[TowerTalk] Balun for Force 12 6 Element 20 Meter Monobander

[Jim Brown]

Hi Jim,

Choking impedance is driven by both noise and power handling. The 
fundamental principle on which my designs are based is that the choke 
should have a high value of RESISTIVE impedance at the frequency(ies) of 
interest. Reactive impedance can be cancelled by the common mode 
impedance of the feedline.

Because the choke is resistive, it dissipates power due to the common 
mode current. And because power is proportional to the resistance but to 
the square of voltage or current, dissipation falls twice as fast as 
choking Z. So Z must be high enough to minimize the dissipation. That's 
why some of my designs for higher power consist of two chokes in series.

I have learned, through extensive work in the lab, that Fair-Rite #31 is 
the only ferrite material suitable for use at HF. The reason is, that 
because it is a very special MnZn mix, the dimensional resonance 
component of its impedance combines with the circuit resonance to make 
the resonance much broader than with NiZn materials like #43, #52, and 
#61. It is the higher Q resonance of the NiZn, combined with the wide 
tolerance of ferrite parts, that every single choke must be measured to 
find it's resonance, AND that the resonance is too narrow to cover 
adjacent bands. In other words, the wide tolerance of the ferrite 
material moves the resonance too much for a cookbook like mine to be 
published.

73, Jim K9YC

On 7/21/2020 3:30 PM, jimlux wrote:
> And that leads to a "choking impedance should be around 5k" which is 100x.
> 
> But if it's 4k, it's still good. If it's 3k, it's still good.
> 
> I think the value of Jim's cookbook is not so much the choking 
> impedance, per-se, but that he's
> 
> a) bounded the core variability (designed for worst observed case 
> performance of the core)
> b) also taken into account side effects like resonance and leakage 
> capacitance
> 
> The latter is MUCH less affected by core properties (I think, Jim can 
> correct me if I'm wrong).