EMRFD Message Archive 7086

Message Date From Subject
7086 2011-12-22 04:54:44 g3wie Equivalent networks for top-coupled circuit filters
Hello everyone,

I'm designing a transverter for the UK/Eu 70MHz band which will be open to all and likely end up as club projects where there will be a shortage of test equipment. Looking at signal frequency filters in the light of the proximity of the 88 - 10MHz broadcast band I'm planning to use triple tuned circuits to the mixed form as per the Ladpac2008 programs. The top coupling capacitors are down in the 1pF region which is unlikely to be repeatable in the expected production environment.

EMRFD gives some information on substituting small value top-coupling capacitors with a T configuration with more tractable values. The text gives a useful equation but then says "The new network will have an equivalent parallel component at each end; you must reduce the capacitance that tunes the resonators accordingly", but doesn't say how to calculate these equivalent values. (It seems to me that in the limit, you could make these equivalent components equal the original resonator tuning Cs, thus reducing component count, but I imagine that there are insufficient degrees of freedom to do this and maintain the required coupling.)

I'm trying to avoid the usual fudges such as omitting the coupling capacitors and placing the resonators "close" to each other for a mix of inductive/capacitve coupling which "works ok". This also won't work in my case as I'm proposing to use -17 grade toroids for the inductors (subject to tolerances but that's another story)

Any help/ideas gratefully received

Chris G3WIE
7095 2011-12-22 12:55:35 in3otd Re: Equivalent networks for top-coupled circuit filters
you can use a wye-delta transformation
<http://en.wikipedia.org/wiki/Y-%CE%94_transform> to convert a pi
7096 2011-12-22 14:04:58 Wes Re: Equivalent networks for top-coupled circuit filters
Hello Chris,

Take a look in the files section at a GIF that I just placed there with the title "bandpass_coupling_stuff.gif". This gives one solution. However, there are many.

Consider an example. We design a filter with one of the many programs and find that we need a coupling cap of Cjk=0.2 pF. We can then pick any equal larger coupling cap above 0.4 pF. Pick Ca=1 pF. Then we find that a shunt cap of 3 pF is needed. We also find that the input capacitance looking into the new network is not 0.2 pF but is 0.8 pF. We must take this into consideration when tuning the resonators.

This procedure has worked well for me up at and around the 2M band. However, lead length is critical here. It should work well if you use SMD parts. Leaded parts that are large, or have long leads is an invitati
7101 2011-12-22 20:38:56 Bill Noyce Re: Equivalent networks for top-coupled circuit filters
Here are the formulas I've used.
C12 = original coupling cap
Calt = larger coupling cap (used twice)
Cpar = shunt cap to ground
Ser = series equivalent

The resonator sees Calt in series with [Cpar in parallel with the other
Calt]. So the equivalent capacitance is
Calt * combo / (Calt + combo)
And combo is simply Calt+Cpar.

Cpar = (Calt/C12 - 2) * Calt
Ser = Calt * (Calt+Cpar) / (Calt+Calt+Cpar)
new Ctune = orig Ctune + C12 - Ser [end sections tuning]
new Cmid = orig Cmid + 2*C12 - 2*Ser [mid section tuning]

I've plugged the resulting filters into AADE Filter Designer and they seem
to work, but I'd love to have someone else validate the formulas...
-- Bill, AB1AV

7102 2011-12-23 02:12:38 g3wie Re: Equivalent networks for top-coupled circuit filters
Thanks everyone for your responses. It shows, as always, that there are multiple solutions to a problem, if only you knew them. After I'd posted my query some faint bells started ringing about how immittance inverters might be used but I'd not had a chance to look that up before your answers appeared (and my brain is too rusty to derive it from first principles!).

Wes's graphic adds an expression for Cin to the information in my EMRFD, and Claudio/Bill yes I see that there is choice enough to pick values to absorb the original tank C. One quick calculation transforms tank parallel C 47pF/series 1.6pF into 50pF series and 1474pF parallel. I see 2 problems with that: a) finding a stable 1474pF capacitor that behaves as a capacitor at 70MHz, and b) if the 50pF is a trimmer because the coils are not trimmable, then neither side is grounded which may make tuning more susceptible to hand effects. However if I absorb only part of the tank C - say 15.5pF - then I get close to 18pF and 180pF for the series/parallel substitutions, with a residual 29pF tank C as the usual grounded trimmer. Now I can finish designing the filters I'll build a few and offer them up to a friend's network analyser for comparison with simulation and to get an idea of repeatability

I see my challenges for the transverter not so much in its design, but in the production engineering to make it repeatable. There have been several previous designs dating from the 80s and 90s whose principles I wouldn't seek to change (eg SBL-1 mixers with 50R terminations on all ports). Unfortunately these designs use now-obsolete Toko coils of various flavours to make construction easier. Winding coils is perceived to be tricky by many newcomers, and the related screening is another constructional complexity for the kitchen-table builder. My design goals are roughly:

1) to make this thing as future proof as possible.
2) to make it buildable by newcomers
3) to make it easy to set up without test gear, and to radiate a clean signal. Apart from repeatable filters, this means some built-in level measurement so you can set the input power at 21/28MHz to a level where the mixer performs well, plus a PIN diode power control *after* the Tx mixer.

One I have it prototyped, I'm going to try and lay out the pcb with combination footprints which will accept through hole and SMD, and also a choice of inductors - air-wound/screening, toko in cans, toroid. Whether those footprints have too many associated strays is something I will have to try out.

I will have more questions and appreciate your combined wisdom.

Unfortunately, word has leaked out into the local 70MHz FM community who want to try ssb/cw and all want to be beta-testers. Now I feel I have to make some progress in between each time I go on the air as I inevitably get asked how it's going. No pressure....

Chris G3WIE