[TowerTalk] To get a truly balanced antenna feed

[jimlux]

On 6/24/16 12:48 PM, Richard (Rick) Karlquist wrote:
>
>
> On 6/24/2016 8:08 AM, jimlux wrote:
>>
>> Or they use multiport VNAs and get to the result by algebra.  If you
>> look at the measured test data for a minicircuits RF transformer, you'll
>> see full 4 or 5 port data.
>>
>> http://www.minicircuits.com/MCLStore/Agreement.jsp?file=ADT1-6T_S5P.zip
>>
>
> I used to design network analyzers for Agilent (now Keysight).
> A 4 port VNA is the gold standard for dealing with any sort of
> balanced system.  I had the use of an E5071C 4 port at work
> and was able to make very definitive measurements of CAT5
> twisted pair with it.
>
> A 4 port VNA actually consists of:
>
> 1.  A 2 port VNA with one source and one or two receivers.
>
> 2.  A switch matrix to connect two of the VNA ports
> to various combinations of two of the four DUT ports
>
> 3.  A bunch of math to calculate common mode, differential,
> and mixed S-parameters.  Also, a calibration routine.
>
> It does NOT use any baluns whatsoever.  It never generates
> pure balanced or pure common mode drive signals nor does
> it have the corresponding receivers.
>
> It is therefore possible to fake a 4 port VNA using only
> a 2 port VNA if you do the various interconnections manually
> and have a math routine to crunch the data.  However, the
> 2 port VNA needs to be a true 2 port VNA.  Many so-called
> 2 port VNA's should probably be called 1.5 port VNA's because
> they don't have full 2 port calibration.
>
> This is way beyond the capability of virtually all hams because
> of the cost of the 4 port VNA (5 figures) and the complexity
> of understanding how to use it.  Let alone faking it
> with a 2 port VNA.
>
> Rick N6RK


ya know, though..

for HF, a box full of suitable relays could make the test set for the 
underlying 2 channel VNA.  I think that's within the scope of an 
ambitious amateur. I'm not so foolish to think that one can calibrate 
out everything from crummy relays, but there's probably some middle 
ground that's realizable.

The math is probably the sticky part.  I started building a 4 port unit 
for phased arrays using the tentec VNA and some surplus coax relays, but 
the integration of the VNA software with the (relatively simple) 
software to switch the relays with the (not so simple) post processing 
software got to be overly complex.   The VNA software for most ham VNAs 
didn't lend itself to being "part" of an overall measurement system, so 
there was a lot of trying to fool it with reloading cal files for each 
relay configuration.

I suspect that had I spend some substantial time on it, I could have 
done all the cal outside the VNA software, but that was also tricky.

OTOH, measuring mutual Z among phased array elements is probably less 
challenging for a VNA than characterizing a balanced transformer. 
There's a reason they charge what they do for N-port analyzers.