[Top] [All Lists]

Re: [Amps] ferrite in RF chokes for PAs

To: "" <>
Subject: Re: [Amps] ferrite in RF chokes for PAs
From: "Jim Brown" <>
Date: Thu, 06 Mar 2008 09:40:53 -0800
List-post: <">>
Lots of good advice in this post.

See comments interspersed, lots of snips to save space in the digest.

On Thu,  6 Mar 2008 08:29:02 -0700, John Lyles wrote:

>Plate RF choke has an appreciable DC current component in it. This 
>lowers the effective mu of the ferrite, and makes it less effective 
>as a choke. 

YES!  The good news is that this can be predicted from mfr data sheets. 
The bad news is that you probably won't like the prediction. More bad 
news -- the DC will also cause heating, very possibly overheating. When 
that happens you have no mu at all (until it cools). 

>Ferrite-loaded inductors still do have parasitic resonances due to the
>stray capacitance from wire turns. 

And also capacitive coupling from turn to turn and end to end through 
the ferrite, which is also a dielectric. 

>They tend to shift down in frequency so they still have to be taken 
>care of that they are not excited at the normal operating frequencies.
>There is no free lunch here. Less turns of wire, for sure, but also 
>the resonances are lower due to the higher permeability in the coil's

The measured data in my tutorial for coils of 1-14 turns on toroidal 
cores of five different materials is a good illustration of this. The 
tutorial also includes a discussion of how DC affects the behavior of 
ferrite chokes and how to use the mfr's data to compute it.

The equivalent circuit of a wire or a coil of wire passing through a 
ferrite core, or wound around a ferrite rod, is a parallel resonant 
circuit. The R, L, and C of that resonance will appear in parallel with 
the rest of the circuit, and do exactly what you would expect from R, 
L, and C in parallel. The R will burn some power and lower the Q, while 
the L and C will do some resonating. 

The values of R, L, and C are not easy to measure, primarily because R 
is usually very large and C is usually quite small (often much less 
than the stray of the test setup). The tutorial shows one way to do it. 


Jim Brown K9YC

Amps mailing list

<Prev in Thread] Current Thread [Next in Thread>