From: Jeroen Belleman
Subject: Re: Wideband 50:200 ohm, 100W, 30MHz-1GHz balun design?
Date: Fri, 22 Nov 2002 14:52:17 +0100
References: <3DDCBD57.898BFFF1@cern.ch> <firstname.lastname@example.org>
X-Mailer: Mozilla 4.7 [en] (X11; I; Linux 2.2.12-20 i686)
"Tony (remove \"_\" from email address)" wrote:
> Another thought - use a single ferrite on the bottom 2 lines
> together, and nothing on the top 2 (as they don't need any
If you don't need the secondary side to float, then yes,
that should work.
> Many thanks for taking the trouble. I can see how that works.
> Would you hazard a guess as to the co-ax length and what
> type/size of ferrite to try?
I'd keep the coax as short as possible, or the parasitic
capacitance of the coax screens which carry an RF voltage
will certainly affect the response. About a decimeter
should be acceptable. Shorter is better.
The main problem will be the connections at the ends,
because at 1GHz, a mere centimeter of straight conductor
is a sizable impedance. (Far too high in this application,
As for the ferrites, a few uH of inductance is plenty to
meet the 30MHz lower cut-off requirement. Let's assume
you use straight bits of coax passing through one or more
ferrite rings. At 100W, you'll have 70Vrms between the
primary and secondary ends of the bottom coax lines.
More or less arbitrarily, we'll pick Brms=0.1T as an
acceptable flux density, so that an effective surface
area of about 4e-6 m^2 of ferrite is enough. Choosing
a core big enough to fit over RG58, say, a Philips TN10/6/4,
with a central hole of 5.5mm diameter, just one core is
enough to handle the power, and choosing 3C11 material,
it will have enough inductance as well. Higher permeability
is fine too. Don't apply power at too much below the design
frequency, or the ferrite will get hot.
The above is quite a mouthful. In practice, I'd simply
slip three or four high permeability rings over the coax
and see if it's enough. No need to shave off every penny
if you make only one or two.