NEWBIE QUESTION: "DSP For Scientists and Engineers." No Maths?

I am reading the free online book "DSP for Scientists and Engineers."
Fantastic.  Concepts are explained clearly, and with a limited amount
of maths.  Seems almost TOO good to be true.  I am asking myself the
question, am I missing out by not having the maths?  I guess it
depends on what you are trying to do...

I'm a radio amateur who is trying to understand DSP sufficient to be
able to program a chip like Microchip's dsPIC33.  I'd like to be able
to decode tones, decode morse and modulate/demodulate PSK and SSB
etc.  I am comfortable with basic embedded system programming and I
have some maths and electrical engineering (20 years ago now)
background.  I am treating this as a learning exercise.

I suppose my question is do I HAVE to dust off my college math to
understand DSP sufficient to do what I have described above, or would
I be able to do it after completing "DSP For Scientists and
Engineers"?  I seem to remember analog filter design being quite
mathmatical, though I was one of those engineers who went into
software pretty quickly, so never actually designed a filter "for
real".

To summarise:

Do I NEED the math in DSP to do the kind of thing I want to do?
If I did need the maths, which book should I choose? (Some
undergraduate level text?)

Hope someone can guide me here...

T.

On Jul 25, 9:33 am, TommieTip...@yahoo.com wrote:
> I am reading the free online book "DSP for Scientists and Engineers."
> Fantastic.  Concepts are explained clearly, and with a limited amount
> of maths.  Seems almost TOO good to be true.  I am asking myself the
> question, am I missing out by not having the maths?  I guess it
> depends on what you are trying to do...
>
> I'm a radio amateur who is trying to understand DSP sufficient to be
> able to program a chip like Microchip's dsPIC33.  I'd like to be able
> to decode tones, decode morse and modulate/demodulate PSK and SSB
> etc.  I am comfortable with basic embedded system programming and I
> have some maths and electrical engineering (20 years ago now)
> background.  I am treating this as a learning exercise.
>
> I suppose my question is do I HAVE to dust off my college math to
> understand DSP sufficient to do what I have described above, or would
> I be able to do it after completing "DSP For Scientists and
> Engineers"?  I seem to remember analog filter design being quite
> mathmatical, though I was one of those engineers who went into
> software pretty quickly, so never actually designed a filter "for
> real".
>
> To summarise:
>
> Do I NEED the math in DSP to do the kind of thing I want to do?
> If I did need the maths, which book should I choose? (Some
> undergraduate level text?)
>
> Hope someone can guide me here...
>
> T.

I think at some point, you need to know some math, but there are good
introductory books. Try "Understanding Digital Signal Processing" by
Lyons.

Jason

You can expect the first practical results in a year or so; in about 5 
years you will be able to do the simple DSP projects. This is what it takes.



Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com


TommieTippee@yahoo.com wrote:

> I am reading the free online book "DSP for Scientists and Engineers."
> Fantastic.  Concepts are explained clearly, and with a limited amount
> of maths.  Seems almost TOO good to be true.  I am asking myself the
> question, am I missing out by not having the maths?  I guess it
> depends on what you are trying to do...
> 
> I'm a radio amateur who is trying to understand DSP sufficient to be
> able to program a chip like Microchip's dsPIC33.  I'd like to be able
> to decode tones, decode morse and modulate/demodulate PSK and SSB
> etc.  I am comfortable with basic embedded system programming and I
> have some maths and electrical engineering (20 years ago now)
> background.  I am treating this as a learning exercise.
> 
> I suppose my question is do I HAVE to dust off my college math to
> understand DSP sufficient to do what I have described above, or would
> I be able to do it after completing "DSP For Scientists and
> Engineers"?  I seem to remember analog filter design being quite
> mathmatical, though I was one of those engineers who went into
> software pretty quickly, so never actually designed a filter "for
> real".
> 
> To summarise:
> 
> Do I NEED the math in DSP to do the kind of thing I want to do?
> If I did need the maths, which book should I choose? (Some
> undergraduate level text?)
> 
> Hope someone can guide me here...
> 
> T.
> 
> 

On 25 Jul, 15:33, TommieTip...@yahoo.com wrote:
> I am reading the free online book "DSP for Scientists and Engineers."
> Fantastic. �Concepts are explained clearly, and with a limited amount
> of maths. �Seems almost TOO good to be true.

It is.

>�I am asking myself the
> question, am I missing out by not having the maths? �I guess it
> depends on what you are trying to do...

No. You need the maths if you want to design or implement DSP
material. DSP is applied maths, so you just can't design anything
without understanding the maths. DSP is appled maths, so you can't
impelement anything without understanding the maths.

...
> I suppose my question is do I HAVE to dust off my college math to
> understand DSP sufficient to do what I have described above,

Yes.

