Which university produces good analog EEs?

[ehsjr]

I honestly think these days there's a lot more paranoia about "intellectual
property" than there was 20+ years ago and companies *instruct* their
employees to be a lot less "visible" than they otherwise might be.

True.
Ed

 

[[email protected]]

Hello Folks,

Happens a lot these days, last time an hour ago: Someone is looking for
an analog/mixed signal engineer (this time low power design). I could do
it but they absolutely want to have someone on staff. Which I can't do.
So, I often try to convince them to settle for a youngster who gets
coached now and then, instead of sitting there a year from now still
trying to find the perfect candidate.

Which US or Canadian university lets off the best analog/mixed EEs? I
know, I know, many can't even solder etc. It ain't like it used to be.
But there has got to be an alma mater that sticks out. Or maybe a
particular institute at one. And please, no pissing contests.

I never interviewed a Georgia Tech grad, but Marshall Leach, Jr.
certain has the credentials.
http://users.ece.gatech.edu/~mleach/

I've interviewed plenty of UC Berkeley grads, and you could find an
analog designer there.This doesn't mean every grad from UCB will know
analog. The worse Ivy League has got to be hands down MIT. I assume it
was a good university at one time given it's reputation. But I
interviewed undergrads that didn't know basic s-plane stability
issues, as if they don't teach classic control theory anymore What
little analog they knew was bipolar.

A real surprise are the University of Toronto grads. These guys know
analog and signal processing.

The trouble with low power (assuming you mean micropower) is you
really need to be a careful designer, especially if the chip is
designed to have low quiescent power but handle high current. You also
need the benefit of seeing a few designs that didn't work, hopefully
not your own but from the company portfolio of goofs. One of the
classic bugs is designing micropower bandgaps, only to have them get
pumped from an on-board switcher. You have to throw in all sorts of
parasitics to make sure nothing sneaks into your reference.

 

[Jim Thompson]

Do you still have some of the first transistors?

Got a CK722 ? Someone said they've gone for > $700
on E-bay. (I find that very hard to believe.) Anyway,
you might be sitting on a goldmine.

Ed

Yep. CK722, CK760, CK761 plus a number of the GE vacuum "pinch" tube
sealed devices.

I'll give them to my kids before the tax man tries to "death tax"
them.

...Jim Thompson

 

[Jim Thompson]

I never interviewed a Georgia Tech grad, but Marshall Leach, Jr.
certain has the credentials.
http://users.ece.gatech.edu/~mleach/

I've interviewed plenty of UC Berkeley grads, and you could find an
analog designer there.This doesn't mean every grad from UCB will know
analog. The worse Ivy League has got to be hands down MIT. I assume it
was a good university at one time given it's reputation. But I
interviewed undergrads that didn't know basic s-plane stability
issues, as if they don't teach classic control theory anymore What
little analog they knew was bipolar.

Probably only the duds applied to your company ;-)

However I never had control systems, per se, undergrad... just a very
good math background to understand it if I needed it. (and that was
more than 45 years ago.)

Took non-linear control systems in grad school. Instructor scared the
piss out of me by announcing that, to weed down the class size, he was
giving an exam in undergrad control systems... pass or walk :-(

I got the best score, and "A" as my final grade ;-)

A real surprise are the University of Toronto grads. These guys know
analog and signal processing.

I don't know any of those.

The trouble with low power (assuming you mean micropower) is you
really need to be a careful designer, especially if the chip is
designed to have low quiescent power but handle high current. You also
need the benefit of seeing a few designs that didn't work, hopefully
not your own but from the company portfolio of goofs. One of the
classic bugs is designing micropower bandgaps, only to have them get
pumped from an on-board switcher. You have to throw in all sorts of
parasitics to make sure nothing sneaks into your reference.

Yep :-(

...Jim Thompson

 

[Joerg]

I never interviewed a Georgia Tech grad, but Marshall Leach, Jr.
certain has the credentials.
http://users.ece.gatech.edu/~mleach/

I've interviewed plenty of UC Berkeley grads, and you could find an
analog designer there.This doesn't mean every grad from UCB will know
analog. The worse Ivy League has got to be hands down MIT. I assume it
was a good university at one time given it's reputation. But I
interviewed undergrads that didn't know basic s-plane stability
issues, as if they don't teach classic control theory anymore What
little analog they knew was bipolar.

