Thinking process involved when designing analog electronic circuits

[Eeyore]

I have been looking for a book that would explain the basic principles
involved in designing electronic circuits (i.e. what a person should
know, how the person should think and so on).

There are many books that seem to explain how analog or digital
components function and as far as the digital ones are concerned, how
to even group them together to create purely digital circuits. But I
have been looking for something that explains the process of placing
analog components together to create an analog circuit.

First off, do you plan to design with discretes or ICs only. That makes a
big difference. And what frequencies are of interest to you (including
DC).

Graham

 

[Ken S. Tucker]

Years ago I was a "Quotations Specialist".

An RFQ (Request For Quotation) lands on my desk,
and me jab was to configure a system meeting the
RFQ specs, then prepare a Quote, using as far as
possible *off the shelf* boards, and prepare a block
diagram, and spec a new board if necessary that I'd
sketch and send to engineering.

I was meticulous about my Block Diagrams that would
go to the customer as part of the Quote, so I'd draw
(draft) them myself. I figured rightly that the engineers
evaluating the Quote would focus on the BD, because
the 10-100 pages extra were spec details and legalities,
the meat was in the BD.

Well the boss (and the drafting dept) were getting a bit
pissed off at me for dwelling on the BD drawing, but I
needed to do that because 5% of my brain was drwing,
90% was studying the interaction and anticipating the
engineers questions, sometimes in conference call with
3 at a time for a few hours, and 5% cigarettes and coffee.

My strategy worked well, it helped to increase sales,
because customers like pictures, especially with
doddlates...a picture is worth a 1000 words.

I could Quote to $20K on my word, to $200K checked
by the boss, and $2M checked by the boss+prez.

All done using a self feeding pre-leaded auto pencil
aka a "sharpie", with an eraser on the end.
Ken

 

[JeffM]

Years ago I was a "Quotations Specialist".
[. . .]Block Diagrams[. . .]
I could Quote to $20K on my word, to $200K checked
by the boss, and $2M checked by the boss+prez.
All done using a self feeding pre-leaded auto pencil
aka a "sharpie", with an eraser on the end.

It sounds like *your* Sharpie was erasable.
This is what most folks mean when they say "Sharpie":
http://www.google.com/images?q=Sharpie
(Felt-tip; permanent ink.)

My strategy worked well, it helped to increase sales,
because customers like pictures, especially with
doddlates...a picture is worth a 1000 words.

You're going to have to explain "doddlates".
(I'm assuming you spelled it correctly.)

 

[Nico Coesel]

[snip]

Also, one thing Spice will NOT do is model reality. Just someone's
simplified model of it.

Ummmmh? IMNSHO, transistor models, particularly MOS devices in an
ASIC, are quite well modeled!

And for a good reason! But I guess the interconnection in an ASIC are
modelled as well.

Yep. Simulators don't "design".


Nonsense! Drop by sometime and I'll show you my etchings, errrh, uh,
libraries ;-)

I bet there is a lot of nifty stuff in your libraries.

If absolute tolerances have any major effect on your circuit's
performance you're NO designer ;-)

Tolerances always have some effect on your circuit. It just depends on
how much of that effect you tolerate. One of my electronics teachers
once said: "electronics is about cancelling errors and tolerances".

 

[Ken S. Tucker]

Hi John.

Lately I've been sketching block diagrams on my whiteboard,
photographing them, and emailing them to prospective customers as a
"proposal." You can even see the markers in the tray below the board.
They are typically amused, and usually buy.

One day, a few years ago, I flew to a customer site in Rockford,
Illinois, to discuss a possible new arbitrary waveform generator. I'd
been scribbling block diagrams for myself on the plane, on a quadrille
pad, in rough weather, and they were barely legible. I walked into a
big conference room with about 20 engineers and programmers and
managers. Somebody saw my papers and grabbed them and made them into
overhead projector slides, and started showing them. We'll be selling
our 500th unit in a few months. PowerPoint and glitzy graphics mostly
put people to sleep.

