Hi Indeed the most stable (long term) standard is a very high end RTD. These devices have a lot of voodoo in their design. Even with all that, they still arrive with a note on the box that reads “ for applications requiring < 10 mk, re-calibrate before use”. More or less, you *also* need a triple point cell to use one. Triple point cells do not bounce when they hit the floor …. I have empirical data on this :(
In terms of *resolution*, even the high end gear goes over to thermistors. You can get about an order of magnitude increase in relative performance running a fancy thermistor rather than a fancy RTD. The Fluke / Hart web site is a wonderful thing to drool at in this regard. Be sure to lock the credit card safely away before going anywhere near it. https://us.flukecal.com/about/fluke-calibration-brands/hart-scientific <https://us.flukecal.com/about/fluke-calibration-brands/hart-scientific> https://us.flukecal.com/products/temperature-calibration/probes-sensors <https://us.flukecal.com/products/temperature-calibration/probes-sensors> Keep in mind that those probes are done with a specified accuracy. The real world numbers likely are better than spec by some factor ….. ======= Getting back to TimeNuts stuff. If you want to make a practical controller for a fixed set point, the thermistor still wins. You get more voltage / lower noise (in miliK) with the thermistor. The fact that you can get a good one (= very low aging) for $2 rather than $200 also helps ….. :) Bob > On Sep 28, 2020, at 2:48 PM, John Moran, Scawby Design > <[email protected]> wrote: > > Thanks for going easy on me Bob ... a case of more haste, less speed! I > focussed on low long-term drift specs without realising I had turned up a > voltage reference, sorry. > > However, I have found some YSI glass encased thermistors that have long-term > drift specs of <10mK at 25C and 75C over a period of 100 months. They are in > the YSI 46000 series - data sheet attached. > > http://www.farnell.com/datasheets/169207.pdf > > There is an interesting paper by NIST on achieving the International > Temperature scale - link attached (it is 196 pages and 10MB) that seems to > indicate platinum sensors are the most stable at less than 1mK and, of > course, to be able to measure these resistors accurately, you need an equally > low-drift voltage/current source. :-) > > https://nvlpubs.nist.gov/nistpubs/Legacy/TN/nbstechnicalnote1265.pdf > > This reference appeared from an EEVblog where I think some Volt-nuts were > discussing temperature. One of them confirmed that the most economical way > was to have a group of lower-cost sensors and characterise them. > > https://www.eevblog.com/forum/metrology/long-term-stability-of-temperature-sensors/ > > John > > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