A real surprise are the University of Toronto grads. These guys know
analog and signal processing.

Bingo! That's exactly where one of my clients finally got they analog
guy from. Now he has to get used to the fact that there is no more ice
skating because it's Southern California.

The trouble with low power (assuming you mean micropower) is you
really need to be a careful designer, especially if the chip is
designed to have low quiescent power but handle high current. You also
need the benefit of seeing a few designs that didn't work, hopefully
not your own but from the company portfolio of goofs. One of the
classic bugs is designing micropower bandgaps, only to have them get
pumped from an on-board switcher. You have to throw in all sorts of
parasitics to make sure nothing sneaks into your reference.

It takes experience. But most importantly it requires to say good-bye to
the temptation to use chips in the can or digital processing for
everything that looks complicated. It's often back to transistors and
you've got to know those things inside out.

 

[Joerg]

Probably only the duds applied to your company ;-)

However I never had control systems, per se, undergrad... just a very
good math background to understand it if I needed it. (and that was
more than 45 years ago.)

Took non-linear control systems in grad school. Instructor scared the
piss out of me by announcing that, to weed down the class size, he was
giving an exam in undergrad control systems... pass or walk :-(

I got the best score, and "A" as my final grade ;-)




I don't know any of those.




Yep :-(

Maybe that's why there are no (cheap) ultralow brethren to the TLV431.
If you want less than 10uA cathode current on the board level it either
becomes very expensive, very large or you can forget about a reference.

 

[Jim Thompson]

Maybe that's why there are no (cheap) ultralow brethren to the TLV431.
If you want less than 10uA cathode current on the board level it either
becomes very expensive, very large or you can forget about a reference.

Hardly a week goes by that I don't design a micro-power BandGap into a
chip ;-)

...Jim Thompson

 

[Joerg]

Hardly a week goes by that I don't design a micro-power BandGap into a
chip ;-)

Dang, I just knew you'd say that ...

For us board level guys the situation looks pretty dire.

 

[John Larkin]

Nah, the Watson Lab in Yorktown Heights. Best sandbox in the world at
one point, still very very good (and improving since the low point in
about 2000).

Cheers,

Phil Hobbs

IBM used to be headquartered in Manhattan. Then they decided that
Manhattan might be nuked by the Soviets, and it would be in the
interest of the USA that IBM survive, so they moved upstate.

John

 

[John Larkin]

Hardly a week goes by that I don't design a micro-power BandGap into a
chip ;-)

...Jim Thompson

The Intersil references are amazing. They're a floating-capacitor
EEPROM sort of cell connected to a fA follower opamp. They charge them
up to the right voltage at the factory, then ship them. They are very
stable and low noise for micro bits of power.

John

 

[Jim Thompson]

IBM used to be headquartered in Manhattan. Then they decided that
Manhattan might be nuked by the Soviets, and it would be in the
interest of the USA that IBM survive, so they moved upstate.

John

Yorktown Heights is "upstate" ?

...Jim Thompson

 

[Joerg]

Joerg wrote:




You can do a MOSIS run on old processes for a few $k. One day you should
put all of the little circuits you have always wanted but can't buy onto
one little chip, the "Joerg special". Design it into one of your products,
making sure that you hook up all of the pins, and manufacture it in the
shadiest overseas contract manufacturer that you can find. Pretty soon
there will be exact copies coming out of 15 different fabs and we'll have
an eternal supply.

But it's already there. Guys like Jim have tons of designs ready to go.
All it would take is getting it into the market. AFAICT something like a
1uA cathode current version of a common ref chip would really take off
in the marketplace, provided it's less than 10c in qties. It doesn't
even need to be adjustable of have a large compliance range.

 

[Chris Jones]

Dang, I just knew you'd say that ...

For us board level guys the situation looks pretty dire.

You can do a MOSIS run on old processes for a few $k. One day you should
put all of the little circuits you have always wanted but can't buy onto
one little chip, the "Joerg special". Design it into one of your products,
making sure that you hook up all of the pins, and manufacture it in the
shadiest overseas contract manufacturer that you can find. Pretty soon
there will be exact copies coming out of 15 different fabs and we'll have
an eternal supply.

