Talk:Infinite impulse response: Difference between revisions

 

Line 8: Line 8:

Is this for real? [[User:Mark Richards|Mark Richards]] 22:18, 1 Apr 2004 (UTC)

Is this for real? [[User:Mark Richards|Mark Richards]] 22:18, 1 Apr 2004 (UTC)

: Sure. Why are you in doubt? — [[User:Pjacobi|Pjacobi]] 16:47, 5 Aug 2004 (UTC)

: Sure. Why are you in doubt? — [[User:Pjacobi|Pjacobi]] 16:47, 5 Aug 2004 (UTC)

:: See: [[Zeros and poles]]. In the [[Infinite_impulse_response#Transfer_function_derivation|transfer function]] section there is ‘H(z)’, if the zero occurs in the divisor (lower part), it is pole. Feel free to add a reference there.[[Special:Contributions/~2025-33113-31|~2025-33113-31]] ([[User talk:~2025-33113-31|talk]]) 20:48, 26 November 2025 (UTC)

:: See: [[Zeros and poles]]. In the [[Infinite_impulse_response#Transfer_function_derivation|transfer function]] section there is ‘H(z)’, if the zero occurs in the divisor (lower part), it is pole. Feel free to add a reference .[[Special:Contributions/~2025-33113-31|~2025-33113-31]] ([[User talk:~2025-33113-31|talk]]) 20:48, 26 November 2025 (UTC)

[[Pole_(complex_analysis)]] [[User:Gah4|Gah4]] ([[User talk:Gah4|talk]]) 22:14, 3 July 2017 (UTC)

[[Pole_(complex_analysis)]] [[User:Gah4|Gah4]] ([[User talk:Gah4|talk]]) 22:14, 3 July 2017 (UTC)

Could an explanation of the term “pole” as used in the context of IIR be added to Wilkipedia? Existing definitions of pole don’t help.

Is this for real? Mark Richards 22:18, 1 Apr 2004 (UTC)

Sure. Why are you in doubt? — Pjacobi 16:47, 5 Aug 2004 (UTC)

See: Zeros and poles. In the transfer function section there is ‘H(z)’, if the zero occurs in the divisor (lower part), it is a pole. Feel free to add a reference; the zero and pole page could be updated. ~2025-33113-31 (talk) 20:48, 26 November 2025 (UTC)[reply]

Pole_(complex_analysis) Gah4 (talk) 22:14, 3 July 2017 (UTC)[reply]

It’s stated in the article that: “This is in contrast to a finite impulse response in which the impulse response h(t) does become exactly zero at times t > T for some finite T”. I think this is not the correct definition, because we can have a filter with impulse response that comes from -inf and stops at zero. Even that T = 0 < +inf, it’s an IIR filter. — Preceding unsigned comment added by 85.68.1.212 (talk) 18:11, 24 November 2013 (UTC)[reply]

I’ve deleted the Truncated IIR filter section. The author had confused the concept of the infinitely long output of IIR filters with some concern that the filter representation itself was somehow infinite and thus needed to be clipped. To which I say:

filter[0] = 10; for (tt=1; tt>0; tt++) { filter[tt] = 0.5 * filter[tt-1]};

There, that wasn’t so bad was it? Fits on one line.

To confuse matters, google does show there is an obscure filter somebody has wickedly (or cleverly) called a “Truncated IIR filter”, but it actually is an FIR filter. If someone wants to write a separate wikipedia article on that filter (along with the text explaining why it isn’t really an IIR filter, and what purpose the filter serves, etc.) that would of course be great & we could add a “See Also” link to it, but it doesn’t merit its own section in the main IIR article. technopilgrim 21:47, 13 Apr 2005 (UTC)

stumbled across this article by accident and found it very hard to find out what it is about. could someone mention some type of context at the beginning? (as in “a filter used in electronics” or so.) —Helge.at 09:34, 15 September 2005 (UTC)[reply]

• FIR is a finer point of digital signal processing. Therefore, a general introduction, like it is used in electronics, should stay in the DSP page. Never the less, I tried to clarify the introduction. Faust o 12:56, 5 March 2006 (UTC)[reply]

Its kinda hard to understand when you dont know about electronics. I wanted to find out about FIR and IIR to understand the “actual” difference between “normal” and linear phase EQ; and I’m lost. How about FIR and IIS for non-scientific recording engineers 🙂

The FIR has a linear response for in-band. Linear in-band signal will be replicated, but delayed. The IIR uses prior results and also some similar math to the FIR. The prior result makes frequencies attenuate, but it may not have linear phase. It will change the timing of different frequencies. So a high pitch sound/tone may arrive at a non-synchronized time to a lower pitch as compared to the original. ~2025-33113-31 (talk) 20:45, 26 November 2025 (UTC)[reply]

In the introduction:
“The simplest analog IIR filter is an RC filter made up of a single resistor (R) feeding into a node shared with a single capacitor (C). This filter has an exponential impulse response characterized by an RC time constant.”

All textbooks I know only consider digital filters and not analogue filters in their definition of an “Infinite Impulse Response Filter”. Im not suggesting analogue filters do not have an infinite impulse response. All analogue filters have an infinite impulse response. Rather I’m suggesting analogue filters are not Infinite Impulse Response (IIR) Filters, as this term to my knowledge usally descibes a digital filter.

Of couse the reason I’m writing this is to provoke discussion, so please comment.

There is nothing wrong with calling analogue filters IIR, as they are indeed IIR. (Actually, FIR analogue filters exist; they may be implemented using tapped delay lines – they are discrete-time, but analogue nonetheless.) Oli Filth 08:27, 12 December 2006 (UTC)[reply]

The article suggests that recursive digital filters are inherently IIR; this is not the case. “Recursive” is an implementation, which produces poles in the frequency response, and usually (but not always) results in IIR behaviour.

A simple counterexample is the following recursive difference equation:

  y [ n ] = y [ n − 1 ] + x [ n ] − x [ n − 2 ] {\displaystyle \ y[n]=y[n-1]+x[n]-x[n-2]}

which can be easily transformed to become:

  y [ n ] = x [ n ] + x [ n − 1 ] {\displaystyle \ y[n]=x[n]+x[n-1]}

which is clearly FIR in behaviour.

The article should be updated to reflect this. What’s more, recursive filter should not redirect here; a separate article will need to be created. Oli Filth 08:16, 12 December 2006 (UTC)[reply]

I have now converted the recursive filter redirect into a stub article. Oli Filth 20:05, 25 December 2006 (UTC)[reply]

seems to be missing! Openlander 00:52, 25 January 2007 (UTC)[reply]

This topic is a tough one for people who haven’t attended courses on the subject. It would be good to link to a page or add a section or external links pointing to online resources for determining filter coefficients, especially java sites. Most of the math involved intimidates engineers that have claimed to have done this in school. Thanks. Hansschulze (talk) 22:46, 15 July 2009 (UTC)[reply]

“Clearly, if a_j ≠0 then the poles are not located at the origin of the z-plane.” –(true for j=0, not all j’s)

“This is in contrast to the FIR filter where all poles are located at the origin” –(there is no pole in FIR. It’s confusing to say “poles are all at zero”. or someone correct me.) — Preceding unsigned comment added by Jimaxtwiki (talk • contribs) 17:06, 21 February 2015 (UTC)[reply]

That equation in the Stability section should have a 1 on the left hand side, not 0, right? Alan Bloom (talk) 12:06, 9 June 2025 (UTC)[reply]

Zero is correct. If the denominator is zero, then the response is infinite. That is what you expect at a pole. Constant314 (talk) 12:58, 9 June 2025 (UTC)[reply]