What does a DSP engineer do?

hi everyone,

Kindly excuse, for writing this long a post.

I'm new to DSP & to DspRelated.com. I currently work as a s/w programmer
in the windows world, and have been for a while now. No offense to anyone,
but i find what i do a little too dry which literally puts my brain to
sleep. 
Hence i've been pondering to switch careers for a while now, and have
decided to take the first step. 

I have always loved math (kinda lost touch now).  I had a course in
Signals & Systems and in DSP back in college, which i was interested in.
Hence, I decided to start learning DSP, did some Google'ing and picked up
this book which had some excellent reviews (rightfully so), "The Scientist
& Engineer's Guide to DSP By Steven W Smith". I have read a few topics,
next is the Fourier analysis. Before i get over excited and quit my job, i
would like to know a few things about this field.


I bumped into the following article, while i was also wondering what a DSP
engineer would do. 
http://www.dspdesignline.com/howto/206502300

Here he mentions how the industry has changed such that, a 'typical Dsp
Engineer' need not even know much about DSP theory or the math involved. 
How true is this? Is it a narrow minded view of Dsp (A field which i
understand is pretty vast)? What would anyone potentially be doing with Dsp
in real world?

Even better, i would greatly appreciate if you could spare some time, and
let me in on what you are doing in this world of Dsp. 
This would really help me understand what i could achieve learning it, and
what path i can take in DSP.

Would also appreciate, any advice/suggestions/information you feel that
could help me as a beginner.

Thanks
ShaQ

There are various past threads, some recent, that should interest you:

http://groups.google.com/groups/search?q=group%3Acomp.dsp+career+-opportunity


Martin

-- 
Quidquid latine scriptum est, altum videtur.

learningDsp wrote:
> hi everyone,
> 
> Kindly excuse, for writing this long a post.
> 
..
> I bumped into the following article, while i was also wondering what a DSP
> engineer would do. 
> http://www.dspdesignline.com/howto/206502300
> 
> Here he mentions how the industry has changed such that, a 'typical Dsp
> Engineer' need not even know much about DSP theory or the math involved. 
> How true is this? Is it a narrow minded view of Dsp (A field which i
> understand is pretty vast)? What would anyone potentially be doing with Dsp
> in real world?
> 

I guess that statement refers to the increasing reliance on wiring 
prefabricated code blocks together, without worrying too much about how 
they work in detail, or how they are designed. So supposedly the 
industry is represented by "users" and "designers", though any 
self-respecting engineer would expect to be both. From the many past 
posts to this list, I suspect most people here would apply that 
description to managers, not engineers.

Most people on this list probably would regard dsp as a means to an end, 
so would ask "what do you want to use it for?". Sonar, comms, forensics, 
military, space, etc. In my case, working on music software, DSP can be 
as much the end as the means - you might discover/invent some hitherto 
unknown dsp process, that becomes a world-leading synthesis or 
processing technique; and you are designing these things as a direct 
creative act, not so much to "solve a problem". So dsp in music and 
audio can feed more than the brain. Read about VST plugins, consider the 
Mac ( lots more fun than Windows these days...) and look therein at 
CoreAudio AudioUnits; and even the iPhone. Some of the top Apps for that 
are music/audio ones. Helps if you like music, of course.

It is however also true that most computer musicians these days are 
users, not designers; they know just about enough to select a particular 
type of filter to get some effect, but would not have any idea how to 
design one from scratch. Guilty as charged!

Richard Dobson

On Thu, 11 Jun 2009 07:26:28 -0500, learningDsp wrote:

