Forums

AM demodulation

Started by Steve October 7, 2003
Having looked through countless websites and books, I've learnt loads,
but not what i'm looking for: An algorithm for AM demodulation...  Can
anyone help?
Steve wrote:
> Having looked through countless websites and books, I've learnt loads, > but not what i'm looking for: An algorithm for AM demodulation... Can > anyone help?
demodulation output =3D lowpass filter ( abs (input) ) Ren=E9
Or

output = sqrt(input^2+HilbertTransform^2(input))

or same as above but with some aproximation of the sqrt(a^2+b^2)

Michal
"Ren&#2013265929;" <username@fel.nl> p&#2013265933;se v diskusn&#2013265933;m pr&#2013265933;spevku news:3f82961f$1@cs1...
Steve wrote:
> Having looked through countless websites and books, I've learnt loads, > but not what i'm looking for: An algorithm for AM demodulation... Can > anyone help?
demodulation output = lowpass filter ( abs (input) ) Ren&#2013265929;
Ren&#2013265929; <username@fel.nl> wrote in message news:<3f82961f$1@cs1>...
> Steve wrote: > > Having looked through countless websites and books, I've learnt loads, > > but not what i'm looking for: An algorithm for AM demodulation... Can > > anyone help? > > demodulation > > > output = lowpass filter ( abs (input) ) > >
This stuff doesn't exactly come naturally to me... All i have at the mo is a list of numbers, a sampled signal... how do i get the above to work on a list of numbers? Can anyone point me in the direction of a C implementation of the above or any other type of AM demodulation?? p.s. i think in C not maths!! i study for uni maths exams by converting it all to C and running the program....so if you need to mention a hilbert trasform can ya give it in C !!!
Hi Steve : when you say sampled , how was it sampled? Is this list a set of
complex numbers or real numbers and what's the spacing between the samples?
Best of Luck - Mike

"Steve" <steven.fox@nuigalway.ie> wrote in message
news:28ce72ef.0310080335.389fd21e@posting.google.com...
> Ren&#2013265929; <username@fel.nl> wrote in message news:<3f82961f$1@cs1>... > > Steve wrote: > > > Having looked through countless websites and books, I've learnt loads, > > > but not what i'm looking for: An algorithm for AM demodulation... Can > > > anyone help? > > > > demodulation > > > > > > output = lowpass filter ( abs (input) ) > > > > > > This stuff doesn't exactly come naturally to me... > All i have at the mo is a list of numbers, a sampled signal... how do > i get the above to work on a list of numbers? Can anyone point me in > the direction of a C implementation of the above or any other type of > AM demodulation?? > > p.s. i think in C not maths!! i study for uni maths exams by > converting it all to C and running the program....so if you need to > mention a hilbert trasform can ya give it in C !!!
steven.fox@nuigalway.ie (Steve) wrote in message news:<28ce72ef.0310080335.389fd21e@posting.google.com>...

> p.s. i think in C not maths!! i study for uni maths exams by > converting it all to C and running the program....so if you need to > mention a hilbert trasform can ya give it in C !!!
Hmmm... I have two objections 'bout that: First, you rely *way* too much on other people's goodwill. Asking for help with your own code is one thing, asking someone else to code your stuff up for you is quite another. What if I or someone else gave you a code littered with bugs? If you don't know why and how the code is supposed to work, you're lost. Second, DSP relies on maths. If you have problems with the maths, people around here are willing to help. Now, I prefer that people at least make an effort on their own part. The maths can be tricky at times, but that's a part of DSP. For better or for worse. Rune
Rune,

I second your whole e-mail. It seems that people who can spell DSP think
they can do it. What you end up with is code that never worked right in the
first place, and no one knows to fix it.

I worked for a while at a major defense contractor that was working with 10
year old DSP code. No one left understood the DSP code, they just got
follow-on contracts to make more copies run and to improve the GUI.  They
should have made the DSP code work right. Amazingly, since the customer was
dumb and happy, the contractor was not real interested in fixing the code to
improve performance or even in bringing it to the customer's attention.
What a shame.

Steve, if you don't do math, you sure as hell don't do DSP.

