"Steve Underwood" <steveu@dis.org> wrote in message news:d3kogu$5q1$1@home.itg.ti.com...> Jon Harris wrote: > > More generally, any kind of filtering low in the band tends to be a lot > nastier than similar filtering high in the band. Intuitively, what else > would you expect? It takes more samples of the low frequencies to see a > similar difference in the signal. > > Regards, > SteveSteve, I'm not following the "nastier" and "it takes more samples of the low frequencies to see a similar difference in the signal" in relation to filter design. But I can imagine a framework where it would: If the reference is in absolute terms then I don't see how it matters. If the reference is in percentage of the absolute frequency at the transition then I can see it. For example, if the transition has to be 1Hz, then it is no more demanding at 10Hz than it is at 1,000Hz. However if the transition has to be 1% of the frequency where it occurs then that would be 0.1Hz at 10Hz and 10Hz at 1,000Hz and that's obviously a big difference. Normally I'm thinking of design of digital filter bands expressed in absolute frequency terms and not in percent. Maybe that's too constrained a viewpoint? Fred
FIR notch filters using the Windowing method
Started by ●April 13, 2005
Reply by ●April 14, 20052005-04-14
Reply by ●April 14, 20052005-04-14
I agree with you but having the offending interferer within my band of interest, using an IIR filter would wreak havoc on the measurements I am attempting to make. Particularly since the measurements are based on the signal envelope and well as the carrier after filtering.
Reply by ●April 14, 20052005-04-14
Fred Marshall wrote:> "Steve Underwood" <steveu@dis.org> wrote in message > news:d3kogu$5q1$1@home.itg.ti.com... > >>Jon Harris wrote: >> >>More generally, any kind of filtering low in the band tends to be a lot >>nastier than similar filtering high in the band. Intuitively, what else >>would you expect? It takes more samples of the low frequencies to see a >>similar difference in the signal. >> >>Regards, >>Steve > > > Steve, > > I'm not following the "nastier" and "it takes more samples of the low > frequencies to see a > similar difference in the signal" in relation to filter design. But I can > imagine a framework where it would: > > If the reference is in absolute terms then I don't see how it matters. If > the reference is in percentage of the absolute frequency at the transition > then I can see it. > For example, if the transition has to be 1Hz, then it is no more demanding > at 10Hz than it is at 1,000Hz. However if the transition has to be 1% of > the frequency where it occurs then that would be 0.1Hz at 10Hz and 10Hz at > 1,000Hz and that's obviously a big difference. > > Normally I'm thinking of design of digital filter bands expressed in > absolute frequency terms and not in percent. Maybe that's too constrained a > viewpoint?I've long (a little longer than I've been hanging around here) taken it for granted that in order to affect a frequency significantly, a filter's significant impulse response needs to last in the order of that frequencies reciprocal. (The realization was thrust upon me when I tried to design a Hilbert transformer for an 22 kHz sample rate with the best low-frequency response possible. After a while it occurred to me that a 90-degree phase shift at 50 Hz takes exactly 5 ms no matter what the sampling frequency. Duh!) A good rule for estimating is (maybe half) the the longest significant period and the sampling period related to the steepest edge. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●April 14, 20052005-04-14
"Fred Marshall" <fmarshallx@remove_the_x.acm.org> wrote in message news:gcSdnadGNomCDMPfRVn-2g@centurytel.net...> > "Steve Underwood" <steveu@dis.org> wrote in message > news:d3kogu$5q1$1@home.itg.ti.com... > > Jon Harris wrote: > > > > More generally, any kind of filtering low in the band tends to be a lot > > nastier than similar filtering high in the band. Intuitively, what else > > would you expect? It takes more samples of the low frequencies to see a > > similar difference in the signal. > > > > Regards, > > Steve > > Steve, > > I'm not following the "nastier" and "it takes more samples of the low > frequencies to see a > similar difference in the signal" in relation to filter design. But I can > imagine a framework where it would: > > If the reference is in absolute terms then I don't see how it matters. If > the reference is in percentage of the absolute frequency at the transition > then I can see it. > For example, if the transition has to be 1Hz, then it is no more demanding > at 10Hz than it is at 1,000Hz. However if the transition has to be 1% of > the frequency where it occurs then that would be 0.1Hz at 10Hz and 10Hz at > 1,000Hz and that's obviously a big difference. > > Normally I'm thinking of design of digital filter bands expressed in > absolute frequency terms and not in percent. Maybe that's too constrained a > viewpoint?What you say makes sense, Fred. However, in audio, filter bands are almost always expressed in terms relative to their center or cut-off frequency, e.g. in octaves. Even in other applications, we tend to _implicitly_ think in these terms as well. Consider, for example, a "steep" IIR (or analog) filter. Generally this thought of in terms of dB/octave (or decade), say 48dB/octave is a steep filter and 6dB/octave is a gentle filter. This is true regardless of whether the filter cut-off frequency is 1Hz or 1000Hz. So, in my experience, filter bands in absolute frequency terms are a rarity.