Congrads! You know any half-ass system has 1,000's
of mathematical calculations, down to Ohm's Law.
Can you imagine presenting a system on those
calculations, LOL, ask for bids on every single circuit,
might work, in a room full of Vulcan's, but that's anthemic
to how humans make decisions.
Go with a BD but be absolutely confident of the details.
Ken

 

[Ken S. Tucker]

Years ago I was a "Quotations Specialist".
[. . .]Block Diagrams[. . .]
I could Quote to $20K on my word, to $200K checked
by the boss, and $2M checked by the boss+prez.
All done using a self feeding pre-leaded auto pencil
aka a "sharpie", with an eraser on the end.

It sounds like *your* Sharpie was erasable.
This is what most folks mean when they say "Sharpie":http://www.google.com/images?q=Sharpie
(Felt-tip; permanent ink.)
Understood.

My strategy worked well, it helped to increase sales,
because customers like pictures, especially with
doddlates...a picture is worth a 1000 words.

You're going to have to explain "doddlates".
(I'm assuming you spelled it correctly.)

LOL, alternatively doodle-lates, are derived from a secret
ancient sanskit language dated back to the 1960's.
It uses symbols like "~" and "==" or "+" but it's secret ;-).
It's function is to communicate.
Ken

 

[krw]

Years ago I was a "Quotations Specialist".
[. . .]Block Diagrams[. . .]
I could Quote to $20K on my word, to $200K checked
by the boss, and $2M checked by the boss+prez.
All done using a self feeding pre-leaded auto pencil
aka a "sharpie", with an eraser on the end.

It sounds like *your* Sharpie was erasable.
This is what most folks mean when they say "Sharpie":
http://www.google.com/images?q=Sharpie
(Felt-tip; permanent ink.)

My strategy worked well, it helped to increase sales,
because customers like pictures, especially with
doddlates...a picture is worth a 1000 words.

You're going to have to explain "doddlates".
(I'm assuming you spelled it correctly.)

I assumed he meant "doodles".

 

[Ben Bradley]

I have been looking for a book that would explain the basic principles
involved in designing electronic circuits (i.e. what a person should
know, how the person should think and so on).

There are many books that seem to explain how analog or digital
components function and as far as the digital ones are concerned, how
to even group them together to create purely digital circuits. But I
have been looking for something that explains the process of placing
analog components together to create an analog circuit.

Does anyone know if "Electronics - Circuits and Systems" by Owen
Bishop is good in this regard, or is there something better?

Thanks in advance

I've got many of the books others listed, though I don't recall Bob
Pease's "Troubleshooting Analog Circuite" mentioned - IMHO, it's as
much about design as anything. And unlike so many of his 'porridge'
articles, the book is (best as I can remember) quite narrowly focused
on the field of electronics.

I'm far from a hotshot designer, but one thing that has given me
ideas is paging through collections of schematics (preferably with
explanations, but often not needed). Any book titled "Encyclopedia of
Electronic Circuits" - there appears to be seven volumes with that
title from Rudolf F. Graf and/or others, but I only have five of them.
I also see an unrelated book with that title.

Many of these schematics are quite simple things you might find
from the Radio Shack/Forest Sims "Engineer's Notebooks" (not that
there's anything wrong with those things), and heck they may have come
from there, but many are from the ED "Design Ideas" section and such
as well. But they all give ideas of how to use transistors in various
configurations, things I might not have seen before or haven't seen in
a long time, and that might be a good fit for some design problem.

Also, "High SPeed Diital Design" because it shows how at some
level, digital design is really analog.

 

[JosephKK]

0a) get some measurement equipment


Referring to point '0a' it is helpfull to see what electrons do in a
circuit

I absolutley agree!
And the instrument of choice is here the oscilloscope.
Also, a volt and amp meter...
But a scope can be all that.
So that should be your first [hobby] project: Design an oscilloscope.

Either you do not have the right smiley for that or your suggestion is
out of line. 'scopes am full of tricky stuff. The whole point of
[hobby] kits is to have easy early success.

 

[JosephKK]

Besides the many good suggestions from others, the very best way to
learn electronic design is with SPICE. The days of designing with
pencil and vellum are over.

If they are ever going to be any good, they must build a few things. I
have tried to teach SPICE crippled fools too often.

And the best spice is free. Get LTspice
and join the Yahoo forum at

http://groups.yahoo.com/group/LTspice/

LTSpice has good value. It is far from the best. The included
libraries are obviously targeted at increasing sales. Nor does it
have even nearly adequate modeling tools. Nor does is it conveniently
do hierarchical designs.