Dirk

Dirk A. Bell
DSP Consultant


"Rune Allnor" <allnor@tele.ntnu.no> wrote in message
news:f56893ae.0310081215.ce60b75@posting.google.com...
> steven.fox@nuigalway.ie (Steve) wrote in message
news:<28ce72ef.0310080335.389fd21e@posting.google.com>...
> > > p.s. i think in C not maths!! i study for uni maths exams by > > converting it all to C and running the program....so if you need to > > mention a hilbert trasform can ya give it in C !!! > > Hmmm... I have two objections 'bout that: First, you rely *way* too much
on
> other people's goodwill. Asking for help with your own code is one thing, > asking someone else to code your stuff up for you is quite another. What > if I or someone else gave you a code littered with bugs? If you don't know > why and how the code is supposed to work, you're lost. > > Second, DSP relies on maths. If you have problems with the maths, people > around here are willing to help. Now, I prefer that people at least make > an effort on their own part. The maths can be tricky at times, but that's > a part of DSP. For better or for worse. > > Rune
Steve wrote:
> Ren&#2013265929; <username@fel.nl> wrote in message news:<3f82961f$1@cs1>... > >>Steve wrote: >> >>>Having looked through countless websites and books, I've learnt loads, >>>but not what i'm looking for: An algorithm for AM demodulation... Can >>>anyone help? >> >>demodulation >> >> >>output = lowpass filter ( abs (input) ) >> >> > > > This stuff doesn't exactly come naturally to me... > All i have at the mo is a list of numbers, a sampled signal... how do > i get the above to work on a list of numbers? Can anyone point me in > the direction of a C implementation of the above or any other type of > AM demodulation?? > > p.s. i think in C not maths!! i study for uni maths exams by > converting it all to C and running the program....so if you need to > mention a hilbert trasform can ya give it in C !!!
If you know the math well enough to translate it -- correctly -- to C, you don't need much help with DSP. Just translate the detector equations to C and voila! You need the Hilbert transformer only if the carrier isn't sufficiently oversampled. There are other ways even then. One is to stuff between the samples with zeros, then filter the result so that it represents no frequencies that the original sampling frequency won't support. The result will be suitable for demodulating by Lowpass(average(absval))) because there are sure to be enough points near the peak of the carrier. The computation may actually tale a little longer than the Hilbert method that Kvasnicka described, but the concepts are simpler. Now get busy translating. Jerry -- "I view the progress of science as ... the slow erosion of the tendency to dichotomize." Barbara Smuts, U. Mich. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
"Dirk Bell" <dirkman@erols.com> wrote in message news:<bm215t$np$1@bob.news.rcn.net>...

> I worked for a while at a major defense contractor that was working with 10 > year old DSP code. No one left understood the DSP code, they just got > follow-on contracts to make more copies run and to improve the GUI. They > should have made the DSP code work right. Amazingly, since the customer was > dumb and happy, the contractor was not real interested in fixing the code to > improve performance or even in bringing it to the customer's attention. > What a shame.
<shudder> Those projects are the worst, aren't they... since the project makes little if any sense in the first place, just about anything goes, which means that just about anyone can make a "contribution"... not a place for skilled people to be. The trick is to recognize those things as early as possible and then get the hell out of there. As for old code, I am not unfamiliar with code written in relatively arcane programming languages (Fortran 66), comprising some 100 variable names like "bca", "ycc", "gwack" and the likes (that version of fortran had a 6 (8?) character limit on variable names), and where the only "documentation" of the code was a two-page prosaic text that only vaugely mentioned the basic concept the code was based on, and then only in the most general terms. I think it should be outlawed to express any opinions whatsoever about computer programming, until one have tried to modify or maintain (or even compile) that kind of code. Rune

Dirk Bell wrote:
> > I worked for a while at a major defense contractor that was working with 10 > year old DSP code. No one left understood the DSP code, they just got > follow-on contracts to make more copies run and to improve the GUI. They > should have made the DSP code work right. Amazingly, since the customer was > dumb and happy, the contractor was not real interested in fixing the code to > improve performance or even in bringing it to the customer's attention. > What a shame. >
That is life. Once ago I had the experience similar to yours. I tried to change the things and I only learned three lessons out of it: 1. Nobody cares what is inside as long as the overall product performance is satisfactory. 2. If the damn thing works somehow, leave it alone. 3. You are not going to gain much for yourself by trying to improve somebody's lousy piece of work. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com