OT: Where did you get your start?

Hey all,

I was just wondering how everybody got their start in DSP. I am personally
geting mine by working with FPGAs and Simulink tools. Also, I am working my
way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
wondering, as I think it's interesting to know where people came from and
such. I hope to especially hear from the *ahem* experienced (prominent,
older, whatever) members. Thanks!

Chris

Chris.Gammell wrote:

> Hey all,
> 
> I was just wondering how everybody got their start in DSP. I am personally
> geting mine by working with FPGAs and Simulink tools. Also, I am working my
> way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
> wondering, as I think it's interesting to know where people came from and
> such. I hope to especially hear from the *ahem* experienced (prominent,
> older, whatever) members. Thanks!
> 
> Chris

While working on my master's thesis 
(http://www.wescottdesign.com/articles/MSK/mskTop.html).  I was having a 
terrible time with the PLLs in the demodulator -- the oscillators 
weren't nearly stable enough.  I spent days trying to figure out a good 
way to make oscillators with enough range and crystal stability.  Then 
my eye strayed over to the crystal attached to the microprocessor.

After spending another day or two trying to figure out how to make the 
microprocessor cough up the necessary time bases, I had a really stupid 
epiphany -- stupid because it was about 40 years after the invention of 
DSP.  I could A/D convert the signal and demodulate it digitally!

So I did.  And it worked.  And after I trimmed the code down just 
enough, it even worked in real time (consuming 98% of the processor 
resources).

This was kind of a letdown, because I thought I had arranged my training 
to be an analog circuits designer.  The next DSP opportunity that came 
up was in a control system.  I fought against it in the preliminary 
design stage, but it was obvious that if I wanted to squeeze the 
application onto the proffered circuit board I'd have to close the loop 
in software.  Sigh.

It took 10 years, but now unless the solution is very simple and needs 
high bandwidth I just start looking at processors...

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

"Chris.Gammell" <cjg11@case.edu> wrote in
news:KdKdnaEh2cF2SFrenZ2dnUVZ_sKdnZ2d@giganews.com: 

> Hey all,
> 
> I was just wondering how everybody got their start in DSP. I am
> personally geting mine by working with FPGAs and Simulink tools. Also,
> I am working my way into hand coding DSP chips (hence the reason I'm
> here!). Anyway, just wondering, as I think it's interesting to know
> where people came from and such. I hope to especially hear from the
> *ahem* experienced (prominent, older, whatever) members. Thanks!
> 
> Chris
> 

I started at as an analog engineer. At one time, my partner called me the 
only analog engineer without grey hair (since this time, my beard has 
betrayed me).

I built a lot of hardware using op amps, Rs & Cs. I remember thinking 
about how cool it would be to have a dsp block that had ADCs, DACs, and a 
DSP that I could actually afford to use in a real product. I don't think 
there is that much difference between solving problems in the s-plane 
than z-plane. 

In the late 1980s, I got a chance to design a diversity modem for the 
Navy. It used an NEC7725 which was a first generation DSP that is 
horrible by today's standards. The modem replaced a rack mount FSK modem 
that the gov't paid $25K for. Ours was 1/10th the size, weight, power, 
and cost. We never sold any, go figure.....

Nevertheless, it did get me started in DSP.

In the early 1990s, I designed a set of noise reduction products 
primarily sold to ham radio. Perhaps, some of you know of these products 
(Timewave DSP-599zx and others). This is when I started using ADI DSPs 
(2105 & 218x). 

I founded Danville in 1998 as a DSP oriented company specializing in ADI 
DSPs. 

-- 
Al Clark
Danville Signal Processing, Inc.
--------------------------------------------------------------------
Purveyors of Fine DSP Hardware and other Cool Stuff
Available at http://www.danvillesignal.com

I have started working with the DSP because of the practical needs. We 
had a project which could be nicely implemented with the DSP, and 
someone had to do it. At that time I was quite experienced with the 
analog and the embedded stuff and could do basic things like DTMF 
decoding with a microprocessor.
After that I did many DSP projects with the different DSPs however the 
real in deep understanding of how this stuff works was achieved much 
later. In 2001 I started a business for DSP and embedded designs. So far 
so good.

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

Chris.Gammell wrote:
> Hey all,
>
> I was just wondering how everybody got their start in DSP. I am personally
> geting mine by working with FPGAs and Simulink tools. Also, I am working my
> way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
> wondering, as I think it's interesting to know where people came from and
> such. I hope to especially hear from the *ahem* experienced (prominent,
> older, whatever) members. Thanks!
>
> Chris

I was studying as a computer engineer, which differed from the
electrical engineering core at my university by only in four courses
(not counting electives). I talked to a professor one day about feeling
like a half-arsed computer scientist and a half-arsed EE. He suggested
I shoot for the discipline that was in the same place: DSP. And so I
did. :-)

Cheers! --M

Chris.Gammell wrote:
> Hey all,
>
> I was just wondering how everybody got their start in DSP. I am personally
> geting mine by working with FPGAs and Simulink tools. Also, I am working my
> way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
> wondering, as I think it's interesting to know where people came from and
> such. I hope to especially hear from the *ahem* experienced (prominent,
> older, whatever) members. Thanks!

