Mmmm . . .
Some food for thought in your various replies.
My replies will be a bit slow, as I'm in the UK and you won't hear from me
until the next day.
Will be checking out some new books, although i find the math a bit heavy
going in some of the ones I've read so far.
I don't have Matlab (rather expensive if you're not a student) but I do have
ScopeFIR, ScopeDSP and Labview 7.
My interest in DSP involves radio receiver design, so the 1000 tap filter is
for narrowband reception.
Incidentally, for those of you who have followed the VMSK arguments
(www.vmsk.org), I believe this is where the thing runs into problems. If you
cannot have an effective receive filter (because it destroys the
modulation), the you are pretty well bound to end up with co-channel
interference problems.
The subject is probably worth starting a new thread, if anyone has anything
new to add on VMSK/MSB.
I'm trying various filters with the (not quite instantaneous) frequency
change, to see the effect. Of course, in most cases the effect is an
amplitude ripple for a short period.
Rick and Jerry, thanks for clearing me up on FIR filters - a look at the
impulse response can be quite enlightening !
Paul
"Paul Lovell" <merit@starnet.freeserve.co.uk> wrote in message
news:clof3m$2lg$1@news6.svr.pol.co.uk...
> Hi all,
>
> Here's a newbie type question, but one which I haven't seen fully
explained
> elsewhere.
>
> Lets say I have a 1000 tap bandpass FIR filter centred on 100kHz and 10kHz
> wide. We'll assume it's perfect (flat passband).
> The group delay is 2.5mS.
> I put a 95kHz continuous sine wave signal at the input, and (after a short
> delay) it appears as a continuous output at the same level and frequency.
>
> Then I suddenly change the frequency of the input signal to 105kHz (as
near
> to instantaneous as possible).
>
> My question is : What happens to the output frequency and amplitude,
> from the filter ?
>
> Does it ramp up to 105kHz in a linear fashion, and which parameters govern
> it's behavior - eg bandwidth of the filter ? group delay ?
>
> What is the maximum rate of change of frequency on the input, which can be
> reproduced, 2.5mS later, on the output ?
>
> Any help much appreciated. My background is (non-DSP) electronic
> engineering rather than mathematical.
>
> Regards, Paul
>
>
>
Reply by Rick Lyons●October 28, 20042004-10-28
On Thu, 28 Oct 2004 15:10:12 -0400, Jerry Avins <jya@ieee.org> wrote:
>Rick Lyons wrote:
>
>> On Thu, 28 Oct 2004 08:42:18 +0100, "Paul Lovell"
>> <merit@starnet.freeserve.co.uk> wrote:
>>
>>
>>>Thanks for your contributions guys,
>>>
>>>I have always assumed that a 1000 tap filter with 400kHz clock would have
>>>2.5uS sample intervals and therefore a 2.5mS delay.
>>>Where am I going wrong here ?
>>>
>>>Please substitute "one clock cycle" for "instantaneous", in my original
>>>request.
>>>Anyway, at my level of knowledge, it appears that the best thing to do here
>>>is try a simulation.
>>>Then I may come back with another question :-)
>>>
>>>Paul
>>>
>>
>>
>> Hi Paul,
>> well, there many different "delays" in the world
>> of DSP. That 2.5 usec of yours is the time between your
>> time-domain samples. Assuming your 1000-tap filter is
>> a tapped-delay line, nonecursive, FIR filter (what else
>> could we assume!), then your filter can be said to have a
>> "delay" of (1000-1)/2 = 999/2 = 499.5 samples.
>> That's a "delay" of 499.5*2.5 usec = 1.2488 msec.
>
> ...
>
>Paul,
>
>Get his book; he's the bloke I wrote about. Get your boss to pay for it.
>
>Jerry
Hi Jer,
Thanks for the "plug".
[-Rick-]
Reply by Jon Harris●October 28, 20042004-10-28
Looks normal to me. Must be on the receive side. Here's what I got:
> There are some good books for beginners like us. (OK: I've been a
> beginner longer than you. Also, I read some if those books.) There's a
> bibliography at http://www.dspguru.com/, and a lot of other good stuff
> too. "Smith" is available chapter by chapter on line (I'm glad I bought
> it). "Lyons" is my favorite, both for it's clarity and for the author's
> being one of the regulars here, with all of the advantages "attendant
> thereto".
"Jerry Avins" <jya@ieee.org> wrote in message
news:2ucsosF28f9ufU1@uni-berlin.de...
> Jerry Avins wrote:
>
> ...
>
> > There are some good books for beginners like us.OKI'vebeena
> > beginner longer th anyou.Also,Ireadsomeifthosebooks.There'sa
> > bibliography at http://www.dspguru.com/, and a lot of other good stuff
> > too. "Smith" is available chapter by chapter on line (I'm glad I bought
> > it). "Lyons" is my favorite, both for it's clarity and for the author's
> > being one of the regulars here, with all of the advantages "attendant
> > thereto".