> hi everyone,
> 
> Kindly excuse, for writing this long a post.
> 
> I'm new to DSP & to DspRelated.com. I currently work as a s/w programmer
> in the windows world, and have been for a while now. No offense to
> anyone, but i find what i do a little too dry which literally puts my
> brain to sleep.
> Hence i've been pondering to switch careers for a while now, and have
> decided to take the first step.
> 
> I have always loved math (kinda lost touch now).  I had a course in
> Signals & Systems and in DSP back in college, which i was interested in.
> Hence, I decided to start learning DSP, did some Google'ing and picked
> up this book which had some excellent reviews (rightfully so), "The
> Scientist & Engineer's Guide to DSP By Steven W Smith". I have read a
> few topics, next is the Fourier analysis. Before i get over excited and
> quit my job, i would like to know a few things about this field.
> 
> 
> I bumped into the following article, while i was also wondering what a
> DSP engineer would do.
> http://www.dspdesignline.com/howto/206502300
> 
> Here he mentions how the industry has changed such that, a 'typical Dsp
> Engineer' need not even know much about DSP theory or the math involved.
> How true is this? Is it a narrow minded view of Dsp (A field which i
> understand is pretty vast)? What would anyone potentially be doing with
> Dsp in real world?
> 
> Even better, i would greatly appreciate if you could spare some time,
> and let me in on what you are doing in this world of Dsp. This would
> really help me understand what i could achieve learning it, and what
> path i can take in DSP.
> 
> Would also appreciate, any advice/suggestions/information you feel that
> could help me as a beginner.
> 
> Thanks
> ShaQ

Richard Dobson nailed it when he said that most people on this list don't 
do DSP, they _use_ DSP to get other stuff done.  Even those of us who are 
consultants applying DSP techniques to a wide variety of problems don't 
"do DSP" so much as we do systems, communications, control, etc.

I disagree to some extent with the Designline article* about DSP work 
diminishing to appliance users.  If you count fielded units you'll find 
that most of the design work is segregated into systems architects and 
coders.  Indeed, if you think about how special it is to have someone 
who's an ace circuit designer as well as an ace coder this makes sense.  
But if you count the total design hours spent I think you'll find a lot 
of people on smaller projects who are embracing more of the work than the 
article indicates.  It's just that the work is embedded into other 
things, rather than being sold as an end in and of itself.

I would suggest that a better use for your current skill set, and one 
that could lead to more DSP work if you aim yourself correctly, would be 
to try to get into doing embedded systems programming.  This is another 
field where the thing you do isn't the end product of the system (no one 
sells microwave oven software, but lots of companies sell microwave 
ovens).  But if you like dealing with a wider variety of issues, and you 
get satisfaction from seeing hardware work, programming embedded systems 
can be a load of fun.  Furthermore, since much of the 'real' DSP work 
that's going on is embedded in the embedded firmware, embedded systems 
programming is a good way to start sneaking up on the field of "DSP".

* Rune Allnor has also made comments to this effect: I respectfully 
disagree with him, too.

