Where can EEs get "practical" classes to be better custom-ic designers?

[Electro Migration]

Where can EEs get "practical" classes to be better custom-ic designers?

If graduating electrical engineers wish to be considered proficient analog,
mixed-signal, or RF designers using commercial tools, what are their
options today?

Here's what I can find so far by googling and asking of others:

They start with an EE degree - then they ...
- add 3-5 years on-the-job training (i.e., design, then lead 5-10 projects)
- attend universities (e.g., MIT open university or UC Berkely extension)
- take technical training (e.g., Besser Associates or SVTII)
- build "in-house training" (e.g., hire consultants for custom classes)
- follow "trade publications" (e.g., IEEE.org journals or EEdesign)
- peruse "designer websites" (e.g., designers guide or analog ic design)
- they ???

Given it would be nice to collect pointers on how to be a better analog,
rf, or mixed-signal designer, the question is two-fold.

(1) What other "options" are there for a custom-IC designer to improve
their job-related skills?

(2) Is there a great list of "practical" design classes, instructors, and
materials available on the web that we could collect here?

 

[Electro Migration]

(1) What other "options" are there for a custom-IC designer to improve
their job-related skills?

I should clarify, I'm NOT looking for pointers on how to push buttons on
tools - nor am I looking for more theoretical knowledge like that already
given in countless universities - I'm looking for pragmatic design skills
training to augment on-the-job training already given.

That is, can you respond with a suggestion or two pointing to whatever
people or material you know of that could help graduating EEs improve their
job-related analog, rf, or mixed-signal DESIGN skills - outside their
regular job?

Where can custom-ic designers go to improve their design skills?

 

[The Master]

I'm looking for pragmatic design skills
training to augment on-the-job training already given.

I know it's not what you are asking for, but I would like to ask new
engineers to pay attention to their layout designers (assuming they don't
do their own layout). I cannot tell you how many times I have been given
a schematic, or series of schematics, that are darn near impossible to
figure out. We have an on going joke about having to "break in" a new
engineer, so they draw schematics properly.

I've actually had one time where I was given a high speed digital
datapath, but the schematics were drawn where each state was arrayed
rather then one schematic for the entire path that's arrayed. In other
words, I had schematics that was 20 inverters. Those 20 outputs went into
another schematic, and were inputs for 20 nor gates... Seriously, I'm not
making that up!

 

[sycochkn]

I know it's not what you are asking for, but I would like to ask new
engineers to pay attention to their layout designers (assuming they don't
do their own layout). I cannot tell you how many times I have been given
a schematic, or series of schematics, that are darn near impossible to
figure out. We have an on going joke about having to "break in" a new
engineer, so they draw schematics properly.

I've actually had one time where I was given a high speed digital
datapath, but the schematics were drawn where each state was arrayed
rather then one schematic for the entire path that's arrayed. In other
words, I had schematics that was 20 inverters. Those 20 outputs went into
another schematic, and were inputs for 20 nor gates... Seriously, I'm not
making that up!

That is what orcad and the likes does. You still have to do a lot of manual
editing to get it readable.

Bob

 

[Kevin Aylward]

Where can EEs get "practical" classes to be better custom-ic
designers?

If graduating electrical engineers wish to be considered proficient
analog, mixed-signal, or RF designers using commercial tools, what
are their options today?

Here's what I can find so far by googling and asking of others:

They start with an EE degree - then they ...
- add 3-5 years on-the-job training (i.e., design, then lead 5-10
projects)
- attend universities (e.g., MIT open university or UC Berkely
extension)
- take technical training (e.g., Besser Associates or SVTII)
- build "in-house training" (e.g., hire consultants for custom
classes)
- follow "trade publications" (e.g., IEEE.org journals or EEdesign)
- peruse "designer websites" (e.g., designers guide or analog ic
design)
- they ???

Given it would be nice to collect pointers on how to be a better
analog, rf, or mixed-signal designer, the question is two-fold.

(1) What other "options" are there for a custom-IC designer to improve
their job-related skills?

(2) Is there a great list of "practical" design classes, instructors,
and materials available on the web that we could collect here?