Get all the help files you can find and learn the terminology. Go
through the FILES folder in the forum and try all the examples. Some
may not make much sense in the beginning, but eventually things will
start to gel and come together. When you have learned enough about
spice to enter schematics yourself, do the examples in

http://www.ecircuitcenter.com/index.htm

This will get you up to speed faster than anything else I know.

With spice, you don't have to worry about destroying components due
to a miswire or other mistake. You can make changes and analyze them
much faster than with hardware.

You don't have problems with grounding, crosstalk, bypassing,
ringing, scope probe loading, probe resonance and ringing,
intermittent connections, component variations, bad components, poor
connections, power supply ripple and noise, offsets due to thermal
drift, interference from SCR dimmers and fluorescent lights,
uncalibrated or bad test equipment, and a host of other problems
when trying to implement a circuit in hardware. And the circuit will
work exactly the same way every time, so you don't have to waste
time trying to figure out what changed since the last time you
turned it on.

Quick translation, no smoke when you make a mistake, no reward when
you correctly handle the real world. Such is simulation.

Learn how a circuit is supposed to work first, then you can diagnose
and solve the other issues much easier.

For a lot of people that is far quicker learned experimentally than
through countless imaginary situations.

Spice is so crucial to electronics that I ask prospective engineers
and technicians to bring along their favorite LTspice files for the
interview. If they don't have any, I can't afford to waste time
having them learn it on the job.

You would not miss some good candidates is you more generically
required SPICE files with _all_ device models included. You would
probably catch some superior ones this way.

 

[Tim Williams]

One day, a few years ago, I flew to a customer site in Rockford,
Illinois, to discuss a possible new arbitrary waveform generator.

Fun -- just a couple ten miles south of me. (Not at liberty to divulge the
party, I suppose?)

Tim

 

[JosephKK]

...

Oh ... a 'lectric eraser! My dad was a civil engineer, before
computers. Drafting table, ink on vellum, etc & a 'lectric eraser.
What I don't see in the pic is the brush - anybody who's serious about
erasing has a brush. Yours must be out of sight.

<G>,
Bob

Not to mention tools like measuring sticks (triangular scale),
parallels, PMP arm, and templates for speed. By the time i got to do
serious drafting, many years ago, they were switching out vellum to
chronoflex. And xerox type photocopiers were making inroads on the
ozalid process at 22 by 34 and 24 by 36 inches.

 

[JosephKK]

[email protected] says...>
[snip]
Also, one thing Spice will NOT do is model reality. Just someone's
simplified model of it.

Ummmmh? IMNSHO, transistor models, particularly MOS devices in an
ASIC, are quite well modeled!

They'd better be or *nothing* works. Since you rely on cross-chip
tracking the models have to be quite good.

"Cross chip tracking"? Ain't no such animal. Only local (accurate"
matching. Special efforts are necessary to send references around the
chip.

Sure there is. Tracking may be a function of distance and
orientation, but there certainly is tracking across a chip. There
is tracking across wafer, as well, though obviously not as strong.
Models can (and do) incorporate any or all such tracking. It's
pretty hard to do manually though.

I was talking of device libraries. Board designers generally don't do
circuit design.

Tell Larkin that. ;-)

"Absolute tolerance" means just that... resistor absolute value
tolerances may be as high as 30%

Or perhaps 2-5% for board designers. Tracking can still be fun.

"Tracking", in terms of temperature, yes. More important is "ratio"
matching, which can be made to 1% (or less with some special effort).

Ratio matching is part of tracking. There are still tolerance
terms.

Or you can be clever, and make a 10-bit ADC, from 3% resistors ;-)

...as long as you have .1% tracking somewhere.

The need for that tight of tracking is topology dependent. Not really
needed for any smart choices of topology.

Samplers can get a little tricky.

Perhaps, though money matters to the owner. I try to keep in mind
that I'm spending his money.

Sure, after 5 to 6 figures of NRE. Kind of makes it real important to
have good models and get it right before committing to silicon.