-- 
www.wescottdesign.com

On Jun 11, 11:28&#4294967295;am, Tim Wescott <t...@seemywebsite.com> wrote:
> On Thu, 11 Jun 2009 07:26:28 -0500, learningDsp wrote:
> > hi everyone,
>
> > Kindly excuse, for writing this long a post.
>
> > I'm new to DSP & to DspRelated.com. I currently work as a s/w programmer
> > in the windows world, and have been for a while now. No offense to
> > anyone, but i find what i do a little too dry which literally puts my
> > brain to sleep.
> > Hence i've been pondering to switch careers for a while now, and have
> > decided to take the first step.
>
> > I have always loved math (kinda lost touch now). &#4294967295;I had a course in
> > Signals & Systems and in DSP back in college, which i was interested in.
> > Hence, I decided to start learning DSP, did some Google'ing and picked
> > up this book which had some excellent reviews (rightfully so), "The
> > Scientist & Engineer's Guide to DSP By Steven W Smith". I have read a
> > few topics, next is the Fourier analysis. Before i get over excited and
> > quit my job, i would like to know a few things about this field.
>
> > I bumped into the following article, while i was also wondering what a
> > DSP engineer would do.
> >http://www.dspdesignline.com/howto/206502300
>
> > Here he mentions how the industry has changed such that, a 'typical Dsp
> > Engineer' need not even know much about DSP theory or the math involved.
> > How true is this? Is it a narrow minded view of Dsp (A field which i
> > understand is pretty vast)? What would anyone potentially be doing with
> > Dsp in real world?
>
> > Even better, i would greatly appreciate if you could spare some time,
> > and let me in on what you are doing in this world of Dsp. This would
> > really help me understand what i could achieve learning it, and what
> > path i can take in DSP.
>
> > Would also appreciate, any advice/suggestions/information you feel that
> > could help me as a beginner.
>
> > Thanks
> > ShaQ
>
> Richard Dobson nailed it when he said that most people on this list don't
> do DSP, they _use_ DSP to get other stuff done. &#4294967295;Even those of us who are
> consultants applying DSP techniques to a wide variety of problems don't
> "do DSP" so much as we do systems, communications, control, etc.
>
> I disagree to some extent with the Designline article* about DSP work
> diminishing to appliance users. &#4294967295;If you count fielded units you'll find
> that most of the design work is segregated into systems architects and
> coders. &#4294967295;Indeed, if you think about how special it is to have someone
> who's an ace circuit designer as well as an ace coder this makes sense. &#4294967295;
> But if you count the total design hours spent I think you'll find a lot
> of people on smaller projects who are embracing more of the work than the
> article indicates. &#4294967295;It's just that the work is embedded into other
> things, rather than being sold as an end in and of itself.
>
> I would suggest that a better use for your current skill set, and one
> that could lead to more DSP work if you aim yourself correctly, would be
> to try to get into doing embedded systems programming. &#4294967295;This is another
> field where the thing you do isn't the end product of the system (no one
> sells microwave oven software, but lots of companies sell microwave
> ovens). &#4294967295;But if you like dealing with a wider variety of issues, and you
> get satisfaction from seeing hardware work, programming embedded systems
> can be a load of fun. &#4294967295;Furthermore, since much of the 'real' DSP work
> that's going on is embedded in the embedded firmware, embedded systems
> programming is a good way to start sneaking up on the field of "DSP".
>
> * Rune Allnor has also made comments to this effect: I respectfully
> disagree with him, too.
>
> --www.wescottdesign.com- Hide quoted text -
>
> - Show quoted text -

I call people who write low-level dsp software or do an equivalent
implementation in hardware, who do not understand the theory and
practical details behind DSP, "Lousy Engineers".  They often make
stupid mistakes, producing systems that seem to them to work, but
whose performance falls short of what it is supposed to be.  In
general, how do you test a system that you do not understand?

Dirk

On 11 Jun, 17:28, Tim Wescott <t...@seemywebsite.com> wrote:

> Richard Dobson nailed it when he said that most people on this list don't
> do DSP, they _use_ DSP to get other stuff done.

Agreed. &#4294967295;

> I disagree to some extent with the Designline article* about DSP work
> diminishing to appliance users.
...
> * Rune Allnor has also made comments to this effect: I respectfully
> disagree with him, too.

Well, I have only expressed what observations I have
made on how the world actually works, which is not at
all how I would want that it worked.

I think Vladimir made the point for me with his list
of emails, published in another thread. To you and
me the people who write that sort of stuff are, well,
stupidents and studiots, to use Vladimir's words.

To the 15-year-olds of today, who might have talents
and interests which might lead them in contact with DSP
within the next decade or so, the same people are future
teachers, bosses and clients. Not a pleasent prospect,
if you ask me.

Rune

On Thu, 11 Jun 2009 08:52:11 -0700, Rune Allnor wrote:

> On 11 Jun, 17:28, Tim Wescott <t...@seemywebsite.com> wrote:
> 
>> Richard Dobson nailed it when he said that most people on this list
>> don't do DSP, they _use_ DSP to get other stuff done.
> 
> Agreed.
> 
>> I disagree to some extent with the Designline article* about DSP work
>> diminishing to appliance users.
> ...
>> * Rune Allnor has also made comments to this effect: I respectfully
>> disagree with him, too.
> 
> Well, I have only expressed what observations I have made on how the
> world actually works, which is not at all how I would want that it
> worked.

Interestingly, I'm expressing what observations I've made, too.  There 
are a _lot_ of "DSP appliance users" out there, who more or less 
mindlessly put together canned routines, and who are aided and abetted by 
The Mathworks.  Yet there are also a lot of folk, not all of them old, 
who are doing small apps, working from the bare metal outward, who meed 
the definition of "old time DSP engineer".  The salient point for this 
thread is that they are rarely called "DSP engineers".  They're called 
"Bob, who always makes things work", or "Systems Engineer", or "FPGA guy 
(give him all the weird stuff)", or "Senior Software Engineer (give him 
all the weird stuff)", etc.