> or would
> I be able to do it after completing "DSP For Scientists and
> Engineers"? �I seem to remember analog filter design being quite
> mathmatical, though I was one of those engineers who went into
> software pretty quickly, so never actually designed a filter "for
> real".

If you thing analog filter design was hard, you have no chance
with DSP. Analog filter design is basically a matter of table
look-up and real-valued computations you can do in minutes with
a simple calculator.

DSP uses complex-valued maths and relies heavily on concepts
from linear algebra.

> To summarise:
>
> Do I NEED the math in DSP to do the kind of thing I want to do?

Yes.

> If I did need the maths, which book should I choose? (Some
> undergraduate level text?)

Look up the literature lists from you local university.

Rune

I appear to be in the minority, but I don't think you need math (or
"the maths" if you aren't American) to solve the types of problems you
mentioned.  If you can read equations other people have dervied, and
translate those into code, you'll be fine.  I'd also disagree it takes
five years to be able to do simple DSP projects.  If you're talking
about problems that aren't already solved, then sure.  Things like
tone decoding don't really fall into that category.

What *might* happen is you implement your simple projects, get a taste
for it, and decide you want to go further.  If so, math is in your
future :-).

Good luck,
Darrell

Rune Allnor wrote:
> On 25 Jul, 15:33, TommieTip...@yahoo.com wrote:
>> I am reading the free online book "DSP for Scientists and Engineers."
>> Fantastic.  Concepts are explained clearly, and with a limited amount
>> of maths.  Seems almost TOO good to be true.
> 
> It is.
> 
>>  I am asking myself the
>> question, am I missing out by not having the maths?  I guess it
>> depends on what you are trying to do...
> 
> No. You need the maths if you want to design or implement DSP
> material. DSP is applied maths, so you just can't design anything
> without understanding the maths. DSP is appled maths, so you can't
> impelement anything without understanding the maths.
> 
> ...
>> I suppose my question is do I HAVE to dust off my college math to
>> understand DSP sufficient to do what I have described above,
> 
> Yes.
> 
>> or would
>> I be able to do it after completing "DSP For Scientists and
>> Engineers"?  I seem to remember analog filter design being quite
>> mathmatical, though I was one of those engineers who went into
>> software pretty quickly, so never actually designed a filter "for
>> real".
> 
> If you thing analog filter design was hard, you have no chance
> with DSP. Analog filter design is basically a matter of table
> look-up and real-valued computations you can do in minutes with
> a simple calculator.

> DSP uses complex-valued maths and relies heavily on concepts
> from linear algebra.

If you are writing the filter software.
OTOH downloading it from a DSP toolbox needs no maths.

-- 
Dirk

http://www.transcendence.me.uk/ - Transcendence UK
http://www.theconsensus.org/ - A UK political party
http://www.onetribe.me.uk/wordpress/?cat=5 - Our podcasts on weird stuff

On Jul 25, 9:33�am, TommieTip...@yahoo.com wrote:

> I suppose my question is do I HAVE to dust off my college math to
> understand DSP sufficient to do what I have described above, or would
> I be able to do it after completing "DSP For Scientists and
> Engineers"? �I seem to remember analog filter design being quite
> mathmatical, though I was one of those engineers who went into
> software pretty quickly, so never actually designed a filter "for
> real".

That book is a good place to start if you're an experimental learner
building your intuition. With your embedded programming experience, I
doubt you'll have any problem implementing the communications
algorithms that you mentioned. DSP is a huge developing field,
however. Don't expect to master everything -- but that's the fun of
it.

> Do I NEED the math in DSP to do the kind of thing I want to do?
> If I did need the maths, which book should I choose? (Some
> undergraduate level text?)

If you learn best by working through problems, I recommend Schaum's
Outlines:

Linear Algebra (Lipschutz & Lipson, 2001)
Matrix Operations (Bronson, 1989)
Digital Signal Processing (Hayes, 1999)
Analog & Digital Communications (Hsu, 2003)
Probability, Random Variables, & Random Processes (Hsu, 1997)

I used the following text to learn DSP:

Discrete-Time Signal Processing (Oppenheim & Schafer, 1999)

Oppenheim (MIT) and Schafer (Georgia Tech) start with a couple of
chapters on discrete-time signals and systems and then go into the
theory of Z-Transforms of LTI systems and Discrete-Time Fourier
transforms of non-periodic signals. The mathematics is presented
electrical engineering style in terms of summations with step and
delta functions rather than matrices and vectors. Here's the kicker:
they don't get around to discussing the DFT of periodic signals until
page 541, but they do spend over 200 pages on the subject. If you
really want to understand DSP, then Z-Transforms and DTFTs are the
place to start -- just don't expect it to be easy.

-Eric

On 2008-07-25, TommieTippee@yahoo.com <TommieTippee@yahoo.com> wrote:
> I am reading the free online book "DSP for Scientists and Engineers."
> Fantastic.

