I regularly use SPICE and other Simulation Software (SystemView, MathCad, CC
Studio, Vissim, FEMM, etc). Moreover, I've worked with a lot of scientists
and first-class engineers who also take advantage of these tools. Therefore,
the comments in this thread which admonish against the use of SPICE leave me
flabbergasted.

I recall the genesis of this thread was the question:
 "............I am really just curious as to what things people run sims on,
how complex those circuits that are simulated are, and if the tests are
worth while?
Thanks,  Afshin"

Gary Packman (6/29/02) gave an objective answer to this question. To further
address the question, the following is a small compendium of actual examples
from my own experiences:

 --Replicate a simulated D/A glitch in SPICE, simulate the prior fixes to
see why they didn't work. Then devise and simulate the fix that did work.
The application was a stroke generator for a CRT display. The company spent
a lot of time and money on expensive consultants who applied intuitive
"brute-force" fixes that didn't work, but actually aggravated the problem.
The answer was very simple, but so subtle that it took a SPICE simulation to
grasp, appreciate, and finally demonstrate its efficacy. A six month
headache was resolved in a couple of days, and most importantly, restored
customer goodwill. An almost trivial application of SPICE brought me a
healthy bonus.

 --A lot of out-of-spec active filter analyses. Judicious parameter
adjustments and low-drift, good accuracy component selection corrected
out-of-spec performance without the need to mangle existing PCAs via
trial-and-error. SPICE is ideal for this type of analysis.

 --Update the design of an old single-wire-loop comm module that replaced
expensive transformers with cheaper, smaller, better electronic devices, and
replaced obsolete analog circuits with DSP functions. This was a complex
simulation which used SPICE in conjunction with other tools. (I certainly
didn't want to use my trusty TI-51 programmable calculator to determine the
coefficients of the DSP FIR filters!).

 --Analyze several competing approaches (for a research project) to
implement extremely low level signal processing circuits (in the presence of
known noise, also simulated). This took 3 days to simulate as opposed to
about an estimated month to build and test. A prototype was built and
worked. A major implication of this is that breadboards (and their haywire
problems) can be demonstratively bypassed for most designs.

 --Show the feasibility of the implementation of a novel, advanced concept
for a "leading edge" product that a (then potential) client wanted to
develop and market. This type of simulation is purely "functional"; it is
also extremely useful for the generation of new business. The simulation
showed that this concept worked. But others which didn't work were thrown
out early, saving considerable time and money, both for me and my client.

I could fill several pages with this stuff, and most of the effort was
highly worthwhile, but instead I'll ask that readers consider that virtually
no modern chip would exist today without simulations (yes, some is even
SPICE). Consider SoS (System on Silicon) and similar devices...how else
could you breadboard/prototype these other than through simulation?

Finally, regarding the story below : " BTW, in engineering school I knew
some straight-A students that did nothing  but simulation.  When they
actually built the circuits, none of them ever  worked as desired.
.......<snip>...... At his request, I looked over his circuit.  He had
connected the transistors wrong.  I pointed this out, and his response was
"shouldn't those transistors work any way you connect them?"."

Straight-A students, indeed...!?  Long ago, when I was first introduced to
computer simulation tools, I was forewarned: "A fool with a tool is still a
fool."

Best regards,
Fred A Rupinski

----- Original Message -----
From: "Bagotronix Tech Support" <[EMAIL PROTECTED]>
To: "Protel EDA Forum" <[EMAIL PROTECTED]>
Sent: Tuesday, July 02, 2002 10:33 AM
Subject: Re: [PEDA] SPICE sim question


