"Jerry Avins" <jya@ieee.org> wrote in message news:u9-dnRq2qamIICPfRVn-jQ@rcn.net...> > I can see from the header that your posts are plain text. However, the > displayed size is larger than what we usually see. It must be the > character set you use: iso-2022-jp. Most of us use something like > ISO-8859-1. Fred owes you an appology for what he wrote and I, for what > I thought. Please accept one from both of us. > > JerryJerry, Unless I have "read in plain text" selected, this message from you is in Arial... ? No apologies, it's just what I'm seeing. Fred
Related to FIR filters
Started by ●June 23, 2005
Reply by ●June 26, 20052005-06-26
Reply by ●June 26, 20052005-06-26
"Jerry Avins" <jya@ieee.org> wrote in message news:SNGdneQED-i0XyPfRVn-jA@rcn.net...> Fred Marshall wrote: > > ... > >> My guess is that upsampling will harm more than help. There is no free >> lunch. > > I agree about the free lunch, but upsampling doing harm isn't intuitively > obvious to me. I suspect that you may have missed the crux of what Ishtiaq > wants to do. I'll put my understanding in different terms than he did to > isolate the core. > > Given: an analog signal to be sampled, and a filter that eliminates > frequencies too high to be "interesting". The filter's cutoff is not > affected by the sample rate we will choose. > > Clearly, the sample rate must be hither than twice the filter's cutoff > frequency. We intend to do digital differentiation on the sample train. > > Ishtiaq asks if there a higher sample rate can provide a more accurate > differentiated signal than a rate close to the minimum. I think it does, > but I haven't taken steps to show it. I know that many IIR filters, > particularly impulse-invariant ones, work best when their critical > frequencies are far below the sample rate. I see no reason to believe that > differentiators can't follow that pattern, and I know that (within limits > set by quantization) very simple ones do. > > There's rarely a free lunch, but one can often buy a good one. A higher > sample rate, either in the ADC or upsampled, is a reasonable price. > > JerryJerry, His clarification says that the sample rate is fixed and he was asking about upsampling those original samples - if I understood correctly. So, I stick with my response. Fred
Reply by ●June 26, 20052005-06-26
Fred Marshall wrote:>"Jerry Avins" <jya@ieee.org> wrote in message >news:u9-dnRq2qamIICPfRVn-jQ@rcn.net... > > >>I can see from the header that your posts are plain text. However, the >>displayed size is larger than what we usually see. It must be the >>character set you use: iso-2022-jp. Most of us use something like >>ISO-8859-1. Fred owes you an appology for what he wrote and I, for what >>I thought. Please accept one from both of us. >> >>Jerry >> >> > >Jerry, > >Unless I have "read in plain text" selected, this message from you is in >Arial... ? >No apologies, it's just what I'm seeing. > >Fred > >Arial? Must be a wireless systems question. :-) Steve
Reply by ●June 27, 20052005-06-27
> I can see from the header that your posts are plain text. However, the > displayed size is larger than what we usually see. It must be the > character set you use: iso-2022-jp. Most of us use something like > ISO-8859-1. Fred owes you an appology for what he wrote and I, for what > I thought. Please accept one from both of us. > > JerryJerry, both of you are very helpful to me always, and there is no need of appology. In fact I want to appologize if my posts cause any inconvinience while reading or replying. I select "plain text" for my posts, but if they go as Arial, then this might be due to some setting, which I am unable to change due to my limited abilities of Japanese Kanji reading. I am using Japanes Windows because some of functions and software (like short cut keys on keyboard and TV/Video play and record) do not work with English Windows. This is actually major problem with Sony's PCs sold in Japan (and might be one of the reasons of drop in their sale as compared to Panasonic and Toshiba notebooks and Dell desktops). Regards, Ishtiaq.
