SV: Re: SV: Re: SV: Re: [sdiy] Simulating SW?

[Magnus Danielson]

From: karl dalen <dalenkarl at yahoo.se>
Subject: SV: Re: SV: Re: SV: Re: [sdiy] Simulating SW? 
Date: Mon, 21 Nov 2005 00:40:03 +0100 (CET)
Message-ID: <20051120234003.55875.qmail at web25508.mail.ukl.yahoo.com>

> 
> --- Magnus Danielson <cfmd at bredband.net> skrev:
> 
> > From: karl dalen <dalenkarl at yahoo.se>
> > Subject: SV: Re: SV: Re: [sdiy] Simulating SW?
> > Date: Sun, 20 Nov 2005 21:29:38 +0100 (CET)
> > Message-ID: <20051120202939.64930.qmail at web25510.mail.ukl.yahoo.com>
> > 
> > > 
> > > > For anyone new to simulation... PLEASE thake this advice. Run >the
> > simulation
> > > AND build the circuit, and compare.  If they >don't look alike, change the 
> > > simulation
> > > > models until it really works like the real thing.
> > > 
> > > Was'nt the whole idea with the simulating
> > > hoopla to actually do the opposite! ;)
> > 
> > Yes, yes... and you know that very well. ;)
> 
> No, cant be! ;)
>  
> > The point is that don't trust your simulation if your models is > too weak to
> > model what you see, but when they start to make sense (and you will learn to
> > do
> > that more straight-forward as you have gained experience) then the simulation
> > tool can be more trusted than in the start.
> 
> How can a model be "more" trusted rather then "less" trusted
> in the start? I would find a simulation/or rater useless if
> all it can provide is a rough aproximation of what i see of
> the plastic breadborad simulation! ;)

Well, if you have simulation models which already proved themselfs to be more
or less right, then you have a model you can trust more than someone which is
untested. When you have a set of models which prooved themselfs, and also made
a simulation considering the non-ideal situations on your breadboard such as
parasitic resistance, capacitance and induction, then you probably can use the
result, otherwise you are potentially in for alot of work to make it work as
expected. But to people think the simulators are "the truth" and therefore get
corny results... ah well...

> Why bother write a FFT software that shows malfunctions?

Which is indeed a good point. The trouble is that FFT is king to alot of
people, who don't even know they can't trust it... sigh!

> > If you *really* understand what you are doing, you don't need
> > to simulate, 
> 
> Speaking of contradiction in terms! ;) 

Well, indeed. Simulation can be a good tool to learn things, but in the end it
is just another little tool in the toolbox. I rarely simulate things myself.
I think that was 5 years ago or something... ;O)

> > but you may occasionally simulate in
> > order to verify in the middle of the night that you are roughly > right.
> 
> Speaking of "really" contradiction in terms! :)) 
> Who? Me or the female? :))

Hey! Those continous curves caught my attention, OK??? ;O)

> > I don't know why I see the picture of Bob Pease tossing his old >computer
> from  the top of Nationals parking-garage building motivating that now THIS 
> computer  will not lye to him again or the use of the simulation output as
> papper in
> > the  birdcage with the comment that SPICE printouts is usefull for at least
> > something... simulation is a bitch if you don't know when you can trust it!!!
> 
> Unfortunately, i have to pop your baloon!
> http://www.national.com/profile/os.cgi?EventID=101805 
> 
> Simulation is still a bitch, no doubth!

Indeed. That National has those seminars is one thing, what Bob thinks about
them is a totally different issue. ;O)

> > > KD (who rather use the plastic strip/bread board to simulate) ;)
> > 
> > That is not simulation, that is doing a test-build, which is
> > certainly a
> > recommended method of development when you can do it.
> 
> It was kidding with Harry!

Yeah, yeah... the hairy Harry... ;O)

Cheers,
Magnus - hacking away on a little filter simulation tool in the middle of the night! :P