The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
References: <3DD2EC20.2FB10158@maine.rr.com> <firstname.lastname@example.org> <3DD4487C.77549C50@maine.rr.com>
Subject: Re: Temperature Measurement Stability - +/-3.5 microdegrees
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
X-Inktomi-Trace: pc1-port1-4-cust150.cos2.cable.ntl.com 1037404501 32227 184.108.40.206 (15 Nov 2002 23:55:01 GMT)
Date: Fri, 15 Nov 2002 23:55:15 -0000
NNTP-Posting-Date: Fri, 15 Nov 2002 23:55:02 GMT
Organization: ntl News Service
I have to ask, what on earth requires that level of accuracy.
I am an engineer, not a scientist.
"Bill Sloman" wrote in message
> "Raymond E. Rogers" wrote in message
> > Thanks Bill, comments embedded
> > Bill Sloman wrote:
> > > For a more comprehensive discussion of temperature stabilisation near
> > > temperature, see A W Sloman, Paul Buggs, James Molloy and Douglas
> Stewart "A
> > > microcontroller-based driver to stabilize the temperature of an
> > > stage to within 1mK inthe range 0-38C, using a Peltier heat pump and a
> > > thermistor sensor" published in Measurement Science and Technology
> > > 7,. pages 1653-64 (1996), which is not only useful in its own right
> > > say so myself) but cites about a dozen earlier papers with useful
> > >
> > > The +/-0.2C Betatherm thermistors we used are claimed to drift less
> > > 11mK per year at 75C. Yellow Springs Instruments offers some +/-0.05C
> > > tolerance parts that ought to be even more stable, and are available
> > > shelf from Newark - they have 16 of the 10k at 25C part 46006 (stock
> > > 46F3197) for $47.60 each.
> > Yes we bought some from YSI that in fact have data to back up the
> > stability claims; and they are costing about $100. Okay for research
> > but probably to much for production.
> I checked out that NIST website, and it does look as if they are talking
> about the YSI 46006 - I don't know of anythihg else that is offered at
> +/-0.005C. The +/-0.05C parts from Newark are expensive. The Betatherm
> that we used are quite a bit cheaper, at about $5. Thermometrics does some
> +/-0.1C parts for similar sorts of prices, and in my current job we are
> actually using Fenwall thermistors - but a very small glass encapsulated
> part with good stability, but a +/-20% resistance tolerance, which we
> calibrate ourselves (and wish we didn't have to).
> > > A couple of the papers we cited achieved much better than millidegree
> > > stability. Larsen's classic paper describes a 50 microdegree stability
> > > bath built around a 500R platinum resistance sensor, and Priel got
> > > +/-3.5 microdegrees with a Fenwall thermistor sensor. Preil complained
> > > dissipating more than 16uW in his sensor produced appreciable long
> > > drift, but probably within the +/0.03C range asked for by the original
> > > poster.
> > >
> > > Interchangeable thermistors from reputable suppliers should be pretty
> > > stable. Their great advantage is their high sensitivity - about -4%
> > I will make a calibration fixture. We can spend money on capital
> > equipment but the shipped instrument must be cheap (relatively
> > speaking). I gave up on interchangability.
> Our calibration bath has been a pain in the fundament - it has been back
> the manufacturere a couple of times, and they don't turn it around fast.
> > > degree C, compared with +0.39% for Pt100 and Pt1000 sensors. This not
> > > makes the job of the measuring electronics much easier, but lets you
> > > away with cheaper and more easily accessible current-setting resistors
> > > so forth. A DIN B/3 Pt100 is a +/-0.04% resistor, and using anything
> > > than $10 Vishay 0.01% resistor throws away most of the accuracy you've
> > > for. The YSI +/-0.05C thermistors are +/-0.2% resistors ...
> > >
> > > The National Semiconductor LM35 semiconductor temperature sensor is a
> > > noiser than a good thermistor or a platinum resistance sensor, but
> > > easier to use. My office mate at Nijmegen University used the
> > > part for some millidegree stability temperature calibration baths and
> > I found the national data sheets .1 - .2 degC/khr , not good enough.
> That is what they are prepared to guarantee, not what they achieve. The
> is good because it was designed right, and it is made in a high quality
> analog fab. Lke all the National stuff it is produced in large volume at a
> low price, and that sort of production can't support the sort of testing
> that YSI does on their thermistors, and charges for.
> The risk always is that someone like National will find a cheaper way of
> meeting the published spec, and you stop getting the non-guaranteed
> performance that you have been relying on. You could have a second,
> design more or less tooled up against such eventualities, but nobody I've
> worked for has ever gone to that sort of trouble.
> The nearest analogy is the display tubes on the electron microscopes sold
> Cambridge Instruments in the 1980s - they were perfectly ordinary TV
> bought out of the huge batches made for the consumer market. The problem
> that the comsumer TV's were always being restyled for a slightly different
> look, and the tubes were never exactly the same shape from batch to batch.
> So our display housings were always big and sloppy, with plenty of room to
> accomodate an oddly shaped tube. The re-engineering costs were peanuts
> compared with what we saved on the display tubes.
> > > very happy with their long term stability. The Dallas parts may be
> suspect -
> > > we had a posting here a couple of years ago about some nasty
> > > drifts that turned out to be humidity driven. Water vapour was getting
> > > through the plastic packaging. I'm sure that the problem has been
> > > now, at least at the level of performance guaranteed on the data
> > I have yet to find the Dallas drift specs. I will try to look up your
> > paper and follow the references.
> > Maybe I can use your paper as an authoritative reference instead of the
> >$100 books.
> No way. I'm quite proud of that paper but it reports a very specific piece
> of work, and is much more oriented to short term stability than long term.
> Some of the papers it refers to are more interested in long term
> you are going to have to do a bit of reading to assemble the ammunition
> seem to want and need.
> If you have any trouble getting a copy of my paper, e-mail me directly at
> email@example.com. At the moment the alias points to Melbourne,
> and it is something like 48 hours before I'll be responding to e-mails at
> the Australian address, but after that I should be able to be helpful.
> Bill Sloman, Nijmegen
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup