** Reply to message from kenneth d lewis <kdlewis@JUNO.COM> on Fri, 25 Sep 1998 16:17:35 EDT

> Hey Math Majors; > How about coming up with a formula that will give me the same information > as the graph on page 24.45 of the Bentley. Preferably with temperature as > the variable and the product in kilohms.

I'm no mathematician... but my calculator does curve fitting. According to it, unless I messed up the data entry there is no algebraic formula that fits that curve. Exponential is the closest fit, but still several hundred ohms off at the high end. I'm pretty sure that calculus (er... the Calculus of Infinitesimals, sorry guys) can deal with it, but I've forgotten what I ever knew about it. Also a spline curve should do, but I don't know how to generate one (spline curve passes through all the given points, looks pretty odd if some of the data are offset, but should be fine here).

Suggestions (pending answer from the calculus boys):

1) restrict the range of temps to what you're really interested in -- maybe the curve would be close enough then (so let me know your range of interest).

2) set up a spreadsheet with a lookup table for the values

3) go to a copy machine and blow that sucker up until you can actually see it.

I suspect #3 may be most practical. Also, replotting on log paper should flatten it out a good deal, so you could do away with the scale shift in the middle.

david

David Beierl <dbeierl@ibm.net> 401 274-5827 voice, -6349 fax OS/2 V4, FP7, JVM 1.1.6, JSM 98.6.3