Just to develop this a little further ....
I referred earlier to a spreadsheet I had produced that allows me to
specify the number of turns, the number of stacked cores and the stray C
of a choke using #31, #43, #52 and #61 ferrite mixes; it will then plot
the complex impedance of the choke between 1 and 30MHz. It takes full
account of the variation of u' and u'' with frequency.
This afternoon I took a look at Jim's (K9YC) measured choke data shown
in Fig 41 of his paper. By adjusting the value of C in my spreadsheet I
can get good agreement with any of Jim's curves in that chart. For
example, 7 turns on 5 stacked FT240-31 cores with 8pF parallel
capacitance is predicted by the spreadsheet to give resonance at 3MHz
(Z=7750 Ohms), Z@2MHz=6050 Ohms and Z@10MHz=2600 Ohms; for comparison,
Jim's measured figures were: resonance at 3MHz (Z=8000 Ohms),
Z@2MHz=6500 Ohms and Z@10MHz=2900 Ohms.
Repeating that exercise for different numbers of turns on the same 5
stacked cores required the following values of C for best match with
Jim's measured figures:
4 turns on 5 cores: 2pF
5 turns on 5 cores: 3pF
6 turns on 5 cores: 4pF
7 turns on 5 cores: 8pF
Looking at those figures it seems counter-intuitive that adding an extra
turn going from 6 turns to 7 turns would double the stray capacitance.
It's not a problem with Jim's data - I get exactly the same sort of
anomalies using my own measured data!
Note that in this case the ferrite mix and the stack size didn't change
- only the number of turns. Initial thoughts are that a simple "parallel
C" model is inadequate to describe what is happening and something more
complex is required.
In the case of large conventional loading coils we have to move from a
"lumped component" model to a transmission line analysis to get accurate
predictions; but I wouldn't have thought that necessary with these
chokes that are physically small compared to a wavelength.
Any thoughts?
Steve G3TXQ
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|