Re: Software for a beginner to design and learn about circuits with?

From: Kevin Aylward (salesEXTRACT_at_anasoft.co.uk)
Date: 12/01/04 

Date: Wed, 01 Dec 2004 08:06:04 GMT

mike wrote:
> msv-groups wrote:
>> Hi,
>>
>> I'm interested in learning about circuitry as a hobby and to
>> eventually create a few small projects that I had in mind, such as
>> guitar pedals and a few audio devices, but obviously I have to start
>> small. I'm somewhat competent with programming/mathematics, have some
>> text
>> books, and I do know a -little- analogue+digital circuit theory, but
>> I've never had a real hands-on attempt by myself before so I'll be
>> pretty much attacking this as though I know nothing.
>>
>> Can anyone recommend some software (free or cheap) that would be good
>> for a beginner to use to learn about circuit behaviour and design
>> some circuits with? Are there programs out there that actually
>> allow you to drop in popular microcontrollers and model their
>> behaviour as well? Thanks,
>> Matt.
>
> Wait while I get up on this her soapbox.
>
> Circuit design is the process of turning an idea into an
> implementation concept into a bunch of hardware and/or software to
> realize the idea.
> For this you don't need software. You need a BOOK!!!
> Go to the library and check out a book on analog circuit design.
> Learn about Ohm's law, norton's theorems etc. Learn about rc/lc/rlc
> circuits. How to bias a transistor, FET etc. Learn how to do Laplace
> transforms. You're rarely gonna do a laplace transform, but the
> knowledge of how, gives you great insight into what's gonna happen
> when you change a circuit paramaeter or topology. It allows you to
> pick component values right out of the air that are very close to
> what you'll finally end up with after you fine tune it. It tells you
> instantly when you're way off base and need a different approach.
> If you need a calculator, you're going too deep. Your objective is to
> be able to scratch out a circuit topology, estimate bandwidths, gain,
> impedance, signal fidelity etc. Learn about component parasitics.
> Unless you're doing audio work, a resistor is not just a resistor. And
> it goes downhill from there. Learn how real components have
> parasitics and vary relative to their specifications and how to
> mitigate the impact of those variations on your objective.

All good general advice.

>
> Learn about Fourier transformations betwee time and frequency domains.
> Again, you're rarely gonna do one by hand, but knowing how gives you
> great insight into topologies and components required to realize your
> design.

All good general advice.

>
> Now, go back to the library and checkout a book on digital circuit
> design.

Nope.....

Digital is for those just to stupid to do analogue:-)

>Learn that ALL curcuits are analog and that most digital
> problems have to do with the ignored analog characteristics of the
> digital simplification. Learn about Karnaugh maps, glitches, races
> etc.
> Now, you know how to design circuits. It's time to start looking at
> software to simulate the circuit you've already designed to fine tune
> it's parameters and take into consideration the simplifications you
> used to design it in your head or with a pencil.

I can't agree at all. This is fundamentally wrong in my opinion, and
that of probably every educator known to man. If this were true, there
would be no lab work until 3rd year of uni. The best way to learn is to
do little bits of theory at a time complemented by practise to reinforce
that theory. A simulator is indeed *real* practise, excluding the few %
issues like parasitics. Too many here are going, well because a
simulator doesn't copy everything exactly from the real world, then it
is sadly lacking for real world learning. This is very narrow minded.
*Most* of what a simulator can do is very good model of the real world.

Spending several years doing the "basics" before even getting started on
trying out the theory is quite daft. One would simply forget what one
first learned. To all intents and purposes, there is no practical
difference between examining ohms law and I=cdv/dt on the bench or in a
simulator. Your brain, eyes and ears are processing essentially, the
same information. There is only this vague psychological issue that some
have about it "not being the same". Essentially, it is.

>
> Circuit simulation replaces the tedious VERIFICATION calculations that
> we used to do by hand.

Unfortunately, in a real design of 100s to 1000s of components, hand
calculations are impossible. Even a one transistor circuit has no exact
analytical solution.

> It does NOT replace the thought processes needed to design circuits.

It can do. This is the reality of how it works in the real world, not
how we kid ourselves it is or should be. Yes, you need a good
understanding independent of simulations, however, it is simple
impossible to understand all of the nuances of a design without some
constructive trial and error on a simulator. Yes, most don't like this
idea, but it is the way it truly is. To illustrate. Consider the design
of a nuclear bomb. These require amazingly powerful computers. Sure,
they are run by *experts* , but even these 20 year experts cant
understand the implications of all of the equations. Thery are simply
too complicated. Or for example, colliding black holes. There is not a
chance to develop a *feel* for happens *without* simulations.

Sure, I agree that without bench work there are many issues that one
doesn't get a feel for in a simulator, but *likewise*, without a
simulator there are *many more* things that you never get a feel for by
bench work. As I said, I've been there 10000's of hours on both the
bench and simulator. I can count countless things that I have missed on
the bench that the simulator showed.

>
> You have only to read the archives of this newsgroup to see countless
> examples where people stuffed random numbers into a simulator and came
> up with something that simulated, but was impractical, unrepeatable,
> or just plain sad.

But this is unintelligent trial and error. Intelligent trail and error
is another matter entirely. *All* design can be reduced to a Darwinian
process, i.g. replication of existing circuits, varying such designs,
and selecting the good ones. Whether this is done on paper, in ones head
(a Darwinian process) or on a simulator don't much matter.

Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

Quantcast