I actually have a problem with the word "training". To put it bluntly, my
view is that if "training" to do analogue design is required, one will never
be much good as an analogue designer. I can't say that I have ever had any
"training" to do analogue i.c. design.

The idea is that you learn and understand the basics that you were taught in
university. e.g. http://www.kevinaylward.co.uk/ee/index.html . That is,
although there are a few bits and pieces here and there that may not be
covered, the bulk is all volts and amps. The amount of new knowledge
required is actually quite limited. Its applying what you should already
know, e.g. cascodes, diff pairs, offset calculations, BW, stability etc, to
construct circuits that matters.

You analyse existing circuits, by *yourself* and figure out, by yourself,
why things were done that way. If you can not figure out why something was
done that way by yourself, you wont be able to design new circuits. Thats
what "training" really is in my view.

So, pick something to design. That is, find a full detailed spec for
something, and try and design it in complete detail using Spice. That is how
you learn design, by doing it. No amount of watching someone kick a ball, or
telling you how to kick a about will turn you into a David Beckham. *you*
need to the kick the ball.

 

[Mobil]

I actually have a problem with the word "training". To put it bluntly, my
view is that if "training" to do analogue design is required, one will never
be much good as an analogue designer. I can't say that I have ever had any
"training" to do analogue i.c. design.

The idea is that you learn and understand the basics that you were taught in
university. e.g.http://www.kevinaylward.co.uk/ee/index.html:-). That is,
although there are a few bits and pieces here and there that may not be
covered, the bulk is all volts and amps. The amount of new knowledge
required is actually quite limited. Its applying what you should already
know, e.g. cascodes, diff pairs, offset calculations, BW, stability etc, to
construct circuits that matters.

You analyse existing circuits, by *yourself* and figure out, by yourself,
why things were done that way. If you can not figure out why something was
done that way by yourself, you wont be able to design new circuits. Thats
what "training" really is in my view.

So, pick something to design. That is, find a full detailed spec for
something, and try and design it in complete detail using Spice. That is how
you learn design, by doing it. No amount of watching someone kick a ball, or
telling you how to kick a about will turn you into a David Beckham. *you*
need to the kick the ball.

Berkeley's class about Analog and RF design is quite good and there're
videos in the website.
Have a look at them: EE142, EE140 and EE240

 

[Joel Koltner]

I actually have a problem with the word "training".

You have a problem withi the word "training" but you suggest people insure
they understand the basics they were taught in university?

In my book, university is just another word for (relatively broad) "training."
Hence the suggestion that "if training is required, you'll never be any good
as an analog designer" is absurd; you might as well state, "if you have to go
to university, you'll never be any good as an analog designer."

Since the quality of universities varies greatly, how well "trained" anyone is
in the basics after graduating varies greatly as well. Individuals who wish
to improve their skill sets should be lauded, and while I agree with you that
a lot of "training" is of the "do it yourself" nature, I also think there's a
lot to be said for studying under the mentorship of a good designer.

---Joel

 

[sycochkn]

In my book, university is just another word for (relatively broad)
"training."

The OP wants to learn to apply that better.

The quality of universities varies greatly, how well "trained" anyone is
in the basics

Does a university no matter what the quality, cover the same basic material?

A lot of "training" is of the "do it yourself" nature.

Probably most of it.

there's a lot to be said for studying under the mentorship of a good
designer.

Is that always available to an indidvidual?

 

[Joel Koltner]

Does a university no matter what the quality, cover the same basic material?

At least in the U.S., I think it's a safe statement that the majority of
engineering schools follow ABET standards and therefore do attempt to cover
the same basic material.

How well they actually do that varies greatly, of course.

Is that always available to an indidvidual?

Unfortunately, no... although it doesn't occur to many new college grads that
it's really even an option. There was certainly something to be said for the
old "apprenticeship" style of training.

---Joel

 

[Kevin Aylward]

You have a problem withi the word "training" but you suggest people
insure they understand the basics they were taught in university?

You don't get taught in uni. Someone writes a few bits and pieces on the
board, and you copy it.

