On 1/30/2011 1:23 PM, Roger (Sub1) wrote:
> I would think those things would heat if tapped with the unused turns
> shorted, the wrong mix, insufficient core material, or a mounting that
> would prevent adequate cooling.
I've been studying ferrites for nearly seven years, most of it looking
at suppression applications, but in doing so, I've learned quite a bit
about ferrites in general. In general, we want ferrites to have LOW
LOSS (high Q) when using them in resonant circuits in transmitters, but
we want them to have HIGH LOSS (low Q) when using them for suppression.
And even when using them for suppression, we try to make the choke
self-resonant at the frequency (ies) where we want suppression (the
equivalent circuit of a ferrite choke is a parallel resonant circuit).
I recently did some experiments with bifilar common mode chokes wound on
LOSSY ferrite cores, tuned (by number of turns) to cover the HF bands.
#31 Fair-Rite material is optimum for that, and yields circuit Qs of 0.2
- 0.5, so the resonance is VERY broad. I built a choke with 16 bifilar
turns of #12 THHN, connected as a parallel wire transmission line, added
SO239 connectors at each end, and patched it in line with the output of
my Titan 425 (between the amp and the antenna tuner). In this
configuration, the core sees only leakage flux from the differential
signal, with no common mode voltage. I then transmitted key-down for
several minutes, then with key-up, immediately felt the choke for
heating. What I observed is VERY interesting.
1) There was virtually NO heating of the ferrite core (a 2.4-in o.d.
toroid, the form factor called FT-240 so that Amidon and others can
charge 4X their cost when they resell it to hams).
2) There WAS modest temperature rise (I'd guess no more than 20-30
degrees F) in the wire where it was wound around the ferrite core.
3) There was NO temperature rise in the leads (1-2 inches) going to the
SO239s.
4) Doing the math, 1,5kW in 50 ohms is 5.5A at 274 volts, so it's easy
to understand why there's no observed heating from copper losses.
5) Because dissipation is small, LEAKAGE FLUX is small.
CONCLUSION: Dissipation occurs IN THE COPPER as a direct result of
losses COUPLED from the core to the copper, but it does NOT occur IN THE
CORE ITSELF!
Note that this experiment ONLY looks at dissipation as a result of FLUX
in the core, and the experiment does not place RF voltage across the choke.
How are things different in a transmitter's resonant circuits? First,
we want to minimize loss, so we (should) use low loss cores (like
Fair-Rite #61 or #67). These are NiZn cores, and have much lower mu than
those designed for RFI suppression (like #31 and #43).
Inductors wound on #61 or #67 can have Qs on the order of 10 or more,
and are typically self-resonant a bit above HF. These resonances CANNOT
be accurately measured on reflection-based analyzers (that is, S11).
Rather, we need to measure S21, with the inductor (or choke) wired as
the series element of a divider.
For more on this, see applications notes and tutorials at
http://audiosystemsgroup.com/publish.htm
Note also that my work has focused on ferrite materials, NOT powdered iron.
73, Jim Brown K9YC
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|