I agree.  I wish I'd had a copy when I first got interested in DSP.

> I suppose my question is do I HAVE to dust off my college math to
> understand DSP sufficient to do what I have described above, or would
> I be able to do it after completing "DSP For Scientists and
> Engineers"?

Programmers use "DSP" all the time without thinking about it in those
terms.  The first time I saw a 1-pole IIR low-pass filter used in software
was in the UNIX kernel's computation of avenrun (load average).  Of course
at the time I didn't recognize it as such, and that meant I had no way of
knowing about the many tools available for analyzing its behavior and
performance.

Lots of programmers (myself included) have simulated PID control loops
and plotted the result to try to tune the parameters.  If you know how
to use the Z-transform you can extract far more information about the
potential performance of your control system in a closed form.

So I would say just reading a book like Smith's gives you a huge leg up.
Knowing the math gives you the flexibility to go beyond cookbook examples.
And the mathematical insight is often a more compact way to understand
a problem.  You could learn a lot about DSP by "rote" just like you could
memorize dozens of trig identities.  But if you can remember Euler's
formula you can work out any trig identity you want.  By the same token
when you understand the math behind DSP you can store it all in a more
compact way.

-- 
Ben Jackson AD7GD
<ben@ben.com>
http://www.ben.com/

I also found the following books interesting:

Who is Fourier? A Mathematical Adventure (Transnational College of
LEX, 1995)
Signals and Systems Made Ridiculously Simple (Karu, 1999)
Digital Filters (Hamming, 1989)

Richard Hamming, developer of the Hamming window, was a mathematician
who worked at Bell Labs until the late 70s. His presentation of
Digital Filters is interdisciplinary, but following his arguments
requires calculus.

Karu's book is based on a review packet for a one-semester course on
signals and systems at MIT. The presentation is mixed continuous and
discrete time.

"Who is Fourier?" is a cute book written by Japanese kids. You go an
adventure with them as they learn calculus and Fourier analysis,
culminating with a derivation of the FFT method. Avoid this book if
you hate all things cute and think learning should be serious and
boring. The same organization also wrote a book on quantum mechanics
and another on DNA.

-Eric

On Jul 25, 5:01&#4294967295;pm, eric <miad...@hotmail.com> wrote:
> If you learn best by working through problems, I recommend Schaum's
> Outlines:
>
> Linear Algebra (Lipschutz & Lipson, 2001)
> Matrix Operations (Bronson, 1989)
> Digital Signal Processing (Hayes, 1999)
> Analog & Digital Communications (Hsu, 2003)
> Probability, Random Variables, & Random Processes (Hsu, 1997)
>
> I used the following text to learn DSP:
>
> Discrete-Time Signal Processing (Oppenheim & Schafer, 1999)
>
> Oppenheim (MIT) and Schafer (Georgia Tech) start with a couple of
> chapters on discrete-time signals and systems and then go into the
> theory of Z-Transforms of LTI systems and Discrete-Time Fourier
> transforms of non-periodic signals. The mathematics is presented
> electrical engineering style in terms of summations with step and
> delta functions rather than matrices and vectors. Here's the kicker:
> they don't get around to discussing the DFT of periodic signals until
> page 541, but they do spend over 200 pages on the subject. If you
> really want to understand DSP, then Z-Transforms and DTFTs are the
> place to start -- just don't expect it to be easy.
>
> -Eric

On Jul 26, 1:33 am, TommieTip...@yahoo.com wrote:
> I am reading the free online book "DSP for Scientists and Engineers."
> Fantastic.  Concepts are explained clearly, and with a limited amount
> of maths.  Seems almost TOO good to be true.  I am asking myself the
> question, am I missing out by not having the maths?  I guess it
> depends on what you are trying to do...
>
> I'm a radio amateur who is trying to understand DSP sufficient to be
> able to program a chip like Microchip's dsPIC33.  I'd like to be able
> to decode tones, decode morse and modulate/demodulate PSK and SSB
> etc.  I am comfortable with basic embedded system programming and I
> have some maths and electrical engineering (20 years ago now)
> background.  I am treating this as a learning exercise.
>
> I suppose my question is do I HAVE to dust off my college math to
> understand DSP sufficient to do what I have described above, or would
> I be able to do it after completing "DSP For Scientists and
> Engineers"?  I seem to remember analog filter design being quite
> mathmatical, though I was one of those engineers who went into
> software pretty quickly, so never actually designed a filter "for
> real".
>
> To summarise:
>
> Do I NEED the math in DSP to do the kind of thing I want to do?
> If I did need the maths, which book should I choose? (Some
> undergraduate level text?)
>
> Hope someone can guide me here...
>
> T.

You need maths alright - maths and further maths. The more the better.
However, you don't need a lot more than complex analysis,Laplace
Transforms,Fourier Transforms and z-transforms.
Should do for starters. Also some simple power series stuff would be
useful.


K.