[CUWiN] good news everyone

[Jim Thompson]

Bill Comisky wrote:
> On 8/5/06, Aaron Kaplan <aaron at lo-res.org> wrote:
>>
>>
>> > hard disk, like a PVR.   How complicated is that?
>> >
>> (...) i did not mean this question :)
>>
>> > For the 1 thousandth time, we need a directional rooftop
>> > device, of which, each house needs two or more, in
>> > order to construct our urban mesh WHOEVER we want.
>> >
>> oops!
>> yes, i agree. But I think I have to clarify what I acutally meant:
>>
>> _given_ that many freifunk/funkfeuer/cuwin meshes are like they are now,
>> and _given_ that our projects attract media attention and therefore - as
>> it happend in VIE now - the attention of the (local) FCC
>> ("funkueberwachung", "regulierungsbehoerde" in germany) and _given_ that
>> - as you are all freaks - you installed high gain antennas.
>> THEN not everything is lost. You can still use these nice 24dBi high
>> gain antennas on the manner i described before.
>
> The FCC is a more lenient for fixed point-to-point installations;
> see Title 47, part 15, section 247 of the Code of Federal Regulations:
> http://tinyurl.com/oa83g
>
> Note also the allowed EIRP by the FCC is a lot higher than 100mW..
> 1000mW with a 6dB antenna for multipoint links == 4000mW EIRP.
>
> See below for a quick summary of the regulations from Tim Pozar
> http://www.lns.com/papers/part15/Regulations_Affecting_802_11.pdf
>
(quote deleted)

All true, but not sufficient for FCC testing.   The gain of an antenna 
also tends to incrase the emissions in the so-called "restricted bands", 
and emissions here are constrained to abide by *strict* limits, (as 
opposed to the relative limits used elsewhere in 47 CFR 15.247).

This doesn't affect using a 'high gain' antenna on receive, unless there 
is sufficient e-coupling that the emissions from the 'high gain' antenna 
are over these limits.

The 'switching in software' is more easily accomplished on Atheros-based 
designs, where each packet descriptor can contain the antenna that is to 
be used.  You can therefore set 'ant 1' as the RX antenna (by default) 
and 'ant 2' as the TX antenna (in each packet for transmission).  

Given a specialized card, you can have up to 8 such "transmit antennas" 
(6 in the 1st generation Atheros chipsets), making such a 'sectorized 
AP' straight-forward to build.

The scheme above (ant 1 for rx, ant 2 for tx) was used in the Vivato 
"802.11g" long-range device.  If you look at the Vivato filings, there 
is a large bandpass (cavity) filter which is used to notch the outgoing 
signal such that one it is "amplified" by the (large) gain of the 
antenna, it is still in-compliance with the FCC regs.

There were plans (before I left) to construct an "omni" (mesh) device 
out of essentially the same parts (sans the beamformer) such that we 
could rx (and sometimes tx) from (to) up to 6 different directions 
(simultaneously), but only transmit out any given antenna if the other 5 
were not already receiving.

If you don't constrain a multiple radio device to *not* transmit while 
any of its other radios is receiving, you get negative results.

jim