When I was young kid, barley 17 (those were the days :-), I worked as a
technician in a film sound studio, in charge of the sound effect
library (on cassette, 1/4 inch tape and vinyl) and responsible for
transfering location recording to 35mm magnetic tape (for mixing).

One day I walked into one of the studios, there was an engineer laying
the basic sounds for a battle scene (of a WW2 film). After a while of
watching I found out that, instead of coming to my effect library for
the sounds of hand guns, machine guns, cannons and whatever, he
recorded all of them from what looked like a musician's keyboard (which
had an amazingly hi-tech looking computer monitor connected to it).

I was stunned. A keyboard that generated gun shot sounds? He explained
that this wasn't an ordinary keyboard, but something called a "digital
sampler". What really blew me out of my socks was the fact that he
generated all of those sounds from one single pistol shot!

I spent many evenings alone with the sampler (a 12bit 15/30 kHz sampler
with maximum sampling time of 14s / 7s, depeding on the sampling rate).
I got hooked.

:-)

Regards,
Andor

Chris.Gammell wrote:
> Hey all,
> 
> I was just wondering how everybody got their start in DSP. I am personally
> geting mine by working with FPGAs and Simulink tools. Also, I am working my
> way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
> wondering, as I think it's interesting to know where people came from and
> such. I hope to especially hear from the *ahem* experienced (prominent,
> older, whatever) members. Thanks!

Journal's.  In the early '70s I got interested in the problem of how one 
would equalize a loudspeaker to a room (naive, I know) using a digital 
computer and I was in a research group at IBM then which granted us a 
good deal of time for investigating tangential technology and theory and 
provided us with a superb technical library and search engine.

Microprocessors were just emerging then and I thought that would be a 
great application.  Little did I know at the inception the amount of 
computation required.  Anyway, it wasn't called digital signal 
processing or even signal processing yet and my first searches led me to 
digital LMS adaptive equalizers for communications.  The IEEE 
Transactions on Communications were the most fecund source then for 
information on what we now call DSP.  I had some prior experience in 
function minimization and realized pretty quickly that steepest descent 
type algorithms would balk at the size problem I wanted to solve.

I saw that I needed to understand the DFT/FFT which I also accomplished 
from journal articles.  I couldn't figure at that time how to use the 
FFT to solve my problem though, mainly because all the considerations of 
periodicity seemed to me then to rule it out for the FIR world I was 
working in.  I just didn't understand yet that the whole periodicity 
thing was an unnecessary artifact.  Not sure anyone else did either.

At a time of particularly vigorous head scratching I encountered a paper 
in the IEEE Transactions on Communications called "Non-iterative 
Automatic Equalization" which showed that the time domain matrix which 
needed to be solved was Toeplitz and went on to derive what we now call 
the Levinson-Durbin algorithm for solving them.  The author did not seem 
to be aware of any previous work on this and thought it was novel.  I'm 
not sure whether he was re-inventing the wheel or that his priority was 
subsequently overlooked.

Turned out that the hardware for audio quality analog->digital->analog 
was _far_ to expensive at that time for the idea to be feasible and that 
it took a mainframe to implement the Toeplitz algorithm, which I did 
successfully program in APL for IBM mainframes.  I had to get time on a 
System\370 Model 91, the fastest machine IBM made, but I did get to work 
on model problems of the size required.  The technology of the day was 
just not up to the application at any reasonable cost so I put the idea 
on the shelf for many years but the investigation kindled an interest in 
signal processing that has persisted for nearly 35 years.


Bob
-- 

"Things should be described as simply as possible, but no simpler."

                                              A. Einstein

Chris.Gammell wrote:
> Hey all,
> 
> I was just wondering how everybody got their start in DSP. I am personally
> geting mine by working with FPGAs and Simulink tools. Also, I am working my
> way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
> wondering, as I think it's interesting to know where people came from and
> such. I hope to especially hear from the *ahem* experienced (prominent,
> older, whatever) members. Thanks!

Actually, I never did get started. I built a lot of digital and analog 
hardware to meet in-house needs at RCA, and eventually, much of the 
digital stuff became easier to implement with microprocessors, which I 
built (at board level sometimes) and programmed. Marrying digital with 
analog involved A/D and D/A conversion, and some of the signals got 
processed digitally with algorithms I designed ad hoc and pro tem. I 
never quite had the time learn how other people did it, but as long as 
my designs worked, I didn't have to. (One job pulled a ramp signal out 
of noise over 100 times its amplitude. That's filtering!)