>
> At least that's what I see in the newsgroup. In my sent file, there's this:
>
> There are some good books for beginners like us. (OK: I've been a
> beginner longer than you. Also, I read some if those books.) There's a
> bibliography at http://www.dspguru.com/, and a lot of other good stuff
> too. "Smith" is available chapter by chapter on line (I'm glad I bought
> it). "Lyons" is my favorite, both for it's clarity and for the author's
> being one of the regulars here, with all of the advantages "attendant
> thereto".
>
> That's been happening to me quite a bit lately. Can anyone suggest why?
> Does my message appear to you as it does to me, or is it normal?
>
> Jerry
> --
> Engineering is the art of making what you want from things you can get.
> �����������������������������������������������������������������������
Reply by Jerry Avins●October 28, 20042004-10-28
Richard Owlett wrote:
...
> It may be a case of "you get what you pay for".
I tried Mozilla because Netscape did it too. The odd circumstance that
not everyone sees it puzzles me. I've not found anything odd in the
received message sources (^U). The spell checker also sees it as odd.
Since my "Sent" image is correct and some see it correctly, it probably
happens when I receive it.
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by Richard Owlett●October 28, 20042004-10-28
Jerry Avins wrote:
> Jerry Avins wrote:
>
> ...
>
>
>>There are some good books for beginners like us.OKI'vebeena
>>beginner longer th anyou.Also,Ireadsomeifthosebooks.There'sa
>>bibliography at http://www.dspguru.com/, and a lot of other good stuff
>>too. "Smith" is available chapter by chapter on line (I'm glad I bought
>>it). "Lyons" is my favorite, both for it's clarity and for the author's
>>being one of the regulars here, with all of the advantages "attendant
>>thereto".
>
>
> At least that's what I see in the newsgroup. In my sent file, there's this:
>
> There are some good books for beginners like us. (OK: I've been a
> beginner longer than you. Also, I read some if those books.) There's a
> bibliography at http://www.dspguru.com/, and a lot of other good stuff
> too. "Smith" is available chapter by chapter on line (I'm glad I bought
> it). "Lyons" is my favorite, both for it's clarity and for the author's
> being one of the regulars here, with all of the advantages "attendant
> thereto".
>
> That's been happening to me quite a bit lately. Can anyone suggest why?
> Does my message appear to you as it does to me, or is it normal?
>
> Jerry
It may be a case of "you get what you pay for".
I've noticed many complaints by users of "free" services [ quotes *very*
significant ] such as Google etc.
My ISP evidently contracts with supernews.com to provide newsgroup
access for its clients. I've never had problems I see others complaining
about. The downside to this nirvana is that my ISP serves a very rural
portion of SW MO USA [ what's an 800 # ? ;]
Reply by Jerry Avins●October 28, 20042004-10-28
Rick Lyons wrote:
> On Thu, 28 Oct 2004 08:42:18 +0100, "Paul Lovell"
> <merit@starnet.freeserve.co.uk> wrote:
>
>
>>Thanks for your contributions guys,
>>
>>I have always assumed that a 1000 tap filter with 400kHz clock would have
>>2.5uS sample intervals and therefore a 2.5mS delay.
>>Where am I going wrong here ?
>>
>>Please substitute "one clock cycle" for "instantaneous", in my original
>>request.
>>Anyway, at my level of knowledge, it appears that the best thing to do here
>>is try a simulation.
>>Then I may come back with another question :-)
>>
>>Paul
>>
>
>
> Hi Paul,
> well, there many different "delays" in the world
> of DSP. That 2.5 usec of yours is the time between your
> time-domain samples. Assuming your 1000-tap filter is
> a tapped-delay line, nonecursive, FIR filter (what else
> could we assume!), then your filter can be said to have a
> "delay" of (1000-1)/2 = 999/2 = 499.5 samples.
> That's a "delay" of 499.5*2.5 usec = 1.2488 msec.
...
Paul,
Get his book; he's the bloke I wrote about. Get your boss to pay for it.
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by Jim Thomas●October 28, 20042004-10-28
Jerry Avins wrote:
> Jerry Avins wrote:
> That's been happening to me quite a bit lately. Can anyone suggest why?
> Does my message appear to you as it does to me, or is it normal?
>
It looked OK on my newsreader (also Mozilla). I haven't seen it on google yet.
