I have the books by Rick Lyons and Steve Smith. Although they differ in detail of approach, they have much in common: 1. targeted at application oriented reader 2. give *one* simple case of filter 3. use that case as leadin to the next sub-topic author covers I end up missing "something". What that "something" is I don't know. Basically I think I need to tie it into by analog orientation. I'm very sure of that as far as recursive FIR filters are concerned. It's seems a natural for emulating SINGLE stage RLC filters. It would have two personal advantages. It likely would get me past an ill defined vague sticking point. Also I have use for a filter that arises out of a simple resonant circuit. Any pointers to WEB sources? TIA
Alternate approach(s) to introducing FIR and IIR filters
Started by ●July 21, 2008
Reply by ●July 21, 20082008-07-21
On 21 Jul, 15:08, Richard Owlett <rowl...@atlascomm.net> wrote:> I have the books by Rick Lyons and Steve Smith. > Although they differ in detail of approach, they have much in common: > 1. targeted at application oriented readerMaybe you have reached the stage where you are ready for the more advanced stuff? Oppenheim & Schafer's basic text?> 2. give *one* simple case of filter > 3. use that case as leadin to the next sub-topic author coversDon't know about the Smith book, but Rick's book is hardly intended as material for deep insight. It covers the basics in a way that let the 'DSP layman' have a chance to understand something from reading it.> I end up missing "something". What that "something" is I don't know. > Basically I think I need to tie it into by analog orientation. > > I'm very sure of that as far as recursive FIR filters are concerned.FIRs have no analog counterpart. Does that help?> It's seems a natural for emulating SINGLE stage RLC filters. It would > have two personal advantages. It likely would get me past an ill defined > vague sticking point. Also I have use for a filter that arises out of a > simple resonant circuit.RLC cirquits are best emulated as IIR filters, not FIRs.> Any pointers to WEB sources?Sorry, no. Rune
Reply by ●July 21, 20082008-07-21
Rune Allnor wrote:> FIRs have no analog counterpart. Does that help?Admittedly picking nits, but search for "analog transversal filter". Greg
Reply by ●July 21, 20082008-07-21
Rune Allnor wrote:> On 21 Jul, 15:08, Richard Owlett <rowl...@atlascomm.net> wrote: > >>I have the books by Rick Lyons and Steve Smith. >>Although they differ in detail of approach, they have much in common: >>1. targeted at application oriented reader > > > Maybe you have reached the stage where you are ready for the more > advanced stuff? Oppenheim & Schafer's basic text?Actually I thought I was aiming in the other direction. In case of IIR I was thinking along lines of something that would parallel freshman calculus and/or physics coursework.> > >>2. give *one* simple case of filter >>3. use that case as leadin to the next sub-topic author covers > > > Don't know about the Smith book, but Rick's book is hardly intended > as material for deep insight. It covers the basics in a way that > let the 'DSP layman' have a chance to understand something from > reading it. > > >>I end up missing "something". What that "something" is I don't know. >>Basically I think I need to tie it into by analog orientation. >> >>I'm very sure of that as far as recursive FIR filters are concerned. > > > FIRs have no analog counterpart. Does that help?I suspected as much. But would there more examples that would be as intuitive as a moving average filter?> > >>It's seems a natural for emulating SINGLE stage RLC filters. It would >>have two personal advantages. It likely would get me past an ill defined >>vague sticking point. Also I have use for a filter that arises out of a >>simple resonant circuit. > > > RLC cirquits are best emulated as IIR filters, not FIRs. > > >>Any pointers to WEB sources? > > > Sorry, no. > > Rune
Reply by ●July 21, 20082008-07-21
Greg Berchin wrote:> Rune Allnor wrote: > > >>FIRs have no analog counterpart. Does that help? > > > Admittedly picking nits, but search for "analog transversal filter". > > GregThat may may be search term I'm looking for. 10 minutes with Google was promising. Thanks.