Reply by ●June 27, 20052005-06-27
Fred Marshall wrote:> "Jerry Avins" <jya@ieee.org> wrote in message > news:SNGdneQED-i0XyPfRVn-jA@rcn.net... > >>Fred Marshall wrote: >> >> ... >> >> >>>My guess is that upsampling will harm more than help. There is no free >>>lunch. >> >>I agree about the free lunch, but upsampling doing harm isn't intuitively >>obvious to me. I suspect that you may have missed the crux of what Ishtiaq >>wants to do. I'll put my understanding in different terms than he did to >>isolate the core. >> >>Given: an analog signal to be sampled, and a filter that eliminates >>frequencies too high to be "interesting". The filter's cutoff is not >>affected by the sample rate we will choose. >> >>Clearly, the sample rate must be hither than twice the filter's cutoff >>frequency. We intend to do digital differentiation on the sample train. >> >>Ishtiaq asks if there a higher sample rate can provide a more accurate >>differentiated signal than a rate close to the minimum. I think it does, >>but I haven't taken steps to show it. I know that many IIR filters, >>particularly impulse-invariant ones, work best when their critical >>frequencies are far below the sample rate. I see no reason to believe that >>differentiators can't follow that pattern, and I know that (within limits >>set by quantization) very simple ones do. >> >>There's rarely a free lunch, but one can often buy a good one. A higher >>sample rate, either in the ADC or upsampled, is a reasonable price. >> >>Jerry > > > Jerry, > > His clarification says that the sample rate is fixed and he was asking about > upsampling those original samples - if I understood correctly. > > So, I stick with my response. > > FredAs far as I know, the result of upsampling while keeping the original bandwidth is precisely the same as sampling faster through the original anti-alias filter. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●June 27, 20052005-06-27
Fred Marshall wrote:> "Jerry Avins" <jya@ieee.org> wrote in message > news:u9-dnRq2qamIICPfRVn-jQ@rcn.net... > >>I can see from the header that your posts are plain text. However, the >>displayed size is larger than what we usually see. It must be the >>character set you use: iso-2022-jp. Most of us use something like >>ISO-8859-1. Fred owes you an appology for what he wrote and I, for what >>I thought. Please accept one from both of us. >> >>Jerry > > > Jerry, > > Unless I have "read in plain text" selected, this message from you is in > Arial... ? > No apologies, it's just what I'm seeing. > > FredI don't know why your machine renders it that way. Mine renders it in Courier (or Courier New) about two point sizes larger than normal. I posted the header on the messages I see. Does it match the header you get? (<ctrl>+U) Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●June 27, 20052005-06-27
I. R. Khan wrote:>> I can see from the header that your posts are plain text. However, the >> displayed size is larger than what we usually see. It must be the >> character set you use: iso-2022-jp. Most of us use something like >> ISO-8859-1. Fred owes you an appology for what he wrote and I, for what >> I thought. Please accept one from both of us. >> >> Jerry > > > Jerry, both of you are very helpful to me always, and there is no need > of appology. In fact I want to appologize if my posts cause any > inconvinience while reading or replying. > > I select "plain text" for my posts, but if they go as Arial, then this > might be due to some setting, which I am unable to change due to my > limited abilities of Japanese Kanji reading. I am using Japanes Windows > because some of functions and software (like short cut keys on keyboard > and TV/Video play and record) do not work with English Windows. This is > actually major problem with Sony's PCs sold in Japan (and might be one > of the reasons of drop in their sale as compared to Panasonic and > Toshiba notebooks and Dell desktops). > > Regards, > Ishtiaq.I don't see Ariel. I see a fixed-width font about two points larger than what I normally see. Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply by ●June 28, 20052005-06-28