Chris

 

[Chris Jones]

But it's already there. Guys like Jim have tons of designs ready to go.
All it would take is getting it into the market. AFAICT something like a
1uA cathode current version of a common ref chip would really take off
in the marketplace, provided it's less than 10c in qties. It doesn't
even need to be adjustable of have a large compliance range.

The 10c part is the problem - the stock market analysts probably won't
knowingly let a semiconductor company embark on a new product with less
that 50% gross margin, and that sounds hard for 10c, tested etc. Anyway
their marketing guys would get some big ideas and try to sell it for $1.50
then cancel it a couple of years later because nobody bought it. On the
other hand, maybe there is a Chinese semi mfg who would be happy to accept
a bit less margin.

Chris

 

[Chris Jones]

The Intersil references are amazing. They're a floating-capacitor
EEPROM sort of cell connected to a fA follower opamp. They charge them
up to the right voltage at the factory, then ship them. They are very
stable and low noise for micro bits of power.

John

It's must be much less than fA.

One day I want to get two of them and get one of them X-rayed a few times,
to see if that affects the voltage. I guess it can't be too bad or they
wouldn't be able to use them in anything that you would take to the
airport.

Chris

 

[John Larkin]

Yorktown Heights is "upstate" ?

...Jim Thompson

Armonk. They were probably thinking of getting out of a-bomb range,
not h-bomb range.

http://www.itjungle.com/tlb/tlb101006-story04.html


IBM has an interesting history. Read "The Maverick and his Machine",
by Maney, about Watson Sr. Another interesting book, almost all
technical, is "IBM's Early Computers" by Bashe et al... some very
weird architectures!

John

 

[ehsjr]

Yorktown Heights is "upstate" ?

...Jim Thompson

By comparison to 590 Madison avenue, it is.

Ed

 

[Benj]

Check how old the fiches are. Probably not very. But if they are really
old make sure it's not the kind of film that can spontaneously combust.
IIRC that burned down a large building somewhere in Russia where they
had stored many spools of old movies.

Fiche is NOT on nitrate film! That comes from the silent movie days or
shortly thereafter. Long before fiche was invented. Projecting a movie
in the old days was a major thrill. If the film stuck in the projector
gate the high intensity lamp set fire to it which traveled up the film
to the upper reel which thence set the whole place on fire! (nitrate
film REALLY burns fast!) Me and a buddy once set a reel on fire in his
driveway just for fun. Made an amazing fire!

 

[JosephKK]

Joerg [email protected] posted to
sci.electronics.design:

Hello Folks,

Happens a lot these days, last time an hour ago: Someone is looking
for an analog/mixed signal engineer (this time low power design). I
could do it but they absolutely want to have someone on staff. Which
I can't do. So, I often try to convince them to settle for a
youngster who gets coached now and then, instead of sitting there a
year from now still trying to find the perfect candidate.

Which US or Canadian university lets off the best analog/mixed EEs?
I know, I know, many can't even solder etc. It ain't like it used to
be. But there has got to be an alma mater that sticks out. Or maybe
a particular institute at one. And please, no pissing contests.

Once upon a time a school called Harvey Mudd did this well. Maybe
they are still at it.
I would consider U-Cal and Cal-State unlikely producers of even solid
based trainables. The real resume as opposed to the normally
prescribed one for getting through HR can really count. There are
still a few non-degreed engineers as well.

 

[Rich Grise]

Fiche is NOT on nitrate film! That comes from the silent movie days or
shortly thereafter. Long before fiche was invented. Projecting a movie in
the old days was a major thrill. If the film stuck in the projector gate
the high intensity lamp set fire to it which traveled up the film to the
upper reel which thence set the whole place on fire! (nitrate film REALLY
burns fast!) Me and a buddy once set a reel on fire in his driveway just
for fun. Made an amazing fire!

Nitrocellulose == guncotton.
http://en.wikipedia.org/wiki/Nitrocellulose

In Jr. High or so, I had a buddy who made some. It does burn real good!
;-)

Cheers!
Rich