Well, it seems the situation is a LOT better with the parabolic 21dBi antennas:





Since these are identical models and the variances seen in the top trace represent twice their error, these are indeed quite nice even leveled antennas, at least at this 2ft distance apart.  Gotta do more testing @ various other locations and separation distances to verify everything is cool.

--Bart

On 12/11/2012 2:23 AM, Bart Kus wrote:

Hello,

So I tried to measure the reference antennas (simple dipoles).  It's kind of a disaster.  With the antennas only 2ft apart:



The received signal is pretty weak:


The marker is down 20dB from the previous measurement:

I guess that pair of sectors I was using were really helping things out.  I'm also concerned about the uneven response, but that could be multipath given the really low signal levels.  Does anyone have any experience with antenna measurement like this?  I'm thinking my next step is using the 21dBi parabolic dishes as the reference antennas instead.  Having thought about it, the high directional gain, although it would need precise alignment to make accurate measurements, would help with ignoring/overcoming multipath, no?  Some of those peaks and valleys might disappear.  A 20dB range over the spectrum is pretty ridiculous.

--Bart