"Jerry Avins" <jya@ieee.org> wrote in message news:scSdnXsiR-t7JV3fRVn-iA@rcn.net...> > As far as I know, the result of upsampling while keeping the original > bandwidth is precisely the same as sampling faster through the original > anti-alias filter. >Hmmmm, really? Let's see .... We start with a signal of bandwidth B. We sample it at fs=2.4*B to get x(n) We also sample it at 2*fs=4.8*B to get y(n) Now, let's take x(n) and upsample it to get x'(n) We now have a sequence that has a repeating spectrum at fs and is no different than x(n) otherwise except for those pesky little interspersed zeros. Now we take x'(n) and lowpass filter it to get rid of the spectral components around fs resulting in x''(n). We have to be realistic. The original sampling to get x(n) actually results in some aliasing because x(n) is a real signal and not strictly bandlimited. The original sampling to get y(n) causes different aliasing. On top of that, the lowpass filtering of x'(n) isn't perfect so we end up with some energy between fs/2 and 3fs/2. And that energy is aliased for sure! On the other hand, the samples y(n) have less initial aliasing and none of the aliasing caused by filtering x'(n). My concern is that ineffective lowpass filtering of x'(n) will allow "jumps" in the data that will differentiate badly. Fred
Reply by ●June 28, 20052005-06-28
Fred Marshall wrote:> "Jerry Avins" <jya@ieee.org> wrote in message > news:scSdnXsiR-t7JV3fRVn-iA@rcn.net... > >>As far as I know, the result of upsampling while keeping the original >>bandwidth is precisely the same as sampling faster through the original >>anti-alias filter. >> > > > Hmmmm, really? Let's see .... > > We start with a signal of bandwidth B. > We sample it at fs=2.4*B to get x(n) > We also sample it at 2*fs=4.8*B to get y(n) > > Now, let's take x(n) and upsample it to get x'(n) > We now have a sequence that has a repeating spectrum at fs and is no > different than x(n) otherwise except for those pesky little interspersed > zeros. > Now we take x'(n) and lowpass filter it to get rid of the spectral > components around fs resulting in x''(n). > > We have to be realistic. The original sampling to get x(n) actually results > in some aliasing because x(n) is a real signal and not strictly bandlimited. > The original sampling to get y(n) causes different aliasing. On top of > that, the lowpass filtering of x'(n) isn't perfect so we end up with some > energy between fs/2 and 3fs/2. And that energy is aliased for sure! > > On the other hand, the samples y(n) have less initial aliasing and none of > the aliasing caused by filtering x'(n). > > My concern is that ineffective lowpass filtering of x'(n) will allow "jumps" > in the data that will differentiate badly.True enough. I thought we were discussing the theoretical effect of oversampling on the quality of a derivative, and I put aside the secondary considerations to concentrate on question. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●June 28, 20052005-06-28
"Jerry Avins" <jya@ieee.org> wrote in message news:f4KdndLzQbq37VzfRVn-qg@rcn.net...> Fred Marshall wrote: >> "Jerry Avins" <jya@ieee.org> wrote in message >> news:scSdnXsiR-t7JV3fRVn-iA@rcn.net... >> >>>As far as I know, the result of upsampling while keeping the original >>>bandwidth is precisely the same as sampling faster through the original >>>anti-alias filter. >>> >> >> >> Hmmmm, really? Let's see .... >> >> We start with a signal of bandwidth B. >> We sample it at fs=2.4*B to get x(n) >> We also sample it at 2*fs=4.8*B to get y(n) >> >> Now, let's take x(n) and upsample it to get x'(n) >> We now have a sequence that has a repeating spectrum at fs and is no >> different than x(n) otherwise except for those pesky little interspersed >> zeros. >> Now we take x'(n) and lowpass filter it to get rid of the spectral >> components around fs resulting in x''(n). >> >> We have to be realistic. The original sampling to get x(n) actually >> results in some aliasing because x(n) is a real signal and not strictly >> bandlimited. The original sampling to get y(n) causes different aliasing. >> On top of that, the lowpass filtering of x'(n) isn't perfect so we end up >> with some energy between fs/2 and 3fs/2. And that energy is aliased for >> sure! >> >> On the other hand, the samples y(n) have less initial aliasing and none >> of the aliasing caused by filtering x'(n). >> >> My concern is that ineffective lowpass filtering of x'(n) will allow >> "jumps" in the data that will differentiate badly. > > True enough. I thought we were discussing the theoretical effect of > oversampling on the quality of a derivative, and I put aside the secondary > considerations to concentrate on question.That is unlike you Jerry! i.e. when did you give way to a theoretical discussion when there were practical considerations needing to be dealt with? (This sir, is a high compliment!) Fred
