Talk:Comb filter

From Wikipedia, the free encyclopedia

	This article is within the scope of the Professional sound production WikiProject, a collaborative effort to improve Wikipedia's coverage of the technology, equipment, companies and professions related to professional sound production. If you would like to participate, you can visit the project page, where you can join the project and see a list of open tasks.
B	This article has been rated as B-Class on the assessment scale.

The spectra of a frequency modulation signal will have equally spaced spectral lines which decrease in amplitude farther away from the center frequency. Communications systems take advantage of this effect by transmitting FM signals with center frequencies that would normally interfere with each other, but are arranged so that their spectral lines interleave. A comb filter can then select the signal of interest. This technique was used to expand monaural FM broadcast radio and monochrome video to stereo and color, by interleaving the extra spectral information. See subcarrier.

(I'm not sure if this is true, so I'm not adding it yet.) - Omegatron 19:42, Feb 26, 2005 (UTC)

• Stanford.edu page talking about feedforward and feedback comb filters

Contents 1 So it's named for the spectrograph it produces.. 2 "Similar to ring modulation"? 3 What if you do stuff to the delayed signal? 4 Half finished?

[edit] So it's named for the spectrograph it produces..

but nobody's found a graphic to illustrate the effect yet? I'm hardly a math geek so I won't try, but it would be nice to see what this means in practice.

[edit] "Similar to ring modulation"?

As far as I'm aware, a comb filter has nothing in common with ring modulation. A comb filter is just a particular discrete-time linear filter, whereas ring modulation involves multiplication, i.e. non-linear. Hence ring modulation will create intermodulation distortion. Oli Filth

[edit] What if you do stuff to the delayed signal?

In my understanding, if you add a delayed signal to the original signal, that's a comb filter. Fine, but what if you do something to the delayed copy? If you invert it, it's obviously still a comb filter, but what if you EQ the delayed signal as well? Is it a comb filter, or is it just kind of like a comb filter? I ask because I'm writing a plugin for a music program that does this sort of thing, and I don't know whether to describe it as a comb filter or kind of like a comb filter. - furrykef (Talk at me) 05:14, 5 January 2006 (UTC)

You might end up with a patent violation. Sounds a bit like Karplus-Strong_string_synthesis, if you feed it with a signal and leave it comb. But sounds pretty neat. Good luck.

A comb filter is just a special filter structure. If you alter the response of the delay line, then you're back to any old filter; the adjective "special" no longer applies, so it's not a comb filter anymore. IMO. Oli Filth 02:04, 19 June 2006 (UTC)

[edit] Half finished?

This article like many Wikipedia seems to be pretty good technically at least in some ways but there seems to be a lot missing. What this article really needs is a proper TV engineer to write at least a section for it (I could be persuaded to do this but I would have to do a lot of research first and I haven’t touched the subject for about ten years). - But there needs to be a section about comb filters in television systems and other analogue electronics.

Comb filters are not specific to TVs; they're used all over the place in signal processing applications. For instance, in interpolation and decimation filters, there is a class of filter known as cascaded integrator comb (CIC), which uses a comb filter.

Anyway, in analogue TV the comb filter isn’t a small peripheral component of the system, its actually pretty much the heart of the whole decoder. (As I remember) It is the comb filter that breaks the signal up into components and it does a kind of magic that allows a very simple circuit to do a very complex task. In fact I seem to remember that it is use of comb filters that made the whole analogue TV system possible. Even monochrome TVs need a comb filter though it is what is called a simple comb - a single resistor.

A resistor does not perform any filtering on its own.

I also suspect that the description of comb filters based and focused on delays is only partially correct or complete, it is more of a ‘simplified’ 'scientific' rather than an engineers explanation, has a lot of math, and may actually confuse people. Yes comb filters can or do use delays but the essence of the filter is that it is recursive (a repeated element) - another name for them I have heard is ‘recursive filters‘. Also I don’t think the shape of the waveform is the place the filter got its name from, its circuit diagram actually looks like a comb or matrix made of resistors, but I think the name came from a more abstract level - the filter orders in a certain way, it was originally conceived as like a comb - it just happens that the waveform looks the same because that is what it does.

No, a comb filter is just one type of filter. It can be feedforward or feedback (recursive). It relies on delays to operate (as do all filters). Almost all digital filters are derived from "repeated elements"; see the image at digital filter, for instance. The comb filter is so called due to the shape of its frequency response.

The delay based version of the circuit is only one type and there are others - the TV circuit for instance is purely driven by phase. In fact I seem to remember reading somewhere that comb filters are considered a ‘universal’ type (this means that combs can be made to perform almost any function and also that almost any given circuit type can be made to work as a comb). This was probably from an early book on cybernetics, or feedback, computation or AI (my own main subject area).

Phase-shift and delay are essentially the same thing. Oli Filth 11:47, 21 October 2006 (UTC)

The final thing not mentioned is the inventor of the comb filter, who I think may be someone like Norbert Werner or one of his friends, but this is purely a guess.

Lucien86 06:36, 21 October 2006 (UTC)

Incidentally, I do agree that the article is severely lacking. There are no diagrams of implementations in terms of delay lines, no pole-zero diagrams, and the frequency response diagrams are unhelpful, at best. The derivation of the frequency response is made more complicated than it needs to be by working with e^jωn rather than z. Oli Filth 11:47, 21 October 2006 (UTC)

I was bored, so I've completely re-written the article to include all this stuff. Oli Filth 16:39, 21 October 2006 (UTC)

What does alpha represent? "Scale factor" does not mean anything to me. 71.168.90.28 (talk) 23:26, 9 June 2008 (UTC)