Re: engineers' brains

krw wrote:
In article <%g5ng.475428$xt.26537@xxxxxxxxxxxxxxxxxxxxxxxxx>,
see_website@xxxxxxxxxxxxx says...
Keith wrote:
In article <0a8o92t1j7dfqf3ukfsovlhnkdlvl8es4q@xxxxxxx>,
jjlarkin@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx says...
On Fri, 23 Jun 2006 12:06:51 -0400, Keith <krw@xxxxxxxxxx> wrote:



What a circuit designer ain't is...

A tech who gets a reference design and fiddles with it a bit.
That's OK, but it's not design because he didn't invent it and
probably doesn't understand it.

Logic designers who work only at the gate-and-up level.

Why? What's the difference between building an analog circuit
using opamps, comparitors, and DACs, vs gates? The same sorts of
skills apply.


Not if the "design" is a purely logical construct, coded most
likely in Verilog or something. There's no "circuit" and no
electrical issues involved, just boolean logic.

No, it's just that the math and tools are a little different.
Again, I don't see the difference between wiring analog blocks
together and wiring digital blocks together. Same swamp, different
alligators.

Its fundamentally different. This is it

To all intents and purpose, a software or digital specification, is
always physically realisable.

Not true. I've seen many timing specifications that cannot be met.
C, Cdv/dt, and all that.

For analogue, a given specification may be out with the laws of
physics. Analogue electronics is fundamentally restricted by:

F(sigma, speed, power) that is is it is impossible to arbitrarily
achieve power, speed and accuracy specification. e.g
http://www.anasoft.co.uk/EE/cmospafl/cmospafl.html

Same swamp, different alligators. I think Intel would love to have
a 1W 10GHz P4, too.

Your missing the point.


There are numerous specifications that all have to be achieved in the
real analogue world, not the virtual world of digital and software.
That is, the design of a digital multiplier is a routine exercise,
the design of an accurate high speed, low power analogue multiplier
with low feed through, is very difficult, or impossible.

Only if you don't have to implement it with real silicon.

And no, don't defend with yeah, software and digital has constraints,
because these same ones are also in analogue, so don't differentiate.

Realizable technology limits both realms in similar ways.

Analogue is constrained by the real world. Its all about trading off
one conflicting specification from another.

And I can have a zero-delay, zero-power, zero-size gate? Get real.

I'll take a coupla hundred million, all wired correctly, thanks.


Ahmmm. Again, you are simply missing the point. Its trivially obvious
that digital hardware may be limited in basic speed and power, but this
limitation is not relevant to the point being made about analogue.

The number of varibles to be satisfied in an anlogue design is way more
than digital. Like, why do you think digital is used in the first place,
like audio, saving data? Why don't we just do it by analogue methods?
That is, why represent anlaogue information as 0,1 in the first place?

Look, digital design is a solved problem. Given that one requires a
register, barrel shifter, accumulator, software can just go and
synthesise the damm thing. This just can't be done with analogue, and
never will. Analogue design requires conciousness. This is reflected in
the salaries companies are prepared to pay analogue engineers,
http://www.electronicsweekly.com/Articles/2006/06/22/39017/Goodanaloguedesigniskeytosuccess%2csaysIntersilCEO.htm
:-)


Kevin Aylward B.Sc.
431infoEXTRACT@xxxxxxxxxxxxx
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

"There are none more ignorant and useless,than they that seek answers
on their knees, with their eyes closed"



.

Quantcast