FIR Cookbook?
in [DSP]
|
From: Ron N. on 13 Dec 2006 17:02 jeff227 wrote: > >If you are going to calculate on the fly, you were probably want to use a > > >windowing method to calculate the coefficients. This approach is routely > > >discussed in many DSP texts. > > Yes. I have now worked up, with all of your help and inputs, exactly what > I was hoping to find. It is a very simple Windowed Sinc FIR with > coefficients calculated at run time. It maintains the same corner > frequency and bandwidth regardless of sample rate (within Nyquist limits). > It may not be brilliant or optimal but it works well and is simple and > fast to compute. So how did you choose your window? And how did you determine whether the attenuation and ripple met your requirements for your full range of sample rates and cutoff frequencies needed? > The first "trick" to making it scale to any sample rate was to scale the > number of taps. Because nTaps is an integer, the bandwidth and corner > frequency will deviate slightly at certain sample rates. These variations > are small, however, and not a problem in my audio application. > > I prefer this approach to pre-calculating and storing coefficients for the > simple reason that it will accommodate ANY sample rate, not just the common > ones. It also allows the filter cutoff to be moved at run time. (seeing if you did your "exercises for the student" as per the comments :-) -- rhn A.T nicholson d.0.t C-o-M
From: jeff227 on 13 Dec 2006 18:10 >So how did you choose your window? And how did you determine >whether the attenuation and ripple met your requirements for your >full range of sample rates and cutoff frequencies needed? >(seeing if you did your "exercises for the student" as per the comments >:-) Well, I will ALWAYS be a student regardless of how much I learn! I am still fiddling with the Window choice but my requirements are fairly simple - I want a transition bandwidth over about a 1 octave (yes octave, not Hz) range and a flat passband. A 1 octave passband means the corner frequency (in Hz) is equal to the transition bandwidth (in Hz). Stopband attenuation of -40dB is PLENTY for this application so I am using the von Hann window along with the appropriate nTaps. It gives the transition band I want with -40dB to the first lobe in the stopband and uses minimal (not optimal) coefficients (for speed). The passband is very flat up to Fc whereafter it rises slightly then drops off. According to the plots I have created, the bandwidth and corner is staying where it should regardless of sample rate (just like I wanted). But, like I said, I'm still fiddling and still learning. Probably change everything tomorrow! :)
From: jeff227 on 13 Dec 2006 22:57 >A lot of people have that background. Others -- me among >them -- don't. > >In my twisted world, DSP is mathematics applied to solve >real-world problems. Whatever connection there is to >hands-on RLC cirquits is mainly of historical ineterst, >and any analogies quickly become obsolete. Rune, I absolutely agree that DSP, in some cases, is completely removed from analog ideas. HOWEVER, some of my most popular products right now contain VACUUM TUBES. Why? Because DSP has yet to duplicate the sound. It's great to forge new ground but don't forget the past while you're doing it. :)
From: robert bristow-johnson on 13 Dec 2006 23:40 jeff227 wrote: > >A lot of people have that background. Others -- me among > >them -- don't. > > > >In my twisted world, DSP is mathematics applied to solve > >real-world problems. Whatever connection there is to > >hands-on RLC cirquits is mainly of historical ineterst, > >and any analogies quickly become obsolete. > > > Rune, I absolutely agree that DSP, in some cases, is completely removed > from analog ideas. > > HOWEVER, some of my most popular products right now contain VACUUM TUBES. > Why? Because DSP has yet to duplicate the sound. i might agree with that, but point out that products such as the Line6 pod, do a pretty damn good job, if not perfect. there are so many weird things to account for when emulating vacuum tube distortion. one that i didn't know until recently was about the mechanical coupling of the amp bottom (the case with the loudspeakers) to the amp head (the amplifier with the vacuum tube components). the behavior and sound was different if the head was mounted on the bottom than if it was set aside on another pedestal. the vibrations of the loudspeakers were coupled to some small degree to the internal electrodes, the cathode, plate, grid (and screen and supressor) of the tube to give it a time varying behavior that was correlated to the signal. modeling that must be a male offspring of a copulating female canine. also, even if i am alone in this, i see great value to conceptually coupling the concepts of digital filtering to analog filtering. in both we have these memoryless components (or operations) that scale quantities and sum quantities. in both we have components that necessarily have memory that are used to discriminate one frequency from another. in analog filters, these are reactive elements, usually capacitors and their memory is cumulative. in digital filters, these memory components are simply computer memory and they act like delays. so in both cases, we assemble circuits (that is an appropriate word for it in both cases) that scale and add quantities, but such would not discriminate different frequencies. so in both, we add a component that has a concept of time and the past, so we can discriminate frequencies. that is where conceptually these analog and digital filters are comparable.
From: Rune Allnor on 14 Dec 2006 03:07
jeff227 skrev: > >A lot of people have that background. Others -- me among > >them -- don't. > > > >In my twisted world, DSP is mathematics applied to solve > >real-world problems. Whatever connection there is to > >hands-on RLC cirquits is mainly of historical ineterst, > >and any analogies quickly become obsolete. > > > Rune, I absolutely agree that DSP, in some cases, is completely removed > from analog ideas. > > HOWEVER, some of my most popular products right now contain VACUUM TUBES. > Why? Because DSP has yet to duplicate the sound. What do you mean by "most popular"? Where are these products? What are their market shares? People who continued to use horses for years after everybody else turned to motorized help, said something similar; "One can talk with a horse, not with a tractor." To them, the social aspect of being two living creatures working together was more important than power, cost efficiency etc. The fact of the matter is that hifi enthusiasts have hard times ahead. I never considered myself to be one, but after I got my MP3 player, maybe I am becoming one. Even my poor 20-year old stereo rack sounds way better than even a half-decent MP3. Sound quality does not matter these days, as long as it is above some some low minimum. What matters is small size of the player, and easy access to low-price music. Rune |