In my book, university is just another word for (relatively broad)
"training." Hence the suggestion that "if training is required,
you'll never be any good as an analog designer" is absurd; you might
as well state, "if you have to go to university, you'll never be any
good as an analog designer."

I disagree. In my view, all the really good analogue designers, essentially,
trained themselvs.

In my uni days I had, maybe, a few 1 hour classes, like, this is a cascode,
this is an emitter follower, out of 4 years of "training". Sure, I got
Maxwell's equations, digital design and shit, but that was about it. I
learnt detailed transistor level design simply by looking at circuits, and
designing them.

A "good" university e.g. Cambridge, might not even have any taught courses
at all, just like doing a PhD, you teach yourself. An instructor is only
there for basic guidance, not to teach you. Its er.. called being a mature
student.

Since the quality of universities varies greatly, how well "trained"
anyone is in the basics after graduating varies greatly as well.
Individuals who wish to improve their skill sets should be lauded,

Yes, but if you want a job done right, do it yourself. Even when I play the
guitar, if I was ever "taught" a song, I would forget it. Teaching yourself
and it sticks in. Get books, read and understand them. If you don't
understand some bit, ask someone that might know, but don't expect that a
"teacher" will be able to give you anything but the basics to get started.

The point is that in actually reality, if you are not a "bod" i.e someone
that learns this stuff on their own, in my view, you just wont become much
good as an analogue designer. That's just my experience. Like, you could
never teach me to paint. I just can't do it. Like, painters teach themselves

and while I agree with you that a lot of "training" is of the "do it
yourself" nature, I also think there's a lot to be said for studying
under the mentorship of a good designer.

The problem is that that a good designer generally don't have the time.
Analogue design takes way much, way too much to explain all the details that
actually have to be done to make a circuit work, and actually manufacturable
..

 

[Joel Koltner]

Kevin -- I do agree with much of what you're said; we're not really that far
apart, I think, in our beliefs. Thanks for the post...

One thing I'd mention is that "learning by looking at circuits and designing
them" is not as applicable today as it was in, e.g., the '70s: Circuits today
are very complex to take in all at once (you can't just open up the back of a
cell phone and figure much out...). Recall the discussion about...

-- Whether or not even the most brilliant minds of the '60s would be able to
figure out how a modern hard drive worked (not so clearcut!)
-- The case where some guy's friend wanted to make a 3D computer game and so
just started entering "code" such as "fire missiles at enemies" in a text
editor and actually thought he was "pretty close" to making something work!

If you open up electronics magazines today, the vast majority of them are
centered around microcontrollers and digital stuff which arguably is much
easier to understand and digest than, e.g., some fancy neutralized tube
amplifier for UHF. Even if you open up a magazine and see, e.g., a guitar
amplifier, in all likelihood it's using someone's all-in-one IC rather than
being a discrete design, right?

Kids who don't have some "guidance" -- and access to a good library! -- are
definitely at a bit of a disadvantage today when it comes to learning analog
design, IMO.

---Joel

P.S. -- Kudos for putting up your own web pages on analog design. Ditto to
folks like Jim, Joerg, and John who do similarly and/or spend a significant
amount of their personal time helping others to learn.

 

[Joerg]

Where can EEs get "practical" classes to be better custom-ic designers?

If graduating electrical engineers wish to be considered proficient analog,
mixed-signal, or RF designers using commercial tools, what are their
options today?

Here's what I can find so far by googling and asking of others:

They start with an EE degree - then they ...
- add 3-5 years on-the-job training (i.e., design, then lead 5-10 projects)
- attend universities (e.g., MIT open university or UC Berkely extension)
- take technical training (e.g., Besser Associates or SVTII)
- build "in-house training" (e.g., hire consultants for custom classes)
- follow "trade publications" (e.g., IEEE.org journals or EEdesign)
- peruse "designer websites" (e.g., designers guide or analog ic design)
- they ???

Given it would be nice to collect pointers on how to be a better analog,
rf, or mixed-signal designer, the question is two-fold.

(1) What other "options" are there for a custom-IC designer to improve
their job-related skills?

(2) Is there a great list of "practical" design classes, instructors, and
materials available on the web that we could collect here?

