"top c coupled" BPF filter design theory

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

jhaible

Well-known member
Joined
Jun 24, 2004
Messages
530
Location
Germany
Hi,

I'm looking for ... well, what the title says.

It's an LC filter topology where you have several grounded parallel LC combinations, whiche are connected on top with capacitors from stage to stage.

I'm looking for a good online article that explains the design, with equations and all.

Please let me know if you have something. Seems that this type is rarely covered in LC filter books. (?)

Thanks,

JH.
 
[quote author="Samuel Groner"]Did you check Zverev, Handbook Of Filter Synthesis?

Samuel[/quote]

No - is it online somewhere?

JH.
 
[quote author="jhaible"][quote author="Samuel Groner"]Did you check Zverev, Handbook Of Filter Synthesis?

Samuel[/quote]

No - is it online somewhere?

JH.[/quote]

Unlikely as it is still in print I think. At least there is a softcover version now, still pricey though.
 
[quote author="Samuel Groner"]Every major technical library shoud have a copy. Not sure if it covers the network you're looking for, but Zverev is still the reference for passive filters I'd say.

Samuel[/quote]

I looked it up on amazon - impressive table of contents for sure!

I'm tempted to buy it, for the whole of it that's inside. And yet, at the moment, I'd love to get just a short (few pages) introduction into top c coupled filters that I can quickly download somewhere ...

JH.
 
j, I think I know what you're envisioning but perhaps a fragment of a schematic would make it certain.

In a glance through my copy of Zverev I didn't see tables etc. for that specific topology (as I understand it from your description).
 
a classic HPF for an RF system, especially those that digitally tune using upconverters.

Check most passive RF literature. Other than that I would simulate since you seem to know what you want already.
 
[quote author="bcarso"]j, I think I know what you're envisioning but perhaps a fragment of a schematic would make it certain.

In a glance through my copy of Zverev I didn't see tables etc. for that specific topology (as I understand it from your description).[/quote]


Something like this (for a 6-pole BPF):

top_c.jpg


I'm not fomliliar with this topology.
I know two similar circuits for BPF:
(1) An inductor in series with C12, and C23, C3, L3 omitted
(that would be BPF transformation on a standard LPF chain),
or
(2) C12 and C23 replaced with resistors
(That is what Bob Moog used on his Fixed Filter Banks - it's easy
to calculate f0 and Q for each partial filter there!)

So in my limited experience with LC filters, I see the Top C Coupled topology as an alternative to (2), but less noisy (not as many resistors), and probably not as easy to calculate (with 8 instead of 6 reactive elements, something in the equations has to cancel in order to result in a 6-pole filter, no?)

JH.
 
PS: I first came across this topology in the "Super Filter" design software we have at work. The program suggests using 3 identical L values even for a staggered tuning (probably taking advantage of the 2 extra degrees of freedom that come with the two extra capacitors, see above) - only then to reject any input with an error message "detecting an asymmetrical s plane" - whatever this means.

JH.
 
[quote author="jhaible"] with 8 instead of 6 reactive elements, something in the equations has to cancel in order to result in a 6-pole filter, no?
[/quote]

Yes it does. You have two capacitive loops in system (closed paths
consisting of only capacitors and voltage generators). If you have capacitive
loop, one of capacitor voltages is lineary dependable of other capacitor
voltages, thus system order is decreased by one. If simulator is not
smart enough it will go beserk when it trys to solve eight order system with
only six independent state variables. Inductive equvalent is when
you have a junction point which conects only inductors and curent sources.
Try conecting very small resistors in series with C12 and C23, that should
help. Why do you need this filter?

cheerz
urosh
 
[quote author="recnsci"][quote author="jhaible"] with 8 instead of 6 reactive elements, something in the equations has to cancel in order to result in a 6-pole filter, no?
[/quote]

Yes it does. You have two capacitive loops in system (closed paths
consisting of only capacitors and voltage generators). If you have capacitive
loop, one of capacitor voltages is lineary dependable of other capacitor
voltages, thus system order is decreased by one.
[/quote]

That makes sense! Thanks for pointing it out.

Try conecting very small resistors in series with C12 and C23, that should
help.

Simulating it in Spice would be no problem. But that would mean trial and error, not design.

Why do you need this filter?
cheerz
urosh

I don't really need it. I just ran across it and want to learn.

Ok, and there _is_ this idea of building a vocoder with completely passive filters. (;->)
Having an LC filter topology that allows the use of 3 identical inductors for a staggered-tuned BPF is intriguing ...

JH.
 
Ha! It's a frigging approximation!

When you design a BPF in top c coupling topology, you don't get the same overall frequency response as with an "ordinary" BPF filter. It's the same around the passband , but the slopes are different. They are shiftet to the right, due to the high pass nature of the C coupling.
(So maybe not everything cancels with these additional reactive parts ...)

No wonder the Super Filter program denied service when I defined the filter via poles and zeros.

I guess when I just go via filter type (Bessel, Cheby, ...) and Bandwidth, it will work allright.

JH. (just learned something.)
 

Latest posts

Back
Top