Clearly we come from the same worlds (unless Mars has it's Nordic 
countries stashed someplace and you have a Very Interesting USENET 
connection).  Yet we see two different views.  I think it's because I 
come more from embedded where things have to be small and efficient, and 
you seem to come more from the "gather a ton of data and process it on a 
PC" world.

> I think Vladimir made the point for me with his list of emails,
> published in another thread. To you and me the people who write that
> sort of stuff are, well, stupidents and studiots, to use Vladimir's
> words.

Except that I absolutely, positively reject Vladimir's assumption that 
just because they don't have knowledge and facility in the difficult 
field of DSP that they are mentally or morally inferior to me.  I can't 
accept that assumption because I know too many people who are a lot 
sharper than me in an infinite number of ways, yet get a different answer 
every time they add two and two.

> To the 15-year-olds of today, who might have talents and interests which
> might lead them in contact with DSP within the next decade or so, the
> same people are future teachers, bosses and clients. Not a pleasant
> prospect, if you ask me.

True, but it's one that I deal with all the time.  Fortunately I run 
across enough people who are willing to call me a systems engineer* and 
give me free rein.  One _can_ function in such an environment, although 
having good people skills as well as technical skills is practically 
mandatory.

* The working definition of "Systems Engineer" is "We don't know what he 
does, or how he does it, but he's obviously smart and valuable, so we'll 
pay him".

-- 
www.wescottdesign.com

Tim Wescott wrote:
> On Thu, 11 Jun 2009 08:52:11 -0700, Rune Allnor wrote:
<snip>
> 
> Clearly we come from the same worlds (unless Mars has it's Nordic 
> countries stashed someplace and you have a Very Interesting USENET 
> connection).  Yet we see two different views.  I think it's because I 
> come more from embedded where things have to be small and efficient, and 
> you seem to come more from the "gather a ton of data and process it on a 
> PC" world.
> 

Last I checked, we can do both. Whether or not to do so
depends on the question being asked.

<snip>

--
Les Cargill

On 11 Jun, 19:17, Tim Wescott <t...@seemywebsite.com> wrote:
> On Thu, 11 Jun 2009 08:52:11 -0700, Rune Allnor wrote:

> Clearly we come from the same worlds (unless Mars has it's Nordic
> countries stashed someplace and you have a Very Interesting USENET
> connection).

You might be surprised...

>&#4294967295;Yet we see two different views. &#4294967295;I think it's because I
> come more from embedded where things have to be small and efficient, and
> you seem to come more from the "gather a ton of data and process it on a
> PC" world.

Seems we have pretty much the same impression of our respecive
worlds, although I disagree with your suggested casues of our
different views and opinions.

I think the main difference is that you deal with measurable,
quantifiable deliveries: Either this controller contains the
temperature (or whatever the quantity is) in the specified
range - which is measured by some independent, well understood,
quantitative method - or it doesn't. Yes/No. Pass/Fail.
Quantitative = numbers. Very simple.

The people I used to work for appeared to have seen all the
Star Treks, Sci-Fis and C.S.Is, and wanted all those cool toys.
OK, C.S.I didn't start until I had already reached Dante's
6th level (I even had a quick decent to the 7th level, but I
fortunately didn't stay there very long), but you get the idea.

Anyway, the bosses and clients wanted a sonar that told them
that this target was, say, a wild salmon, while that target
was a farmed salmon. And they were only interested in salmons.
Cods, trouts, hallibuts - not intersting.

There were various economic, political, zoological, environmental
etc reasons for people wanting to make that distinction, and
the argument all too often went like "We are approaching/in
the 3rd millennium and this type of system ought to be available
by now. Rune is good at these things, we'll have him make
this data processing system."

When I told them that all a sonar can do is to locate *some*
*nondescript* target in space and *possibly* track it over
time (I certainly can not come up with a quantitative difference
between 'wild salmon' and 'farmed salmon' by means of sonar
data), *I* was all of a sudden blamed for being uncooperative.
Of course, it didn't take much of a discussion before bosses
and clients got the impression that I thought they were stupid[*].