--
Jim Thomas Principal Applications Engineer Bittware, Inc
jthomas@bittware.com http://www.bittware.com (603) 226-0404 x536
A pessimist is an optimist with experience
Reply by Rick Lyons●October 28, 20042004-10-28
On Thu, 28 Oct 2004 08:42:18 +0100, "Paul Lovell"
<merit@starnet.freeserve.co.uk> wrote:
>Thanks for your contributions guys,
>
>I have always assumed that a 1000 tap filter with 400kHz clock would have
>2.5uS sample intervals and therefore a 2.5mS delay.
>Where am I going wrong here ?
>
>Please substitute "one clock cycle" for "instantaneous", in my original
>request.
>Anyway, at my level of knowledge, it appears that the best thing to do here
>is try a simulation.
>Then I may come back with another question :-)
>
>Paul
>
Hi Paul,
well, there many different "delays" in the world
of DSP. That 2.5 usec of yours is the time between your
time-domain samples. Assuming your 1000-tap filter is
a tapped-delay line, nonecursive, FIR filter (what else
could we assume!), then your filter can be said to have a
"delay" of (1000-1)/2 = 999/2 = 499.5 samples.
That's a "delay" of 499.5*2.5 usec = 1.2488 msec.
[If you perform the derivative of your filter's phase response,
with respect to frequency, you should obtain a constant that's
equal to 1.2488 msec.)
That 1.2488 msec "delay" goes by several different names
in the literature, but I like to think of it as
the "transient time" of the filter. Digital filters,
like analog filters, have a "transient response" and a
"steady state" response. When an abrupt change occurs
at the input of your filter
the "effect" of that change doesn't "fully" show up at
the output of your filter until 1.2488 msec later.
After that first 1.2488 msec, then it takes another
1.2488 msec for the filter to reach "steady state".
If you suddenly apply a 105 kHz sinewave input signal
to the filterit will take 1000 input samples before
your filter output reaches a "steady state",
maximum amplitude, output of a 105 kHz sinewave.
The delay between the input 105 kHz sinewave and
the output 105 kHz sinewave will be 499.5 samples.
By the way, do you really need a 1000-tap filter.
That sure sounds like a lot of taps to me!
(Such a FIR filter, if it's linear-phase, would require
at least, roughly, 500 multiplies and 1000 additions
per input sample.) I'll bet you a beer there's a way
to perform your desired filtering at a lower
computational workload per input sample.
Good Luck,
[-Rick-]
Reply by Jerry Avins●October 28, 20042004-10-28
Jerry Avins wrote:
...
> There are some good books for beginners like us.OKI'vebeena
> beginner longer th anyou.Also,Ireadsomeifthosebooks.There'sa
> bibliography at http://www.dspguru.com/, and a lot of other good stuff
> too. "Smith" is available chapter by chapter on line (I'm glad I bought
> it). "Lyons" is my favorite, both for it's clarity and for the author's
> being one of the regulars here, with all of the advantages "attendant
> thereto".
At least that's what I see in the newsgroup. In my sent file, there's this:
There are some good books for beginners like us. (OK: I've been a
beginner longer than you. Also, I read some if those books.) There's a
bibliography at http://www.dspguru.com/, and a lot of other good stuff
too. "Smith" is available chapter by chapter on line (I'm glad I bought
it). "Lyons" is my favorite, both for it's clarity and for the author's
being one of the regulars here, with all of the advantages "attendant
thereto".
That's been happening to me quite a bit lately. Can anyone suggest why?
Does my message appear to you as it does to me, or is it normal?
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by Jerry Avins●October 28, 20042004-10-28
Paul Lovell wrote:
> Thanks for your contributions guys,
>
> I have always assumed that a 1000 tap filter with 400kHz clock would have
> 2.5uS sample intervals and therefore a 2.5mS delay.
> Where am I going wrong here ?
...
The delay of a symmetric or anti-symmetric FIR -- such symmetry is so
nearly universal that it is assumed unless otherwise specified -- it the
time for a sample to reach the middle tap. (Filters with an even number
of taps have a half-period delay. That's why, when I gave the
approximate delay without the fraction, I cited the error in my usual
nit-picking style.) The filter coefficients are generally large in the
center and small at the ends, making that behavior intuitively
reasonable. Even when the coefficients are all the same, this is easily
seen. Such an arrangement is called a "boxcar averager" (it's not
usually the best implementation) and you can sort of see that the
average of the last n values applies best to the middle one of the n. If
you want the original and averaged curves to line up, you need to delay
the original one by n/2.
There are some good books for beginners like us. (OK: I've been a
beginner longer than you. Also, I read some if those books.) There's a
bibliography at http://www.dspguru.com/, and a lot of other good stuff
too. "Smith" is available chapter by chapter on line (I'm glad I bought
it). "Lyons" is my favorite, both for it's clarity and for the author's
being one of the regulars here, with all of the advantages "attendant
thereto".
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������