Question on setting tx-power in ath10k.

Question on setting tx-power in ath10k.

[Ben Greear]

I have added some debugging in my CT firmware based on 10.1.467, and I have noticed
that anything below 5db is treated as zero by the firmware.  This is due primarily to
subtracting a fixed amount based on the NIC's configured chainmask.

I am guessing this is not entirely correct, since at /b rates, for instance, we
will be transmitting on only a single chain, so it does not matter how many
chains the hardware has?

Anyone have any insight on how this stuff is *supposed* to work?

If you have access to firmware source, plz contact me off the mailing list
and I can provide more details about the code that looks broken.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Question on setting tx-power in ath10k.

[Adrian Chadd]

Yeah I remember this stuff. ath9k did the same thing too; it subtracts
TX power based on 2-chain and 3-chain; assuming that the max transmit
power in the EEPROM is actually the 3-chain value so it can get the
per-chain transmit power.

Thing is, whenever I actually looked at the output on a spectrum
analyser, the target power differed from what the card was rated at by
exactly 3 and 5 dBm respectively. Grr.

In FreeBSD I look at the number of configured chains for that TX
attempt and set the TX power cap appropriately per-packet. Grr.


-adrian

On 25 February 2015 at 10:24, Ben Greear <greearb@candelatech.com> wrote:
> I have added some debugging in my CT firmware based on 10.1.467, and I have noticed
> that anything below 5db is treated as zero by the firmware.  This is due primarily to
> subtracting a fixed amount based on the NIC's configured chainmask.
>
> I am guessing this is not entirely correct, since at /b rates, for instance, we
> will be transmitting on only a single chain, so it does not matter how many
> chains the hardware has?
>
> Anyone have any insight on how this stuff is *supposed* to work?
>
> If you have access to firmware source, plz contact me off the mailing list
> and I can provide more details about the code that looks broken.
>
> Thanks,
> Ben
>
> --
> Ben Greear <greearb@candelatech.com>
> Candela Technologies Inc  http://www.candelatech.com
>
>
> _______________________________________________
> ath10k mailing list
> ath10k@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/ath10k

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Question on setting tx-power in ath10k.

[Ben Greear]

On 02/25/2015 11:55 AM, Adrian Chadd wrote:
> Yeah I remember this stuff. ath9k did the same thing too; it subtracts
> TX power based on 2-chain and 3-chain; assuming that the max transmit
> power in the EEPROM is actually the 3-chain value so it can get the
> per-chain transmit power.
> 
> Thing is, whenever I actually looked at the output on a spectrum
> analyser, the target power differed from what the card was rated at by
> exactly 3 and 5 dBm respectively. Grr.
> 
> In FreeBSD I look at the number of configured chains for that TX
> attempt and set the TX power cap appropriately per-packet. Grr.

We have a spectrum analyzer now, but I'm not sure if I can even force
the NIC to run at a specific HT or VHT rate, so not so sure how to
test the changes properly.

Either way, even if I adjust this as I am trying to do now
(on a per-rate basis based on number of chains that rate uses)
the min/max power restraints stay in place, so it should still
be fine as far as regulations are concerned.

Thanks for the info.

Ben

> 
> 
> -adrian
> 
> On 25 February 2015 at 10:24, Ben Greear <greearb@candelatech.com> wrote:
>> I have added some debugging in my CT firmware based on 10.1.467, and I have noticed
>> that anything below 5db is treated as zero by the firmware.  This is due primarily to
>> subtracting a fixed amount based on the NIC's configured chainmask.
>>
>> I am guessing this is not entirely correct, since at /b rates, for instance, we
>> will be transmitting on only a single chain, so it does not matter how many
>> chains the hardware has?
>>
>> Anyone have any insight on how this stuff is *supposed* to work?
>>
>> If you have access to firmware source, plz contact me off the mailing list
>> and I can provide more details about the code that looks broken.
>>
>> Thanks,
>> Ben
>>
>> --
>> Ben Greear <greearb@candelatech.com>
>> Candela Technologies Inc  http://www.candelatech.com
>>
>>
>> _______________________________________________
>> ath10k mailing list
>> ath10k@lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/ath10k
> 


-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Question on setting tx-power in ath10k.

[Ben Greear]

On 02/25/2015 01:15 PM, Ben Greear wrote:
> On 02/25/2015 11:55 AM, Adrian Chadd wrote:
>> Yeah I remember this stuff. ath9k did the same thing too; it subtracts
>> TX power based on 2-chain and 3-chain; assuming that the max transmit
>> power in the EEPROM is actually the 3-chain value so it can get the
>> per-chain transmit power.
>>
>> Thing is, whenever I actually looked at the output on a spectrum
>> analyser, the target power differed from what the card was rated at by
>> exactly 3 and 5 dBm respectively. Grr.
>>
>> In FreeBSD I look at the number of configured chains for that TX
>> attempt and set the TX power cap appropriately per-packet. Grr.
> 
> We have a spectrum analyzer now, but I'm not sure if I can even force
> the NIC to run at a specific HT or VHT rate, so not so sure how to
> test the changes properly.
> 
> Either way, even if I adjust this as I am trying to do now
> (on a per-rate basis based on number of chains that rate uses)
> the min/max power restraints stay in place, so it should still
> be fine as far as regulations are concerned.
> 
> Thanks for the info.

I have uploaded a new beta CT firmware with this change.  If anyone has
time to do some detailed testing on it, that would be welcome.

I've added a release notes text file to the same directory as well, for
those who have asked.

http://www.candelatech.com/downloads/ath10k-fw-beta/

Thanks,
Ben


> 
> Ben
> 
>>
>>
>> -adrian
>>
>> On 25 February 2015 at 10:24, Ben Greear <greearb@candelatech.com> wrote:
>>> I have added some debugging in my CT firmware based on 10.1.467, and I have noticed
>>> that anything below 5db is treated as zero by the firmware.  This is due primarily to
>>> subtracting a fixed amount based on the NIC's configured chainmask.
>>>
>>> I am guessing this is not entirely correct, since at /b rates, for instance, we
>>> will be transmitting on only a single chain, so it does not matter how many
>>> chains the hardware has?
>>>
>>> Anyone have any insight on how this stuff is *supposed* to work?
>>>
>>> If you have access to firmware source, plz contact me off the mailing list
>>> and I can provide more details about the code that looks broken.
>>>
>>> Thanks,
>>> Ben
>>>
>>> --
>>> Ben Greear <greearb@candelatech.com>
>>> Candela Technologies Inc  http://www.candelatech.com
>>>
>>>
>>> _______________________________________________
>>> ath10k mailing list
>>> ath10k@lists.infradead.org
>>> http://lists.infradead.org/mailman/listinfo/ath10k
>>
> 
> 


-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k