On 11/21/2015 4:53 AM, Jim Lux wrote:
On 11/21/15 4:04 AM, Richard (Rick) Karlquist wrote:
Not with just "any network analyzer", like the 8711; you need a 4 port
network analyzer, such at the Agilent E5071C with 4 port option.
Isn't there a process (perhaps tedious and manual) to do this with a 2
port analyzer? Where you connect each pair of two ports in turn (so
there's 6 of them AB, AC, AD, BC,BD, CD) and then in each configuration
connect a short, open, load (or any 2 or 3 known Z) to the other ports.
Then you can solve algebraically for the full 4 port matrix.
Yes, that is correct. Once you calculate
s11,s12,s13,s14,s21,s22,s23,s24,s31,s32,s33,s34,s41,s42,s43 and s44
as you described, you then need to convert them to common
mode and differential mode two port s parameters.
And the "secret" is that most "4 port"
analyzers are just 2 port analyzers with a switch matrix
attached to the 4 external ports to do exactly what you said.
The 4 ports, like the programmable port impedances, are just
a mathematical abstraction. Some of Agilent's competitors
have made a big deal about having a true balanced source,
but I have not heard a convincing argument about why that
is necessary for reasons other than marketing.
Given the number of inexpensive two port analyzers out there these days,
it might be worth figuring out how to do this kind of measurement
Yes, I keep waiting for someone to market a cheap 4 port analyzer,
since "it's just software".
Just out of curiosity, how do you connect the 4 ports to a Balun, which
is sort of a 3 terminal device? Where do the "ground/shield"
connections go?
You should fixture the balun under test on a ground plane with
4 connectors grounded to that plane. The 4 analyzer cables go
to these connectors. You don't connect anything from the DUT
to the ground plane. Alternately, you can do a 3 port measurement
in which case the unbalanced port goes directly to the analyzer.
Again, the Agilent software walks you through the setup possibilities.
Rick N6RK
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|