On 4/16/2012 11:15 AM, Steve Hunt wrote:
> Here's one commercial balun manufacturer that_does_ publish useful
> performance data:
> http://www.balundesigns.com/images/Model%201171%20Choking.jpg
It's good to see that. One concern and one quibble. First, the quibble
-- I have a strong preference for data like this to be plotted with a
log frequency axis. Second, the concern. I have a strong mistrust of
impedance data obtained by reflection-based methods (i.e., S11). I don't
know the AIM unit well enough to know how this measurement may have been
taken, but from the work I've done, some form of S21 measurement is
needed to get good data. I've addressed this in the tutorial.
3-4 years ago, several of us did a "round-robin" measurement of an
unknown choke (as I recall, 14 turns on a 2.4-in toroid that I suspect
was #61 or something like it) by a variety of methods. The least
accurate measurements were done with an AIM unit, and the data were
WILDLY inaccurate. The best agreement was between my simple
generator-voltmeter method and by someone using a Q-meter.
I just looked at a photo of the product in question. Choking impedance
looks quite good, but I'd be surprised by this data if the core is
Fair-Rite #31. Nothing wrong with that if they've got a better
material. I AM concerned with a 5kW power rating for a choke with this
graph of choking Z vs frequency. While it would probably do fine with
1.5kW .in a well balanced system, I would expect it to fry pretty
quickly with much imbalance and the "wrong" length of transmission line.
It's nothing more complicated that E squared / R, where E is the common
mode voltage across the choke and R is the parallel equivalent R in the
equivalent circuit of the choke. G3TXQ, W8JI, and I have had some
pretty spirited discussions on this matter, and there's a discussion of
this both in the RFI tutorial. and the Chokes Power Point.
73, Jim Brown K9YC
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|