I bought a DSK board some time after I retired to see what this DSP 
stuff was all about, came to comp.dsp to find out how to use it and get 
recommendations for books, and I'm still here. I'd like to say "Now I 
are one," but it only looks like I know something, and that not always.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Bob Cain <arcane@arcanemethods.com> wrote in
news:dq9d4m02rmr@enews1.newsguy.com: 

> 
> 
> Chris.Gammell wrote:
>> Hey all,
>> 
>> I was just wondering how everybody got their start in DSP. I am
>> personally geting mine by working with FPGAs and Simulink tools.
>> Also, I am working my way into hand coding DSP chips (hence the
>> reason I'm here!). Anyway, just wondering, as I think it's
>> interesting to know where people came from and such. I hope to
>> especially hear from the *ahem* experienced (prominent, older,
>> whatever) members. Thanks! 
> 
> Journal's.  In the early '70s I got interested in the problem of how
> one would equalize a loudspeaker to a room (naive, I know) using a
> digital computer and I was in a research group at IBM then which
> granted us a good deal of time for investigating tangential technology
> and theory and provided us with a superb technical library and search
> engine. 
> 
> Microprocessors were just emerging then and I thought that would be a 
> great application.  Little did I know at the inception the amount of 
> computation required.  Anyway, it wasn't called digital signal 
> processing or even signal processing yet and my first searches led me
> to digital LMS adaptive equalizers for communications.  The IEEE 
> Transactions on Communications were the most fecund source then for 
> information on what we now call DSP.  I had some prior experience in 
> function minimization and realized pretty quickly that steepest
> descent type algorithms would balk at the size problem I wanted to
> solve. 
> 
> I saw that I needed to understand the DFT/FFT which I also
> accomplished from journal articles.  I couldn't figure at that time
> how to use the FFT to solve my problem though, mainly because all the
> considerations of periodicity seemed to me then to rule it out for the
> FIR world I was working in.  I just didn't understand yet that the
> whole periodicity thing was an unnecessary artifact.  Not sure anyone
> else did either. 
> 
> At a time of particularly vigorous head scratching I encountered a
> paper in the IEEE Transactions on Communications called "Non-iterative
> Automatic Equalization" which showed that the time domain matrix which
> needed to be solved was Toeplitz and went on to derive what we now
> call the Levinson-Durbin algorithm for solving them.  The author did
> not seem to be aware of any previous work on this and thought it was
> novel.  I'm not sure whether he was re-inventing the wheel or that his
> priority was subsequently overlooked.
> 
> Turned out that the hardware for audio quality analog->digital->analog
> was _far_ to expensive at that time for the idea to be feasible and
> that it took a mainframe to implement the Toeplitz algorithm, which I
> did successfully program in APL for IBM mainframes.  I had to get time
> on a System\370 Model 91, the fastest machine IBM made, but I did get
> to work on model problems of the size required.  The technology of the
> day was just not up to the application at any reasonable cost so I put
> the idea on the shelf for many years but the investigation kindled an
> interest in signal processing that has persisted for nearly 35 years.
> 
> Bob


And now you are little more than a washed-up, ex computer programmer with a 
over-inflated ego who is about as inept as they come in the areas of hard 
science, math, physics and engineering.  In fact, you are so technically 
inept that not too long ago you had to go begging in sci.physics for 
someone to lead you hand in solving a simple second-order partial 
differential equation that is discussed in virtually every introductory 
textbook on acoustics.  Worse yet, you were ultimately told by Zigoteau, 
your hand-holder in sci.physics, that you really had no idea what you were 
doing.  

Who cares how technically incompetent frauds like you got started.    

"Chris.Gammell" <cjg11@case.edu> writes:

> I was just wondering how everybody got their start in DSP. I am personally
> geting mine by working with FPGAs and Simulink tools. Also, I am working my
> way into hand coding DSP chips (hence the reason I'm here!). Anyway, just
> wondering, as I think it's interesting to know where people came from and
> such. I hope to especially hear from the *ahem* experienced (prominent,
> older, whatever) members. Thanks!

I didn't know anything about signal processing until the final year of
my bachelors degree in engineering.  I found that, generally, I really
understood the theory of it pretty well.

My main interest is in the algorithms and the analysis and development
of those.

Unlike many of the people on this newsgroup, I do not have much
experience with working on DSP chips (except some early playing with
the Motorola 56k).  

Most of my signal processing has been of the non-real-time variety:

* General frequency estimation
http://home.comcast.net/~kootsoop/freqalgs.htm

* Condition Monitoring of low speed rolling element in a steel mill.
http://home.comcast.net/~kootsoop/CM/index.html

* Estimation and modelling problems in sugar mill dryers.
http://home.comcast.net/~kootsoop/SugarDryer.htm

* Measuring height and volume of solder paste deposits.
http://home.comcast.net/~kootsoop/SPI.html

* The closest I got to real-time was working with Darren Ward on some
array signal processing problems:
http://home.comcast.net/~kootsoop/Array/index.html

Except that Darren did all the hard work. :-)

And these days, I tend to do more "Systems Engineering" than anything
DSP related. :-(

Ciao,

Peter K.

-- 
"And he sees the vision splendid 
of the sunlit plains extended
And at night the wondrous glory of the everlasting stars."