Only this way: Build stuff. Buy parts at Digikey or other places, fire
up the soldering iron, put it together, make it work. IMHO you cannot
become a good analog IC designer unless you have a lot of experience
with discrete circuits under the belt.

Nowadays I encounter a lot of fresh grads who think that mastering SPICE
and VHDL is all they need. Wrong. If someone can't solder I usually
advise my clients against hiring that engineer and keep looking.

 

[Kevin Aylward]

Kevin -- I do agree with much of what you're said; we're not really
that far apart, I think, in our beliefs. Thanks for the post...

Yes and thanks.

One thing I'd mention is that "learning by looking at circuits and
designing them" is not as applicable today as it was in, e.g., the
'70s: Circuits today are very complex to take in all at once (you
can't just open up the back of a cell phone and figure much out...).
Recall the discussion about...

I was really restricting my reply to analogue ic design itself. The circuits
are usually more manageable. I agree, that many actual products are
impossible to analyse.

-- Whether or not even the most brilliant minds of the '60s would be
able to figure out how a modern hard drive worked (not so clearcut!)
-- The case where some guy's friend wanted to make a 3D computer game
and so just started entering "code" such as "fire missiles at
enemies" in a text editor and actually thought he was "pretty close"
to making something work!
If you open up electronics magazines today, the vast majority of them
are centered around microcontrollers and digital stuff which arguably
is much easier to understand and digest than, e.g., some fancy
neutralized tube amplifier for UHF. Even if you open up a magazine
and see, e.g., a guitar amplifier, in all likelihood it's using
someone's all-in-one IC rather than being a discrete design, right?

Yes. Transistor level discrete design is all but extinct.

 

[Kevin Aylward]

Only this way: Build stuff. Buy parts at Digikey or other places, fire
up the soldering iron, put it together, make it work. IMHO you cannot
become a good analog IC designer unless you have a lot of experience
with discrete circuits under the belt.

I probably have to agree, maybe, sort of. It was my own particular path,
however, I dare say its possible in principle to become good without
discrete circuit experience. What is quite important though is getting real
experience with any physical hardware, even if it is only with your ic fabed
ones.

Nowadays I encounter a lot of fresh grads who think that mastering
SPICE and VHDL is all they need. Wrong. If someone can't solder I
usually advise my clients against hiring that engineer and keep
looking.

Although I have spent many, many, years on the bench, for the last 10 years
or so, it has been entirely in the virtual cadence world. I can honestly say
that I can design relatively complex analogue chips, entirely by computer,
and have first pass successes. However, its hard to evaluate just how much
of my prior discrete bench work contributed to this. I believe it was
significant, but I just can't really say for definite.

Generally, I find those that can't solder are not much use, because
soldering is what bods do, and in my experience, its only the bods that end
up being any good. You need to have done things on your own accord, to be
good at anything, in my view. e.g your ice skaters, violinists, etc. If mum
has to force you, their wasting their time.

 

[Joerg]

Yes and thanks.


I was really restricting my reply to analogue ic design itself. The circuits
are usually more manageable. I agree, that many actual products are
impossible to analyse.


Yes. Transistor level discrete design is all but extinct.

Nope, it sure ain't. I make a living with it. What is quickly dwindling
is the required talent pool. Because most students believe in this
extinction myth they gravitate towards chip design, FPGA, embedded or
software. A client had searched a full two years for an analog guy with
discrete design capabilities and finally had to import one. And I am
still coaching him because young grads haven't had our level of hobby
exposure.

 

[Joel Koltner]

You need to have done things on your own accord, to be good at anything, in
my view. e.g your ice skaters, violinists, etc. If mum has to force you,
their wasting their time.

In general I agree, but I do think some people are so extraordinarily bright
that they end up being very good at particular things with only, say,
pecuniary interest (rather than, e.g., hobbyist interest) -- someone like
Tonya Harding (clearly quite talented) perhaps might fit this category.

I also imagine that someone like Gary Kasparov, even if he hadn't been playing
chess all his life, could still beat most players within, say, one year of
beginning to play. It's hard to overstate just how much better someone like
Kasparov is at what he does best than the average Joe is.

