9.12.14 (AR5210, AR5211, AR5212)
ath_rate_onoe: 1.0
ath_pci: 0.9.4.12 (EXPERIMENTAL)

The chipset signatures are:
ath0: mac 5.6 phy 4.1 radio 3.6
ath0: mac 5.6 phy 4.1 radio 4.6


I have searched the archives of this mailing list and found many people
asking this question with mostly no responses, and a few questions asked
where Sam Leffler answered in a few responses in ways that I lead me to
conclude (but with some doubt & much confusion) that setting txpower will
not effect the actual transmit power output.

The confusing and seemingly conflicting answers were:

> From: Sam Leffler <sam@er...>
>  Re: Changing txpower has no effect. =A0
> 2004-04-23 00:46
>=20
>  On Thursday 22 April 2004 04:46 pm, Kirk R Webb wrote:
>> On Thu, Apr 22, 2004 at 04:11:01PM -0700, Sam Leffler proclaimed:
>>> Note also that "off" does not turn off the radio.
>>=20
>> Is there a corresponding power level it sets the device to then?  Packet=
s
>> still go through which seems counterintuitive.
> =20
>  The per-packet txpower value is used by the h/w to index into a table fr=
om
>  which the h/w settings are fetched to use in sending the frame.  The con=
tents
>  of this table is complicated to explain and not readily accessible.  How=
ever
>  it is supposed to map txpower values in the range [0-63] (6 bits) into
>  equivalent output power for the frame (and yes I did check it"s contents
>  before claiming TPC was "working").
> =20
>  I need to special-case a zero txpower setting to mean "turn off the radi=
o".
>  Alternatively I guess I could simply discard packets in the driver if th=
e
>  power is set to zero but that would be a hack.


This does imply that setting txpower should do something and that at this
point at least, setting txpower to zero would not do what is expected (turn
off the transmitter) but that he planned to make it do that. Yet I never
found a message saying it was ever put in. And lots of other people still
questioning if txpower will control the transmit power.

A little later there is a response that makes me believe that maybe there
still isn't a way to set the transmit power:

> From: Sam Leffler <sam@er...>
>  Re: Txpower problem =A0
> 2004-07-03 23:06
>=20
>  On Jul 3, 2004, at 12:37 PM, Jim Thompson wrote:
> =20
>> I think it is.   Back when I was at Vivato, Tong Chia submitted a HAL
>> change that, IIRC, dealt with a signed/unsigned issue that was
>> directly related to controlling tx power.
> =20
>  Unfortunately that change was just the tip of the iceberg.  We"ve been
>  busy with other issues and not had the time to follow through on
>  completing TPC support.
> =20

Then on the MADWIFI developers mailing list a few  months ago, there is a
pretty definite "no":

> From: Sam Leffler <sam@er...>
>  Re: Question about txpower not changing in 5354  Aries miniPCI card =A0
> 2004-10-07 03:58
>=20
>  There"s no TPC support at the moment.
> =20

So is this a correct conclusion, that at this time there is no ability to
select the transmit power (i.e. To select a particular PDAC table entry in
the hardware)? Or am I missing something that I should be doing.

The reason I need this is I am working with a new piece of hardware
(Ubiquiti high powered card) and the vendor may need some help to make sure
they have their EEPROM tables set up properly to select different power
output levels. At this time I can not seem to control this via the hooks
that I know of in the Wireless Tools / MADWIFI Driver. The Ubiquiti board
does not currently come up in its most powerful transmit power mode and
never reaches it while using the MADWIFI driver.

Any assistance or pointers to more info would be greatly appreciated.

Thanks!
--
Robert J. Berger - Internet Bandwidth Development, LLC.
Voice: 408-882-4755
http://www.ibd.com