You mention one of the situations that I have a bit of concern about. The curve fit is achieved by using the internal curve fitting routine of Excel in its X-Y chart menu. I am fortunate that it is a quadratic equation that is required and not a higher order. If excess heat were an issue I feel confident that it would impact the calibration accuracy with its non linear behavrior versus temperature becoming evident.
The closeness of my calculated input power to the actual is an indication that excess power is not having a large impact. Also, there are enough pairs of points covering enough of the axis to rule out luck in obtaining the proper coefficients. As you know a perfect fit is always possible if only 3 pairs of points is available. We used 8 pairs if I recall correctly. Dave -----Original Message----- From: Daniel Rocha <danieldi...@gmail.com> To: John Milstone <vortex-l@eskimo.com> Sent: Wed, Feb 6, 2013 1:55 pm Subject: Re: [Vo]: MFMP Null Result That was not my question. I want to know if he is also fitting excess heat with his curve and thus giving false negatives. 2013/2/6 Jed Rothwell <jedrothw...@gmail.com> Daniel Rocha <danieldi...@gmail.com> wrote: How can you tell whether these are falso positives and not false negatives? 0.2 to 0.6 W with this system is zero. Not positive or negative. That is within the noise. As I said before, no instrument can produce exactly zero. - Jed -- Daniel Rocha - RJ danieldi...@gmail.com
