[OLSR-users] OLSRd configuration with multiple channels

[John Gorkos]

> Cool.  One more question on this: I plan to implement that like this:
>
> Card1--"out1"--+                     +--"out1"--Antenna1
>                +-single LMR400 cable-+
> Card2--"out2"--+                     +--"out2"--Antenna2
>
> Sorry about the poor ascii art :( .  In that graph there are two
> splitters, one acting as a mixer, and the other splitting.  Will that
> work? (I have never done anything like that).  Each of the antennas will
> be about 120º apart (they are 14º panels).  There may be a third antenna
> for "local" coverage.  well, I'll not redact all of that again, here is
> the question I asked in another group (still no answer from them, and I
> want to double check this).
You will wind up with two dead radios within seconds of applying power.
The signal will follow the path of least resistance, which will basically come 
out of one radio, bypass the splitter, and fire straight into your other 
radio.  Replace LMR-cable with steel tubing and Card1 and Card2 with Gun1 and 
Gun2, and pull the trigger.  The bullet will not go where you want it to.  On 
the other side, you don't want to use a splitter like that for your antennas, 
either.  The antenna's are designed to provide a 50 ohm impedance to the 
transmitter.  the splitter will monkey up that impedance and cause your SWR 
to rise at your transmitter.  There are ways to do impedance matching with a 
splitter, but they are beyond the scope of this discussion.. :)


>
> Antenna1 will be "aiming" to one node that is located at one moutain,
> say, 150º east of true north, and the other to another node that is
> located at another moutain, say, 274º east of true north. Each
> antenna will be a 14dBi pannel.
>
> At the "repeater" nodes I'll have the same configuration, two antennas
> (in that case two 14dBi 90º Horizontal polarization sector
> antennas,90º one from the other), two wireless card, "mixed" into a
> single cable, then splitted into two antennas.
>
1 and 4 are barely separated enough for co-located antennas.  try 1-8 and 
4-11, but even then you could be in trouble.
> What I'm assuming is that, provided that the channels are suficently
> separated (say, channel 1 and channel 4) I'll have the output from the
> two cards into the two antennas, and I'll have the reception from any
> antenna into the two cards. Is this correct?. What I want to do is
> to use one channel for "upstream" and the other for "downstram".
if you are doing things that require that much bandwidth and separation, you 
need to get out of 2.4 GHz and use either 5.8 or licensed freqs.
>
> Also, what would happend If I have, say, one panel and one omni
> antenna (the omni in the "back" and over the omni, at least one feet
> above (30cm)), I assume that they will see each other, but.... will
> there be any "bizzare" interference like a feedback, something like a
> "reflection". The panel and the omni will be horizontally polarized
>
> Well, the omni will be there only if I can build those slotted
> waveguide omnis :-S (I want to build the 8 slots per side version,
> wich I think that will have enough vertical beam width (can anybody
> confirm this?)), does anybody knows another horizontal or circular
> polarized omni?, with at least 7dBi gain and a beamwith of at least
> 15º (adding the up and down, ie, 7.5º up, 7.5º down, something like that).
>
ok, I give up.  you are trying to do some bizarre stuff here with your radios.  
Slotted waveguide antennas are simple to design and mercilessly tricky to 
build.  your best bet is to use a CNC milling machine.  a dremel tool and a 
yardstick will yield nothing but frustration.
The bottom line is OLSR will handle your routing, with the appropriate 
configuration.  I'm perhaps too dense to understand the network you are 
trying to physically construct.  Perhaps your best bet would be to install a 
minimal network, and add some of the more robust features once you have the 
basic connections talking.

John Gorkos
Wildcat Wireless Internet