[ previous ] [ next ] [ threads ]
 From:  Jim Thompson <jim at netgate dot com>
 To:  Michael Mee <mm2001 at pobox dot com>
 Cc:  Douglas Stringer <dougstringer at mac dot com>, Justin Wilson <j2sw at mtin dot net>, m0n0wall at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall] Wireless Questions
 Date:  Sun, 13 Nov 2005 13:48:39 -1000
Michael Mee wrote:

> Douglas Stringer wrote:
> > On Nov 11, 2005, at 11:00 PM, Mike Mee wrote:
> >> I've done this many times and it does work quite well (though you lose
> >> approx 1/2 the bandwidth due to the radios overloading each other,
> >> even if they're on channels 1 and 11..)
>> What do you mean by "radios overloading each other." Could you be  
>> more specific?
> The radio frequencies are close enough for the broadcast on one to 
> effect the receive on the other. This affects throughput.
> The only real solution is two radios mounted 2-3 meters apart, 
> including antennas (Or notch filters tuned to the specific channel 
> (expensive) which still won't help much if the miniPCI cards are 
> physically proximate).

2-3 meters isn't enough if the directivity of the antenna (gain) is 
large and/or pointed toward each other (or a common third point). 

I've never tested the 2511MP cards in a lab, but they don't 'leak' that 
much, the biggest problem is probably injection locking via the LO if 
the radios are operated co-channel (or close).  Ths is something we 
battled a lot on the first-gen Vivato
switch, which uses the Agere (Lucent/Orinico) chipset, a (ahem) highly 
similar design.

If you can take care of the other issues (to your satisfaction, anyway), 
then running the two radios on ch1 and ch11 should be fine in terms of 
intermod and injection locking (inside the case).   If you care about 
performance (range, throughput, whatever), then two radios (and 
antennas) nearby each other in the same band "doesn't work".

For instance, the little "bridge/router" I did for Vivato uses two 
2511MP (see previous message) cards on a little MIPS-powered board.   
The applications for this are essoteric at best, but it does pass FCC 
(and wouldn't with
any real amount of intermod.)   We did have to put some little ferrite 
coil "chokes" on the pigtails through, as the FCC wants to arrange the 
antennas in the "worst" orientation (this ends up being with them 
alongside each other at a distance of about 1mm on that product), and 
the chokes were required to pass FCC in this configuration.

> This is a well recognised though often ignored (and often debated) 
> effect. Its easy enough to verify however!

Yeah, lets not go there.   I've been int too many on-line fights where 
my simple math gets labeled "theory", and the other side is finally 
forced to admit that "it works" really translates to "I get some packets 
through under some circumstances".

I'd rather not make a (bigger) ass of myself here.