On May 10, 7:45*pm, tom wrote:
On 5/10/2010 9:34 PM, Ralph Mowery wrote:





The computer program should know its limits. *Anytine a program allows the
data entered to be too large or small for the calculations, it should be
flagged as being out of range. *Also many computer programs will use
simplified formulars that can mast the true outcome. *Usually it is not very
much, but as all errors start to add up the end results may be way off.

I often enter data that I know will be difficult for programs to use. *If
the program gives an answer then I usually don't use that program expecting
a exect answer.
Back in the Windows 3.1 and 3.11 days the simple calculator would give wrong
answers to simple problems. *I think if you entered 3.11 and subtracted 3.1
from it you got the wrong answer. *That program was not corrected by
Microsoft.

I disagree. *The program cannot "know" its limits if the problem it's
modeling is complex enough. *So the user must understand the program and
especially the math related to what the program is modeling.

Blaming the program for giving you the "wrong" answer is like blaming
the tires for hitting the guard rail because you exceeded their limits.
* Those limits are not the same under varying conditions and must be
filtered by experience and understanding.

tom
K0TAR

I've found it in my best interest to check the consistency of results
in various ways, whenever I can. Often there's more than one way to
think about a problem, and if the answers I get differ, I want to know
why. Until I can resolve the differences, I distrust both (or all...)
answers. I also like to have an idea about the tolerance on the
answers, and many programs (and formulas you use to calculate answers
for yourself) don't give much of a clue about the tolerance. Some are
"exact," and some should be considered only approximations, but often
they don't bother to tell you which. One example is formulas for
calculating the impedance of TEM transmission lines; it's common to
see, for air-dielectric two-wire line, Z0=276*log10(2D/d), but this is
an approximation whose error becomes significant as d approaches D.
Even the better formula, Z0=120invcosh(D/d), is not exact: the 120
isn't exactly correct, there's no consideration of finite conductor
resistance (and resulting skin depth), and there's no consideration of
the atmospheric pressure and relative humidity...

I mostly agree with Tom: don't expect the program, or formula, to
know how you are going to misapply it. Try to be aware of what the
answers you get imply. Learn the limits of your tools (programs;
formulas), and apply them wisely so they will serve you well.

Do I get stung by my own foolishness in not paying proper attention to
things like this? You bet I do! Just last night, I entered a coil
into the Hamwaves inductance calculator and it was happy to give me an
answer. The coil? Ten turns of 1mm wire in a coil 10mm diameter and
10mm long... Duh, that's a 1mm winding pitch and the turns will short
together. I didn't think to check that at first. The calculator
complains and won't give you an answer if the pitch is less than the
wire diameter, but not if it's just equal. Considering the same very
useful inductance calculator, I've learned to ignore the answer for
the effective shunt stray capacitance: it in general doesn't come
close to matching the value calculated from the self-resonance and the
inductance. To see what I mean, try entering D=10mm, N=10, len.=20mm,
d=1mm, and check what C(L,p) is reported. Now try changing D in 1mm
increments up and down. OK, so I don't trust the reported C(L,p)
value, but because I've checked several cases of all the other
reported values against measurements of actual coils and against one
or two other programs I use, I've learned to trust those other
reported values, within some tolerance (that's a lot looser than the
reported precision in the calculator!). -- I don't mean to pick on
that inductance calculator, just to use it to illustrate what applies
to pretty much all calculation programs and formulas.

Cheers,
Tom