Reply by ●April 14, 20052005-04-14
"Jon Harris" <jon_harrisTIGER@hotmail.com> wrote in message news:115u1p2nl0drb8@corp.supernews.com...> "Fred Marshall" <fmarshallx@remove_the_x.acm.org> wrote in message > news:gcSdnadGNomCDMPfRVn-2g@centurytel.net... >> >> "Steve Underwood" <steveu@dis.org> wrote in message >> news:d3kogu$5q1$1@home.itg.ti.com... >> > Jon Harris wrote: >> > >> > More generally, any kind of filtering low in the band tends to be a lot >> > nastier than similar filtering high in the band. Intuitively, what else >> > would you expect? It takes more samples of the low frequencies to see a >> > similar difference in the signal. >> > >> > Regards, >> > Steve >> >> Steve, >> >> I'm not following the "nastier" and "it takes more samples of the low >> frequencies to see a >> similar difference in the signal" in relation to filter design. But I >> can >> imagine a framework where it would: >> >> If the reference is in absolute terms then I don't see how it matters. >> If >> the reference is in percentage of the absolute frequency at the >> transition >> then I can see it. >> For example, if the transition has to be 1Hz, then it is no more >> demanding >> at 10Hz than it is at 1,000Hz. However if the transition has to be 1% of >> the frequency where it occurs then that would be 0.1Hz at 10Hz and 10Hz >> at >> 1,000Hz and that's obviously a big difference. >> >> Normally I'm thinking of design of digital filter bands expressed in >> absolute frequency terms and not in percent. Maybe that's too >> constrained a >> viewpoint? > > What you say makes sense, Fred. However, in audio, filter bands are > almost > always expressed in terms relative to their center or cut-off frequency, > e.g. in > octaves. Even in other applications, we tend to _implicitly_ think in > these > terms as well. Consider, for example, a "steep" IIR (or analog) filter. > Generally this thought of in terms of dB/octave (or decade), say > 48dB/octave is > a steep filter and 6dB/octave is a gentle filter. This is true regardless > of > whether the filter cut-off frequency is 1Hz or 1000Hz. So, in my > experience, > filter bands in absolute frequency terms are a rarity.We were talking about digital filters in the context of windowing design - meaning FIR filters. I'm used to using something like P-M for FIR filters where the band edges are expressed in absolute terms. But, if one is used to using analog prototype designs and bilinear transformation and IIR filters (i.e. with poles), then it may be natural to think in those other terms. Yep. Fred
Reply by ●April 14, 20052005-04-14
Jerry Avins wrote:> I've long (a little longer than I've been hanging around here) taken it > for granted that in order to affect a frequency significantly, a > filter's significant impulse response needs to last in the order of that > frequencies reciprocal.How about h[0] = 1, h[1] = -1? That has rather profound effects at low frequencies and is kinda short. :-) Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein
Reply by ●April 15, 20052005-04-15
"Greer" <Greer.GRAY@oxinst.co.uk> wrote in message news:q5WdnZYwX6dsysDfRVn-sw@giganews.com...> Hi, > > I am trying to obtain a FIR notch filter which is very narrow. For > example, a 50 Hz notch filter to remove mains noise from a 1250Hz signal > say.You can make a notch filter directly from a symmetric window: 1) Negate all the coefficients and adjust the center coefficient so that the sum of all coefficients is 0. This will make a notch at DC. 2) cosine-modulate with the desired center frequency. The longer the window you use, the narrower the notch will be. -- Matt
Reply by ●April 15, 20052005-04-15
