[NoCatNet] Testing the separation of rcv/xmt radio cards

[Matt Townsend]

This afternoon Curt, Max, and myself went up to the Grandview site to try 
an experiment where we would try to put a little distance between the two 
radio cards. For this Curt brought his DLink AP with the intent of using it 
in place of the Sending Senao card in the Soekris box.

First, we verified the existing situation:
- Speed test, using Sonic FTP, of the incoming signal from Gold Ridge. 3 to 
3.5 mbps. This was measured connected to the hub going to the house.
- Speed test, of the outgoing signal as presented at the NoCat-GV antenna. 
1.2 to 1.5 mbps.

With the AP about 4 feet from the Soekris and the Senao card unplugged:

-Connecting the AP to the hub yielded speeds of 3 to 3.5mbps.
- Connecting the AP from the hub to the GV sector antenna yielded speeds of 
3 to 3.5mbps

With the AP next to the Soekris and the Senao Card unplugged:
- The output from the Sector ant drop to 2.3-25mbps.

All tests above were repeated several times so to avoid the Reverse Polish 
Luck Syndrome (RPLS.)

We conclude from above that placing the two radio cards next to each other 
is the cause the speed drop in our relay nodes.

We then asked Roger Weeks in Graton if he could connect using the new GV 
signal and he could not. This probably means that, in his case, he requires 
more output power (the current card is a 23dB Senao) than the AP (probably 
35mw) is putting out. We had intended to leave the setup as is but, based 
on Roger's results, rebuilt the node to its original configuration.

Max and Curt may have more to add. We still haven't addressed the science 
behind why these cards interfere with each other when in close proximity 
but it seems every one of our relay nodes exhibit this problem to some 
degree. But we do have a solution to the problem.

Matt