---Joel

 

[Jim Thompson]

Joel Koltner wrote: [snip]

Even if you open up a magazine
and see, e.g., a guitar amplifier, in all likelihood it's using
someone's all-in-one IC rather than being a discrete design, right?

Yes. Transistor level discrete design is all but extinct.

Nope, it sure ain't. I make a living with it. What is quickly dwindling
is the required talent pool. Because most students believe in this
extinction myth they gravitate towards chip design, FPGA, embedded or
software. A client had searched a full two years for an analog guy with
discrete design capabilities and finally had to import one. And I am
still coaching him because young grads haven't had our level of hobby
exposure.

Spread the word: If at all possible use a uP... keep us old farts
working ;-)

...Jim Thompson

 

[Joerg]

Joel Koltner wrote: [snip]
Even if you open up a magazine
and see, e.g., a guitar amplifier, in all likelihood it's using
someone's all-in-one IC rather than being a discrete design, right?
Yes. Transistor level discrete design is all but extinct.

Nope, it sure ain't. I make a living with it. What is quickly dwindling
is the required talent pool. Because most students believe in this
extinction myth they gravitate towards chip design, FPGA, embedded or
software. A client had searched a full two years for an analog guy with
discrete design capabilities and finally had to import one. And I am
still coaching him because young grads haven't had our level of hobby
exposure.

Spread the word: If at all possible use a uP... keep us old farts
working ;-)

Yep

But there will come a time when we aren't around anymore. Even among
analog chip designers I can tell the difference between the old ones and
newer grads. The young folks lean heavily towards building blocks and
have a hard time with true device level stuff. In the discrete world
it's worse, only very few young engineers who would dare to jump into
transistor level at all.

 

[Joerg]

I probably have to agree, maybe, sort of. It was my own particular path,
however, I dare say its possible in principle to become good without
discrete circuit experience. What is quite important though is getting real
experience with any physical hardware, even if it is only with your ic fabed
ones.

I do not think one can become good without getting the hands dirty. Only
by building stuff can one gain a true appreciation about inductive
coupling mechanisms, EMI, inductor saturation, ESR, datasheets cooked by
marketing people, and so on.

Although I have spent many, many, years on the bench, for the last 10 years
or so, it has been entirely in the virtual cadence world. I can honestly say
that I can design relatively complex analogue chips, entirely by computer,
and have first pass successes. However, its hard to evaluate just how much
of my prior discrete bench work contributed to this. I believe it was
significant, but I just can't really say for definite.

Your bench experience is what give you a dose of reality when doing
SPICE. For example, a rookie who never saw a bench will not likely know
why and when tantalum capacitors become spacecraft.

Generally, I find those that can't solder are not much use, because
soldering is what bods do, and in my experience, its only the bods that end
up being any good. You need to have done things on your own accord, to be
good at anything, in my view. e.g your ice skaters, violinists, etc. If mum
has to force you, their wasting their time.

Exactamente. If a client asks me to interview a candidate and he or she
can't solder the interview is de facto over.

 

[Jim Thompson]

Kevin Aylward wrote:
Joel Koltner wrote: [snip]
Even if you open up a magazine
and see, e.g., a guitar amplifier, in all likelihood it's using
someone's all-in-one IC rather than being a discrete design, right?
Yes. Transistor level discrete design is all but extinct.

Nope, it sure ain't. I make a living with it. What is quickly dwindling
is the required talent pool. Because most students believe in this
extinction myth they gravitate towards chip design, FPGA, embedded or
software. A client had searched a full two years for an analog guy with
discrete design capabilities and finally had to import one. And I am
still coaching him because young grads haven't had our level of hobby
exposure.

Spread the word: If at all possible use a uP... keep us old farts
working ;-)

Yep

But there will come a time when we aren't around anymore. Even among
analog chip designers I can tell the difference between the old ones and
newer grads. The young folks lean heavily towards building blocks and
have a hard time with true device level stuff. In the discrete world
it's worse, only very few young engineers who would dare to jump into
transistor level at all.

We need to hold on to our health as long as possible.

...Jim Thompson