> I would do more simulation if there were more and better models for the
> chips I want to use.  Also, if the simulator software was easier to use
(not
> talking about just Protel here).
>
> A lot of the designs I work on are mixed-signal, not strictly analog or
> digital.  It's not easy to simulate these types of systems.  Of the analog
> components, very few have good models.  For example, I could use a generic
> diode statement and tweak the parameters, but I DON'T WANT TO.  I should
be
> able to get accurate SPICE model tailored for that particular diode from
the
> manufacturer.  But they don't offer them, because they think it is a
> "trivial exercise left to the reader".  Of the digital components, there
are
> IBIS models for some fancy chips, but not for commodity standard logic.
>
> To gather and generate all the models I need would take a lot of time.  I
> can't justify it for the return on time investment.  I use a simple method
> of manual simulation:  derive the equations for various parts of the
> circuit, and apply worst case values to them.  Find out what is out of
> bounds, and tweak the design until every component is happy and all the
> specs are met.  All I need for this is the fundamental circuit formulas
> (Ohm's law, energy, power, inductance, capacitance, RC, RL, RLC formulas,
> etc.) and my trusy HP-28C calculator.  Why, just a couple of months ago I
> designed a custom power supply for a customer.  The supply had some truly
> bizarre requirements, such as:  constant current into load (voltage varies
> over a 2:1 ratio), able to run on half-wave AC, 90%+ power factor, up and
> running within 40 ms, etc.  Did I run SPICE simulations - heck, NO!  Guess
> what:  it worked on the first try (no smoke)!  Lucky?  I'd like to think
> it's because I know what I'm doing.  So, good designers (like me, I hope)
> don't necessarily use simulation.  But if we don't, it's not because we
> don't believe in the value of it, but because it's too much trouble.
>
> BTW, in engineering school I knew some straight-A students that did
nothing
> but simulation.  When they actually built the circuits, none of them ever
> worked as desired.  It seems that there is more than "book larnin' "
> required ;-)  I didn't make A's in all my courses, but all my circuits
> worked.  I guess playing around with electronic junk since the age of 8
had
> some advantages.  There were 2 other guys that were electronics whizzes.
> Some of the A students came to us for help.  One had an interesting design
> methodology:  he would plug resistor values into the SPICE circuit and
keep
> simulating it until it worked.  Then he built it on a solderless
breadboard.
> Guess what - it didn't work.  At his request, I looked over his circuit.
He
> had connected the transistors wrong.  I pointed this out, and his response
> was "shouldn't those transistors work any way you connect them?".
>
> Best regards,
> Ivan Baggett
> Bagotronix Inc.
> website:  www.bagotronix.com
>
>
> ----- Original Message -----
> From: "Gary Packman" <[EMAIL PROTECTED]>
> To: "Protel EDA Forum" <[EMAIL PROTECTED]>
> Sent: Tuesday, July 02, 2002 5:54 AM
> Subject: Re: [PEDA] SPICE sim question
>
>
> > Mira,
> >
> > The few good designers I have know in my 25 years of design *always*
> > simulate their designs.  Engineers who cannot design a functioning
> > product without a simulator shouldn't be engineers in the first place.
> > However, every engineer that takes simulation seriously can *improve*
> > any design with simulation. I don't consider simulation a debug tool,
> > rather an optimizing tool.   Have you ever purchased a bad piece of
> > electronic equipment. If yes, this underscores the importance of
> > simulation.  If no, you have had the good fortune of purchasing gear
> > that probably received a thorough regimen of simulation before going
> > into production.  Do you think that camcorder you take so much for
> > granted just fell out of somebody's schematic generating package onto
> > the production line?  People who argue that simulation takes too much
> > time to be worthwhile are, in my experience, not proficient at
> > simulation.  It takes me less time to simulate a circuit than it takes
> > to fill out the purchase orders for the parts that go into the
> > prototypes.  It does takes time to become proficient and you need tools
> > that are easy to use.  Pspice is easy, Protel is not (probably because I
> > haven't taken the time to learn properly, I already know PSPICE so why
> > bother with Protel). Also, with more experience you learn to recognize
> > which portions of a circuit actually merit simulation.  Voltage
> > dividers, not necessary. High-gain multi-stage process control
> > amplifiers that must operate in a plus or minus 100 degree temperature
> > environment, a must.
> >
> > Gary Packman
>
>
>
> ************************************************************************
> * Tracking #: 99992F2219A8D846B5BA9C6BEDB07DADA4164ADF
> *
> ************************************************************************
>

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
*
* To leave this list visit:
* http://www.techservinc.com/protelusers/leave.html
*
* Contact the list manager:
* mailto:[EMAIL PROTECTED]
*
* Forum Guidelines Rules:
* http://www.techservinc.com/protelusers/forumrules.html
*
* Browse or Search previous postings:
* http://www.mail-archive.com/proteledaforum@techservinc.com
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to