Reply by ●July 21, 20082008-07-21
On 21 Jul, 15:51, Greg Berchin <gberc...@sentientscience.com> wrote:> Rune Allnor wrote: > > FIRs have no analog counterpart. Does that help? > > Admittedly picking nits, but search for "analog transversal filter".I did, found only patents, this one at the top: http://www.freepatentsonline.com/5325322.html From the text: "A continuous analog input signal is filtered to produce a discrete time analog output signal" I have no idea what a "discrete time analog" signal might be. Does anybody know what this might be? Rune
Reply by ●July 21, 20082008-07-21
On Mon, 21 Jul 2008 09:10:38 -0500, Richard Owlett wrote:> Rune Allnor wrote: > >> On 21 Jul, 15:08, Richard Owlett <rowl...@atlascomm.net> wrote: >> >>>I have the books by Rick Lyons and Steve Smith. Although they differ in >>>detail of approach, they have much in common: 1. targeted at >>>application oriented reader >> >> >> Maybe you have reached the stage where you are ready for the more >> advanced stuff? Oppenheim & Schafer's basic text? > > Actually I thought I was aiming in the other direction. In case of IIR I > was thinking along lines of something that would parallel freshman > calculus and/or physics coursework. >(snip) I'm not sure what you're looking for when you say that, but O & S is certainly going to give you the rigorous treatment, from which you can derive anything you need. If you're looking for a bazillion examples to try out you may need to just whomp up your own 'homework' problems. This actually has an advantage that you won't get from a book -- book homework problems are rarely malformed, and rarely take bites bigger than you can chew. Your own whomped-together homework problems do, and so help you recognize such in your own work. Or just see if you can do the homework in O & S with the knowledge from Lyons's book. -- Tim Wescott Control systems and communications consulting http://www.wescottdesign.com Need to learn how to apply control theory in your embedded system? "Applied Control Theory for Embedded Systems" by Tim Wescott Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 21, 20082008-07-21
On Mon, 21 Jul 2008 08:08:33 -0500, Richard Owlett wrote:> I have the books by Rick Lyons and Steve Smith. Although they differ in > detail of approach, they have much in common: 1. targeted at application > oriented reader 2. give *one* simple case of filter > 3. use that case as leadin to the next sub-topic author covers > > I end up missing "something". What that "something" is I don't know. > Basically I think I need to tie it into by analog orientation.This can be a bit of a trap. There are a lot of s-domain tricks that apply to the z domain, and a lot of rules of thumb that can be extended from analog signal processing to DSP. But while DSP and continuous-time signal processing both rest on the same bedrock, they're as different as the Statue of Liberty and the rusted-out warehouses on the NY city docks.> I'm very sure of that as far as recursive FIR filters are concerned. > It's seems a natural for emulating SINGLE stage RLC filters.I think you mean IIR? FIR filters aren't normally recursive, and even when they are they are still fundamentally different from IIR systems such as RLC filters.> It would > have two personal advantages. It likely would get me past an ill defined > vague sticking point. Also I have use for a filter that arises out of a > simple resonant circuit.Systems with significant resonances are one of the places that DSP and analog signal processing definitely diverge. All of the 'simple' methods for approximating a filter's continuous-time domain response in the discrete-time domain tend to get less accurate as the poles get more highly resonant and as the poles get faster with respect to the sampling rate; both of these effects can be killers for migrating a resonant filter design from continuous-time to discrete time. Having said that, there are people here (Hey Jerry!) who are comfortable with such approximate techniques. I'm not -- when I need to implement a highly resonant filter I do one of two things: I either specify the filter's performance in the discrete-time domain and design the thing directly, or I modulate the signal down to baseband (using I & Q demodulation), filter it there, then either extract the information it carries or modulate it back up to its original frequency if that's what's called for.> Any pointers to WEB sources? > TIAAlas, no. But if you find some interesting ones please post links here! Ask for comments and, even though we're a pretty taciturn group when it comes to giving our opinions _someone_ may say something interesting. -- Tim Wescott Control systems and communications consulting http://www.wescottdesign.com Need to learn how to apply control theory in your embedded system? "Applied Control Theory for Embedded Systems" by Tim Wescott Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 21, 20082008-07-21
On Jul 21, 11:18�am, Rune Allnor <all...@tele.ntnu.no> wrote:> I have no idea what a "discrete time analog" signal might be. > Does anybody know what this might be?A continuous-amplitude discrete-time signal? Even that doesn't describe what I mean. I did a quick search on "analog transversal filter" and, like you, found a lot of patents. There looked to be some papers deeper in the search, though I did not have time to dig very far. But in the good old days they used to build filters out of analog delay lines. They were true analog filters -- continuous time and continuous amplitude -- but they suffered from aliasing just like discrete-time filters. A while back, someone here on comp.dsp mentioned that SAW filters also qualify, but I am not terribly familiar with them. Greg
Reply by ●July 21, 20082008-07-21
Rune Allnor wrote:> On 21 Jul, 15:51, Greg Berchin <gberc...@sentientscience.com> wrote: >> Rune Allnor wrote: >>> FIRs have no analog counterpart. Does that help? >> Admittedly picking nits, but search for "analog transversal filter". > > I did, found only patents, this one at the top: > > http://www.freepatentsonline.com/5325322.html > > From the text: > > "A continuous analog input signal is filtered to produce a discrete > time analog output signal" > > I have no idea what a "discrete time analog" signal might be. > Does anybody know what this might be?Sampled (in time) but not digitized is my bet. Jerry Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
