From: email@example.com (Bill Sloman)
Subject: Re: Temperature Measurement Stability - +/-3.5 microdegrees
Date: 17 Nov 2002 21:48:59 -0800
References: <3DD2EC20.2FB10158@maine.rr.com> <firstname.lastname@example.org> <3DD4487C.77549C50@maine.rr.com> <3DD65C67.8C2CE98C@maine.rr.com>
NNTP-Posting-Date: 18 Nov 2002 05:48:59 GMT
Jim Thompson wrote in message news:...
> On Sun, 17 Nov 2002 03:40:28 GMT,
> "Walter Driedger" ,
> In Newsgroup: sci.electronics.design,
> Article: ,
> Entitled: "Re: Temperature Measurement Stability - +/-3.5
> Wrote the following:
> |37 C is an average, typical human body temperature. Firstly it is an
> |average; actual temperatures may vary. Secondly it is typical. Daytime,
> |nightime, time of the month, oral, rectal, arm pit; change is the only
> |constant. Finally, it is the human average. As such it does not apply to
> |dogs, gerbils, parakeets or iguanas.
> |It is not some universal constant like the speed of light or even the triple
> |point of water. The is no more point in measuring human body temperature to
> |an accuracy greater than .1 degree than there is in declaring the
> |temperature of the weather in a given city, at a given time to be 37.01. It
> |varies more than 2 degrees from one side of the street to the other.
> I think what the OP *really* needs is stability/repeatability; *not*
> absolute accuracy.
What the OP wants is *demonstrable* stablity and repeatability. If you
deliver +/-0.05C absolute accuracy, you short-circuit a whole lot of
the tedious and expensive testing reqired to prove that you have
achieved short term stability when you can't offer long term
+/-0.005C is relatively expensive, at $47.60 in small volume. +/-0.01C
is cheaper - I think we were quoted about $6.75 100-up.
If you genuinely only want short term stability, using a Peltier
junction (clamped to a well-insulated thermal mass) as your
temperature sensor gives pretty spectacular results. I've seen this
described for calorimetry applications.
Bill Sloman, Nijmegen