On Fri,4/29/2016 7:35 AM, jimlux wrote:
I'm building some cookbook choke strings for generalized antenna
testing, from a box of 2.4" 31 cores. The idea is to build up a
generic broadband choke that is used for antenna measurements (doesn't
have to handle any significant power). The antennas are not
necessarily matched (e.g. if you want to measure a dipole cut for 20
meters over the entire 3-30 frequency range at some points, the Z of
the antenna is pretty reactive)
There's a fairly complex tradeoff between number of turns and number
of cores and I'll almost certainly need multiple chokes in series.
It's fairly straight forward to figure out what the series impedance
of the chokes is using K9YC's handy measured data.
I think my real question has to do with "how much impedance is enough"
and "where should that impedance be placed"
That's a real good question, and the answer is exactly as you would
expect, "it depends." Specifically, it depends on the common mode
circuit of the system into which it is inserted when modeled as an
antenna, what other antennas (and other conductors) are nearby, and the
frequency(ies) of interest.
In general, for a transmitting antenna, the first choke should always be
at the feedpoint, or as close thereto as practical. Any length of
feedline between the choke and the feedpoint is part of the antenna. I
recommend and use additional chokes along the line for two reasons.
First, as "egg insulators" to break up the feedline just as we would a
guy wire to prevent interaction with nearby antennas (primarily
verticals). Second, to distribute some of the common mode voltage
between multiple chokes to prevent overheating. VE7RF has suggested
another reason to add a choke near the shack so that the feedline does
not act as an RX antenna for equipment with RFI susceptibility,
especially when that equipment has a Pin One Problem. Jim has a rack
full of audio gear with Pin One Problems, and reports that a good choke
at the shack end of the coax solves it.
I'm not particularly concerned about RFI, for instance, but I am
concerned about coupling to the feedline and any asymmetry in the
system (and the antenna surroundings) perturbing the measurement.
So, from that basis, I would think that you want chokes periodically
along the feedline, so that no piece of the feedline is "significant"
in terms of coupling to the antenna under test.
And, for lower frequencies, the "significant length of a unchoked
piece of wire" in the near field is longer than for higher
frequencies. A 5 meter long conductor near a dipole cut for 10 meters
is a big deal because it's a half wavelength, but probably
insignificant for 80 meters, where's 1/16th wavelength.
So this would imply that "low frequency chokes" (e.g. 7 or 8 turns
through 5 cores) could be farther apart than "high frequency chokes"
(3-4 turns through 5 cores).
Agreed on all counts.
Or, as Jim recommends for a 40-10 meter - 4 turns on 5 cores and 3
turns on 5 cores = this gets you >5k from 7 to 25MHz (fig 46 in the pdf)
Is there an advantage in stacking cores other than ease of
building/mechanical? I would think that 5 turns on 5 cores is about
the same as the series combination of five separate 5 turn on one core
chokes.
The reasons are both mechanical (getting enough turns of the cable
you're using through the core to provide the needed inductance and
capacitance) and for power handling. Inductance and coupled Rs is
approximately proportional to the length of the cable within the core(s)
and, of course, the square of the turns.
The complete equivalent circuit of a NiZn ferrite choke (#43, #61) is Ls
and Rs in parallel with C stray, which within an octave or so of
resonance convolves to a simple parallel resonant circuit. For purposes
of discussion, I call this the "circuit resonance" of the choke. Chokes
on a MnZn ferrite core (#31, #77, #78) have this resonance plus a second
"dimensional" resonance that appears in series with the circuit
resonance. The placement of the dimensional resonance in #31 material
serves to provide a double humped impedance curve below about 5 MHz
that, in effect, increases the effective bandwidth of the choke by
nearly an octave (a 2:1 frequency ratio). The dimensional resonance in
#31 is pretty broad (low Q), while the dimensional resonance in #77 and
#78 is quite narrow (high Q), and occurs in the range of 1-1.5 MHz. This
is clearly shown in the lab data for these cores (the families of curves
for 1-14 turns). As you know, these are measured data for small diameter
wire, not computed.
I suspect you know all this stuff, but I'm repeating it for others
reading the list.
My "rule of thumb" for choking Z in the range of 5K Ohms is based on
nothing more than "brute force" -- rather than solve each application by
modeling the common mode circuit, throw a large enough resistance into
the circuit so that in most practical circuits, common mode current is
reduced enough so that noise is reduced enough and so that dissipation
is reduced enough.
From what I think I understand about the structure of your experiment,
you might also be thinking of shield current as a result imperfections
in the shield coupling differential current to the shield, as described
by the transfer impedance of the coax. That sounds like an interesting
project (and also a non-trivial one). For my work measuring Pin One
Problems and SCIN in audio circuits, I was lucky enough to be living in
a wood frame house, so I was able to string enough audio cable around it
to get useful data, and there wasn't enough AM broadcast RF around to
pollute the data. I DID, however, get enough TV Channel 2 from a
transmitter about 5 miles away to clearly see it. :) The AES papers
documenting that work are all on my website. I did that initial work in
2003, and didn't know nearly as much about ferrites as I did two years
later (2005) when I published the paper on using them to suppress RFI
(and it was then that I received that great family of curves that opened
my eyes to what was going on.
73, Jim K9YC
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|