Which usually (but not always) was wrong. I thought they were
incompetent.

[*] This has nothing to do with people skills. It is impossible
to point out a fact on a basic level similar to 2+2 =/= 5 without
the opposite party getting that impression. On one occasion I
essentially exposed a professor in radars and electromagnetics
as a fraud. In order to get his 'measurements' to fit, he guy
'calibrated' the EM refractive index of air to a value of 1.4.
I looked up a high-school physcs book to find a value of 1.0004
(or maybe it was 1.004, I don't remember).

I just can't see how people skills can prevent that this guy
gets the impression that I think he is incompetent/a fraud/a
total twat. But that was totally typical for the kind of
discussions I used to have with bosses and clients.

> > I think Vladimir made the point for me with his list of emails,
> > published in another thread. To you and me the people who write that
> > sort of stuff are, well, stupidents and studiots, to use Vladimir's
> > words.
>
> Except that I absolutely, positively reject Vladimir's assumption that
> just because they don't have knowledge and facility in the difficult
> field of DSP that they are mentally or morally inferior to me. &#4294967295;I can't
> accept that assumption because I know too many people who are a lot
> sharper than me in an infinite number of ways, yet get a different answer
> every time they add two and two.

Do you read a superiority syndrome into Vlad's posts? Really?

Interesting. I don't. I see somebody who is intensely
frustrated with people who want quick'n dirty, canned
solutions for free, be it economics or efforts. That's
a frustration I most certainly can recognize and sympathize
with. Maybe it's a cultural thing; some dividing line
down the midst of the Atlantic.

> > To the 15-year-olds of today, who might have talents and interests which
> > might lead them in contact with DSP within the next decade or so, the
> > same people are future teachers, bosses and clients. Not a pleasant
> > prospect, if you ask me.
>
> True, but it's one that I deal with all the time. &#4294967295;Fortunately I run
> across enough people who are willing to call me a systems engineer* and
> give me free rein. &#4294967295;One _can_ function in such an environment, although
> having good people skills as well as technical skills is practically
> mandatory.

What do you prefer the next time you travel by aeroplane?
A pilot who has people skills but only rudimentary knowledge
about planes and flying? Or a pilot who knows all the nooks
and crannies of his plane?

> * The working definition of "Systems Engineer" is "We don't know what he
> does, or how he does it, but he's obviously smart and valuable, so we'll
> pay him".

That particular situation - being approached as somebody who
has skills in my own rights, as opposed to being an operator
on somebody else's behalf - has occured to me exactly two (2)
times over the past 20 years. On both occasions with rather
stunning success, compared to all the BS I have been dragged
into. The first time I managed to use acoustics to do measurements
in rather complicated structures. That project started with
my clients being pissed off by some professor, who had suggested
an approach that obviously did not work, and my clients wanted
to cut him loose. I basically pointed out that the guy was a
fraud (it was the professor I mentioned above), told my clients
why, and what should be done instead. The other time I managed
to sort out a number of operational snags and glitches in a
complex operation.

Again, the problem is that people (bosses, clients) near my
line of work, who do not have basik skills, knowledge or the
hands-on experience I have, *think* they understand what might
be reasonable to expect, when in fact they do not.

That's a lose-lose situation for everybody involved.

Your greatest business asset is that your clients know
they are laymen in your field of work. Enjoy that situation
as long as it lasts. It will not remain like that forever.
Maybe you personally will stay 'top dog' due to age and
seniority, but from a community POV your profession, too,
will end up in the degenerate hell where mine has been
for decades already.

Rune

>Richard Dobson nailed it when he said that most people on this list don't

>do DSP, they _use_ DSP to get other stuff done.  Even those of us who are

>consultants applying DSP techniques to a wide variety of problems don't 
>"do DSP" so much as we do systems, communications, control, etc.

Correct me if i am wrong, 
Basically what it means for me is, i not only have to learn DSP, also need
to be familiar with other technologies where it is used. Hence, I have a
lot of reading coming up.