On 13 Apr 2005 14:35:34 -0700, "JS" <jshima@timing.com> wrote:> But I also need to use a FIR filter for linear >phase reasons.So, to follow through, why is this important? Our hearing and all natural world resonances are *not* linear phase. Where should design goals point? We are the product of billions of years of design in a different direction. Chris Hornbeck 6x9=42 April 29
Reply by ●April 15, 20052005-04-15
Hi All, Many thanks for all your input. Regards, Greer This message was sent using the Comp.DSP web interface on www.DSPRelated.com
Reply by ●April 15, 20052005-04-15
Fred Marshall wrote:> "Jon Harris" <jon_harrisTIGER@hotmail.com> wrote in message > news:115u1p2nl0drb8@corp.supernews.com... > >>"Fred Marshall" <fmarshallx@remove_the_x.acm.org> wrote in message >>news:gcSdnadGNomCDMPfRVn-2g@centurytel.net... >> >>>"Steve Underwood" <steveu@dis.org> wrote in message >>>news:d3kogu$5q1$1@home.itg.ti.com... >>> >>>>Jon Harris wrote: >>>> >>>>More generally, any kind of filtering low in the band tends to be a lot >>>>nastier than similar filtering high in the band. Intuitively, what else >>>>would you expect? It takes more samples of the low frequencies to see a >>>>similar difference in the signal. >>>> >>>>Regards, >>>>Steve >>> >>>Steve, >>> >>>I'm not following the "nastier" and "it takes more samples of the low >>>frequencies to see a >>>similar difference in the signal" in relation to filter design. But I >>>can >>>imagine a framework where it would: >>> >>>If the reference is in absolute terms then I don't see how it matters. >>>If >>>the reference is in percentage of the absolute frequency at the >>>transition >>>then I can see it. >>>For example, if the transition has to be 1Hz, then it is no more >>>demanding >>>at 10Hz than it is at 1,000Hz. However if the transition has to be 1% of >>>the frequency where it occurs then that would be 0.1Hz at 10Hz and 10Hz >>>at >>>1,000Hz and that's obviously a big difference. >>> >>>Normally I'm thinking of design of digital filter bands expressed in >>>absolute frequency terms and not in percent. Maybe that's too >>>constrained a >>>viewpoint? >> >>What you say makes sense, Fred. However, in audio, filter bands are >>almost >>always expressed in terms relative to their center or cut-off frequency, >>e.g. in >>octaves. Even in other applications, we tend to _implicitly_ think in >>these >>terms as well. Consider, for example, a "steep" IIR (or analog) filter. >>Generally this thought of in terms of dB/octave (or decade), say >>48dB/octave is >>a steep filter and 6dB/octave is a gentle filter. This is true regardless >>of >>whether the filter cut-off frequency is 1Hz or 1000Hz. So, in my >>experience, >>filter bands in absolute frequency terms are a rarity. > > > We were talking about digital filters in the context of windowing design - > meaning FIR filters. I'm used to using something like P-M for FIR filters > where the band edges are expressed in absolute terms. But, if one is used > to using analog prototype designs and bilinear transformation and IIR > filters (i.e. with poles), then it may be natural to think in those other > terms. Yep. > > Fred > >Fred, If you look at the equations to predict the filter length for the PM algorithm( there are a couple common forms), the length is based largely on the percetage of the transition width to the sampling frequency. The other driving factors are the attenuation and dB ripple specifications. Another way of thinking about it, if you change the samping rate but keep the same filter coefficients, the frequency response is exactly the same but just dilated/scaled by the ratio of the sampling frequencies. Cheers, David
