[Lowfer] Re: That filter can't be right!!!

[James Moritz]

At 04:01 05/09/2002 -0400, you wrote:
>These filter design variables are most intriguing! My latest discovery is
>that when the I insert the 12 to 1 loop step-down transformer with coupling
>coefficient of .95 in the Pspice filter model the frequency response gets
>really messed up. Then, in place of the transformer I used a fairly standard
>representation with a leakage inductance of 72 uh in series and the
>reflected 74 ohms in parallel with a 720uh primary inductance and the plot
>came out the same messed up way. Thus, it looks like this extra series
>inductance (the leakage inductance) is the real villain. Maybe I can design
>a filter that assumes this inductance is part of the filter. Ahh. I yern for
>the simpler filterless barefoot approach!
>
>Thanks for all the help!
>
>.....And now to these filterless signal reports <g>
>
>Bill

Dear Bill, Lowfers,

You could

a) add a capacitor in series with the 50ohm side of the transformer to tune 
out the 72uH leakage reactance, or

b) de-tune the loop slightly so that it has capacitive reactance to cancel 
out the leakage inductance - this is probably what you are actually doing 
in reality when you tune the loop for maximum current; or

c) If you know that the antenna + transformer combination actually does 
have 50ohm + j0 impedance when tuned to resonance, just use a transformer 
with ideal unity coupling factor and enormous winding inductance (say 1H) 
for the purposes of simulation.


Whatever you do, don't expect to get a Chebyshev frequency response - as I 
alluded to in the previous e-mail, the source and load impedances 
terminating the filter are an integral part of the filter design - you can 
design a Chebyshev filter with 0, 1, or 50 ohm source impedance and 50ohm 
load impedance, but in reality you have a non-linear source impedance and a 
high Q series resonant load impedance, so what you actually get won't be a 
Chebyshev filter.

Cheers, Jim Moritz
73 de M0BMU