 

[JosephKK]

Besides the many good suggestions from others, the very best way to
learn electronic design is with SPICE. The days of designing with
pencil and vellum are over. And the best spice is free. Get LTspice
and join the Yahoo forum at

http://groups.yahoo.com/group/LTspice/

Get all the help files you can find and learn the terminology. Go
through the FILES folder in the forum and try all the examples. Some
may not make much sense in the beginning, but eventually things will
start to gel and come together. When you have learned enough about
spice to enter schematics yourself, do the examples in

http://www.ecircuitcenter.com/index.htm

This will get you up to speed faster than anything else I know.

With spice, you don't have to worry about destroying components due
to a miswire or other mistake. You can make changes and analyze them
much faster than with hardware.

You don't have problems with grounding, crosstalk, bypassing,
ringing, scope probe loading, probe resonance and ringing,
intermittent connections, component variations, bad components, poor
connections, power supply ripple and noise, offsets due to thermal
drift, interference from SCR dimmers and fluorescent lights,
uncalibrated or bad test equipment, and a host of other problems
when trying to implement a circuit in hardware. And the circuit will
work exactly the same way every time, so you don't have to waste
time trying to figure out what changed since the last time you
turned it on.

Learn how a circuit is supposed to work first, then you can diagnose
and solve the other issues much easier.

Spice is so crucial to electronics that I ask prospective engineers
and technicians to bring along their favorite LTspice files for the
interview. If they don't have any, I can't afford to waste time
having them learn it on the job.

Without spice, you will find many people who are highly skilled at
bs in electronics. With spice, there is no faking it. You very
quickly find if they know their stuff or not.

So put most of your effort into learning spice. It will pay handsome
returns later.

Best Wishes

Mike Monett
pstca.com

Just on the side, i like nice schematics so i use CAD. With the right
packages SPICE and some other simulators are included (or more often
the other way around). Either way, i get the tools together. Though
i can do it with a pencil and paper when i desire. As noted elsewhere
in this thread, the tools do not do the design, they are just tools
and have limitations.

 

[JosephKK]

Lord, those were the years. I was using a Model 33(?) TTY machine and GE
timeshare to do some microstrip and quarter-wave pin diode attenuator
calculations - the "modem" was a box that you put a telephone handset into
for acoustic input/output coupling. As I vaguely recall, we either used
Fortran or the original Dartmouth Basic as the language. It was either that
or go up to the computer department ("IT" hadn't been invented yet) with a
stack of punch cards and do the debugging of each individual instruction by
hand...eccchh.

I also had a machine in my office, about the size of a desk phone, that used
a plain old Rat Shack cassette tape as the storage device. WOrked fine if I
had half an hour to load and execute each program.

Jim

Must have been situational, at least i had serial line 300/1200 baud
modems when i started. A few years later i put a new (at the time)
2400 baud modem on my first PC. I downloaded the Clarkson uni packet
driver collection (about 1/2 megabyte) long distance, over night,
about $100 phone bill. Bless FTP.

 

[JosephKK]

As others had said, it's all based on practical experience. You can
know all the theory in the world, yet still be a terrible designer.

And you *really* learn electronics when the things you design and/or
build *don't* work and you have to troubleshoot them.
So here's hoping your next project doesn't work and it's a real !@#$%
to troubleshoot!

Dave.

Hell, you almost sound like Jeff L. 'e hosts "LearnByDestroying".

 

[krw]

[email protected] says...>
[snip]
Also, one thing Spice will NOT do is model reality. Just someone's
simplified model of it.

Ummmmh? IMNSHO, transistor models, particularly MOS devices in an
ASIC, are quite well modeled!

They'd better be or *nothing* works. Since you rely on cross-chip
tracking the models have to be quite good.

"Cross chip tracking"? Ain't no such animal. Only local (accurate"
matching. Special efforts are necessary to send references around the
chip.

Sure there is. Tracking may be a function of distance and
orientation, but there certainly is tracking across a chip. There
is tracking across wafer, as well, though obviously not as strong.
Models can (and do) incorporate any or all such tracking. It's
pretty hard to do manually though.

So one still needs a lot of practical
knowledge and experience.

Yep. Simulators don't "design".


IMHO using Spice is a bit of an art in itself. All the components in
Spice are ideal but real components aren't.

Nonsense! Drop by sometime and I'll show you my etchings, errrh, uh,
libraries ;-)

Take a look at the crap that board designers tend to get stuck
with. :-(

I was talking of device libraries. Board designers generally don't do
circuit design.

