Where do the control engineering graduate?

On Sun, 21 Feb 2010 18:13:39 -0800, workaholic wrote:


Don't forget the math department.

I think it depends on what you want to do, and who's hiring.  I know that
when _I_ was looking to replace myself at my last direct-hire job I was
seeing good candidates from both EE and ME programs.  But I think it
depends a lot on the university.

ME programs tend to be more practice-oriented "do this and that will
happen", without a firm grounding on theory -- but there's so much
overlap that both ME and EE programs will graduate folks with a good mix
of theory and practice.

I think the best two things you can do are to fill holes, and get
practical experience.  So, figure out where the holes in your knowledge
are, and go to that department.  If you're strong in ME kind of stuff,
get your graduate degree from an EE department, or visa-versa.  At the
same time, look for a department where you're going to make something
real work.  Ultimately employers don't want you to come churn through
pretty math -- they want you to find real world solutions to real world
problems, which is a task that 'pretty math' types fail at often enough
that it's a cliche.

--
www.wescottdesign.com

I would add physics to Tim's list.

The local university offers degrees in mechatronics.  I think this is
a good place too start.  I have seen too many mechanical engineers
design machines that can't be controlled and electrical or control
engineers that don't know enough about the whole system that they
don't know it can't be controlled.

[rant]
I have a few control books.  Most assume that the system is known and
go through a whole lot of theory as to how to control it.  It must be
a shock when the poor students get out in the real world and find that
there isn't a transfer function stamped to the machine they want to
control.  I think the ultimate is to understand the control theory and
use that to influence the design of the machine so that it is
controllable and the degree to which it is controllable.  Aircraft
manufacturers must do this.

If this is not possible then being able to do real system
identification is necessary to obtain the transfer functions or
differential equations that define the system.  This also requires
understanding of the physics of the machine.  For instance a heat
exchanger is non linear because of the log mean temperature difference
but one can compensate for changing gains as a function of
temperature. Likewise, we get involved with lifting pipe from
horizontal to vertical. The gain changes as the pipe is lifted into
position.   The effect of the gravitation force becomes less as the
pipe becomes more vertical.  The inertia remains the same but the
mechanical advantage or the actuators on the mechanism raising the
pipe changes.    Forces and gains should change for every angle.

I have seen too many control people try to control the impossible and
fail after much effort.   One can't blame them for failing to control
the impossible but they can be blamed for not knowing it is
impossible.
[/rant]

The point is that simply knowing the control equations and theory is
not enough.  Knowing the machine is a must.  That is why I would look
into a mechatronics course.

I believe you have posted about robotics with arms that are very
compliant.  How did that turn out?  I think that is a good example of
the kind of problems to expect in real life.

Peter Nachtwey

pnachtwey wrote:

I agree with the rant.  On the one hand it's sad, because there are a
lot of folks who graduate with lots of control smarts on paper who can't
do the practical work.  On the other hand it's an opportunity if you can
learn what you need to know to fill in the gaps.  Once your boss figures
out that when _you_ say "I have an advanced degree and I can do real
things with it" means what you say and not "I'm a bloviating idiot with
a meaningless piece of paper" you'll go far.

Knowing some physics is a good thing, but I was assuming that getting
through at least 2nd-year physics was required for both EE and ME
courses, and certainly there's a lot of applied physics in 3rd- and
4th-year ME and EE courses.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

Thanks a lot.

My project does not go well. It is a typical problem for EE department
to design a much too difficult 3D flexible arms, that cannot be used
at all in my lab. I only do some simulation based on a naive model,
and complete my thesis. It is a sad story.

Now I have two choices for my PhD study, one is EE department with
good control theory research reputation and the other is ME department
in another university. I don't know how can I choose. I admit that I
am very weak in mechanical engineering knowledge. But I even doubt
myself to be a good control engineer or get  a well-paid job hired as
a control engineer. Also I have still a to-be-a-scientist dream.

workaholic wrote:
graduate school now. I have some research experience

I figured out within a few months of finding my Master's thesis advisor
that I had been profoundly lucky -- I just went shopping by sticking my
head into offices, and settled on the first match.  What I realized was
that my success at doing my thesis was strongly influenced by the style
and personality of the guy I was working for, and there was some god
smiling down on naive little old me the day I went looking.

I'd suggest that you look at the faculty.  Look for someone who sees his
grad students defend successfully, look for someone whose name shows up
on papers in the company of a bunch of his grad students, and if you can
hunt down old grad students of his (look in older journals) and _ask_
them what the guy is like.

Then once you find your thesis advisor, you have found your school.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

Yes, I figured.  Perhaps if the robot was much stiffer you would have
had a chance. The kinematic equations should have been easy.   It is
easy to find them on the internet or in books but compensating for the
change in intertia, mechanical advantage as a function of position in
3D is not easy.  At least now you know what I am talking about.  The
machine must be designed to be controllable.

I have an EE degree but knowing what I know or don't know I would be
tempted to learn more mechanical stuff.  If you have an EE degree then
why not learn the mecahical,.

It seems to me that if you want to work for a small company it would
be best to have a broader education,  Small companies often rely on
one person to play multiple roles.   Big comanies already have people
for each role.  Then the object is to be the best at that role.  So
where to want to work?  So where are the best  opportunities where you
are?

I can see that Tim has a completly criteria.   I agree that the
advisor you chose will affect the getting the PhD.   I was thinking
about what you do afterwards.

Peter Nachtwey

pnachtwey wrote:

graduate school now. I have some research experience

Big companies still need systems engineers that can span disciplines.
There may not be anyone who _does_ every job, but there needs to be
someone who has at least a "Scientific American" level of understanding
of every job to coordinate -- otherwise everyone just asks the
impossible of everyone else, or they each try to solve the whole problem
in their own domain without pushing off their expensive stuff into
another realm where it would be cheap.

As an example, I used to work on a system that had a processor that had
access to every signal, that calculated the secant of an angle and put
it out on a DAC so that said secant could be multiplied _in analog
hardware_ by another signal available to the microprocessor and applied
to another part of the circuit.  Had the fellows designing the analog
electronics realized that if you can take the secant of an angle in a
processor then you can _certainly_ do a multiplication, then they would
have saved several $$ on the board and unending years of grief for the
manufacturing crew, who had to tweak component values every time the
semiconductor vendor changed processes on the multiplier chip.

So there are certainly places for systems engineers who at least have a
notion of how a job may be accomplished in the mechanical, electrical
and software domains (and chemical, psychological, biological, etc., as
applicable), and knows the right language to ask questions in so that
each task gets steered to the domain where it can be done most
effectively and cost effectively.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com