Ok Eric very fancy words but..............
Let us take a simple spice model, which I am sure you agree comes under the
generic term 'computer modelling'. Forgive me if I'm wrong but the spice
parameters for any device can never be exact. A simple NPN transistor for
example, the data is provided by the manufacturer who builds to a certain
tolerance and verifies the performance data of the device by laboratory
measurement to some arbitrary standard. This data is analysed and presented
as a typical set of performance parameters for that device. It cannot be
exact, so I don't think that there is any problem with semantics at all. It
was not me who originally got pedantic about terminology. I really don't
care how you use the word exact. I think that we have reached a
philosophical impasse.
Regards
Conrad
-----Original Message-----
From: amps-bounces@contesting.com [mailto:amps-bounces@contesting.com]On
Behalf Of MorgusMagnificen@aol.com
Sent: 11 February 2003 17:06
To: conrad@g0ruz.net
Cc: amps@contesting.com
Subject: Re: Spam Alert: Re: [Amps] The Philosophy of Science
Okay, I sincerely apologize for the aggressive broadside. I am just sick and
tired of hearing all of the distortions of theoretical science and
engineering that I hear EVERYWHERE . I hoped this group would have a little
more understanding of it. If your world ends at the 4th significant figure,
fine for you. For many others, the action doesn't even begin until the 6th -
or 10th.
One statement by you and others ( in some of those OTHER armchairs) regards
the term 'computer modelling'. There is somewhat of a semantic problem here,
as follows. The computer models which we use are EXACT, precise physical
devices whose electronic equations we can write precisely. We can then apply
them in circuits and solve the circuit equations to any desired degree of
accuracy. In the limit (this is a profound mathematical statement, which
forms the basis of all numerical computation algorithms) these solutions
converge to the exact answer (if the algorithm designer has not screwed
up!).
The approximation comes in when we attempt to apply this exact model to a
practical circuit. Again, the degree of agreement between the two is limited
by our ability to measure the real-world components, which we all know has
practical as well as theoretical limits. So it is not the modelling process
which is 'inexact'. The error comes from our measurment limits, which we
know, control, and can accurately predict.
The laws of physics themselves are models. I posed the very relavent
question
"is the formula R=E/I an exact model" and no one wants to take a stand on
that, the most basic of all of our electrical 'laws'. That we can approach
exactness only in the limit sense does not make it any less useful to us.
I want to close this (although I am sure you would like to conrtinue to hear
me rant) by going back to where it began, and show how all of those who have
argued against me have badly distorted the issue. It started when Jeff
posted
a very simple solution to a somewhat complex problem - the calculation of
filter capacitance in a PS. I was, like others, initially suspicious of his
results but I wanted to check it out as accurately as possible before
attacking his work. To do so, I made the most accurate calculation I could
of
the same problem, so that if I were to raise a complaint, no one could
accuse
me of basing it on an inexact calculation (i.e. an approximation, with which
the older power supply literature is filled .) So by comparison, my
calculations were so precise (let's say they produced results accurate to
.01%) that they were effectively exact in comparison to older data. To most
engineers I know, that constitutes an exact calculation. (What you may not
realize is that this 'old' data which I always refer to was based on highly
approximated models - with our modern computers we do not have to severely
approximate our models.)
Does it really change anything if I change the wording to read 'highly
precise' calculations instead of 'exact'? Would it convey any more or less
useful information to you? Would it make any difference when you finally get
back to your workshop to build your amp, for which you will be doing well to
get a filter cap that is within 10% of the predicted EXACT value?
I would like to ask for a polling by everyone reading this (if you are still
awake) on the following: Does the fact that my calculations were terminated
at an accuracy of .01%, as opposed to the known errors of 10% or greater in
old data, mean that my calculations are not exact? And if not, how precise
would I have to make them in order to qualify as a standard against which
to
measure simple approximated calculations, such as Jeff's? Does it bother you
that I use the word 'exact' in the context of "high-accuracy, so high that
its estimated error is too low to be of any concern" ?.
Eric K8LV
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|