Tell Larkin that. ;-)

The question is always: is
the simulation close enough to reality? Sometimes I even reverse the
process to get a simulation that is close to the measured values and
start working for there (Spice is great for investigating effects from
component tolerances).

If absolute tolerances have any major effect on your circuit's
performance you're NO designer ;-)

Depends on the definition of "absolute tolerances" is.

"Absolute tolerance" means just that... resistor absolute value
tolerances may be as high as 30%

Or perhaps 2-5% for board designers. Tracking can still be fun.

One of the
"absolute tollerances" you depend on is tracking. ;-)

"Tracking", in terms of temperature, yes. More important is "ratio"
matching, which can be made to 1% (or less with some special effort).

Ratio matching is part of tracking. There are still tolerance
terms.

Or you can be clever, and make a 10-bit ADC, from 3% resistors ;-)

...as long as you have .1% tracking somewhere.

The need for that tight of tracking is topology dependent. Not really
needed for any smart choices of topology.

You need a reference at least as good as what you're measuring.

Samplers can get a little tricky.


Sure, after 5 to 6 figures of NRE. Kind of makes it real important to
have good models and get it right before committing to silicon.

Not to mention TAT.

<snip>

 

[JosephKK]

Referring to point '0a' it is helpfull to see what electrons do in a
circuit

I absolutley agree!
And the instrument of choice is here the oscilloscope.
Also, a volt and amp meter...
But a scope can be all that.
So that should be your first [hobby] project: Design an oscilloscope.

Either you do not have the right smiley for that or your suggestion is
out of line. 'scopes am full of tricky stuff. The whole point of
[hobby] kits is to have easy early success.

Why, building a simple scope is easy.
I have done several.

OK, in that time you used a statically deflected tube,
http://www.hupse.eu/radio/tubes/DG7-32.htm
and my first one had some simple tubes, Y gain double triodes IIRC,
http://www.radiomuseum.org/tubes/tube_ecc85.html
X timebase with self oscillating penthode it was I think,
Z just an AC coupling to the grid.
Power supply not stabilised, but you could see things to way over a MHz.
Before that, I used an old TV tube, with magnetic deflection, with the EHT
created by a oscillation audio tube amp running into a car ignition coil
with a tube rectifier with the filament supplied by a battery.
Bandwidth of that scope was at least a few kHz.
And a nice big screen (> 40 cm).
Later I did a RTL version with delayed timebase, followed by a 300MHz
version (last one I tried).

So, yes it is easy to make a basic scope.
These days with a FPGA you van make a nice 100MHz digital scope:
http://www.fpga4fun.com/digitalscope.html
All very doable for the beginner, lots of documentation and help online,
quite different from the old days where you had to look for info
in old magazines etc, and basically had to adventure by trying your
own ideas.
So, DO NOT LET ANYONE TELL YOU SOMETHING CANNOT BE DONE, because
it is 'complicated'.
Doing these sort of things is way to gain invaluable experience.
Sure, these days you can buy better stuff for less on ebay perhaps,
but what have you learned?
If you are millionaire you can buy anything you want, so there you
have your 100TeraHz scope sitting on your desk, getting an erection
from it, and now what shall you do with it, if you never even wired up a transistor.
Buy ready made stuff and put it next to it, and impress your friends?
Too bad you will not know how to use that scope.

Gosh with all that wonderful success why don't you go into the
business of making hobbyist kits? Who knows, you might become the
next "Heathkit". Try it, why don't ya?

 

[Rich Grise]

...
Oh ... a 'lectric eraser! My dad was a civil engineer, before computers.
Drafting table, ink on vellum, etc & a 'lectric eraser. What I don't see
in the pic is the brush - anybody who's serious about erasing has a brush.
Yours must be out of sight.

The first time I saw one of those, I asked the guy, "Is that so you can
make 100 mistakes a minute?"

He didn't laugh.

Cheers!
Rich

 

[Rich Grise]

Hamilton Sundstrand, with a couple of P&W guys there, too. I have never
met a doofus or a jerk from either company... I don't know how they do
that.

A personnel guy who knows his elbow from a hole in the ground? ;-)

Cheers!
Rich