Why has no one mentioned thermocouples? I had some experience with thermistors a few years back designing thermal attenuators and equalizers for CATV. NTC thermistors can have a large change of resistance for a unit change in temperature. They aren't linear, but there are formulas for computing resistance vs temp. PTC thermistors have a much smaller change per unit change in temp., but are much more linear. And, they are susceptible to self heating, which makes things interesting. If I remember correctly, in my research something called an RTD was supposed to be the king when it came to accuracy and repeatability. As someone else has stated, the IC devices are supposed to be quite good, but you have to interface with them.

On 5 April 2018 at 15:44, John Green <wpxs472@gmail.com> wrote: > Why has no one mentioned thermocouples? > I had some experience with thermistors a few years back designing thermal > attenuators and equalizers for CATV. NTC thermistors can have a large > change of resistance for a unit change in temperature. They aren't linear, > but there are formulas for computing resistance vs temp. PTC thermistors > have a much smaller change per unit change in temp., but are much more > linear. And, they are susceptible to self heating, which makes things > interesting. If I remember correctly, in my research something called an > RTD was supposed to be the king when it came to accuracy and repeatability. > As someone else has stated, the IC devices are supposed to be quite good, > but you have to interface with them. > Keysights nano volt / micro ohm meter takes https://www.keysight.com/en/pd-1000001296%3Aepsg%3Apro-pn-34420A/micro-ohm-meter?cc=ZA&lc=eng can do direct SPRT, RTD, Thermistor, and Thermocouple measurements SPRT = Standard Platinum Resistance Thermometer I don't claim to know much about this, but the uncertainty quoted for the SPRT probes with that meter is 0.003 deg C. When I Google SPRT probes, I see they are incredibly expensive - many thousands of USD each. I'm a bit puzzled there seem to be a number of 3-wire platinum resistance thermometers. I can understand using 4-wires for a Kelvin connection, but can't understand the use of 3-wires. Of course, for environmental monitor in a lab, one is most unlikely to need very high accuracy. Dave

Hi Thermocouples are *really* low output voltage devices at “normal” temperatures. That gets you in to fairly fancy measurement setups ( no “just strap it to an ADC input” stuff). They also require cold junction compensation. They measure offset temperature to that junction. You do a lot of work to get just one sensor to operate. That can be done with various chips from various outfits. It’s still not cheap. The net result is no more accurate or higher resolution than the other stuff we have discussed. The place thermocouples come in are for very high or very low temperatures. If you want to know if it’s really 2,876C, a thermistor is not what you want to use. RTD’s are indeed very stable long term. Properly handled they can do a real good job. They are basically a strain gauge. Mount them wrong (yes, I have a few hundred examples of this) or treat them wrong ….. they don’t do quite as well. If you really want to get accuracy out of them, you run a triple point cell to calibrate them before use. Yes, this is getting off into temperature nuts territory. Heading back a bit to the original question: What *does* impact your frequency or time standard? 1) Temperature 2) Pressure 3) Humidity 4) Load 5) Supply voltage 6) Acceleration (gravity) 7) Stress / strain (bending) 8) Time ( aging ) 9) EMI ( injection locking ) I suppose that list could go on for quite a ways. The obvious one on the list that has not been addressed yet is voltage. Bob > On Apr 5, 2018, at 10:44 AM, John Green <wpxs472@gmail.com> wrote: > > Why has no one mentioned thermocouples? > I had some experience with thermistors a few years back designing thermal > attenuators and equalizers for CATV. NTC thermistors can have a large > change of resistance for a unit change in temperature. They aren't linear, > but there are formulas for computing resistance vs temp. PTC thermistors > have a much smaller change per unit change in temp., but are much more > linear. And, they are susceptible to self heating, which makes things > interesting. If I remember correctly, in my research something called an > RTD was supposed to be the king when it came to accuracy and repeatability. > As someone else has stated, the IC devices are supposed to be quite good, > but you have to interface with them. > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there.

I've used a few Pt100 RTDs over the years for gas flow temperature control in the 200-300C range.  At 0.385 ohms per K, the challenge of managing the noise is a lot tougher than a cheapo NTC thermistor, but I needed repeatability and stability. http://www.farnell.com/datasheets/2278399.pdf I use a 4-wire Kelvin connection, with 1 mA current drive. £14 at 10-off Neil On 05/04/2018 16:27, Dr. David Kirkby wrote: > > Keysights nano volt / micro ohm meter takes > > https://www.keysight.com/en/pd-1000001296%3Aepsg%3Apro-pn-34420A/micro-ohm-meter?cc=ZA&lc=eng > > can do direct SPRT, RTD, Thermistor, and Thermocouple measurements > > SPRT = Standard Platinum Resistance Thermometer > > I don't claim to know much about this, but the uncertainty quoted for the > SPRT probes with that meter is 0.003 deg C. > > When I Google SPRT probes, I see they are incredibly expensive - many > thousands of USD each. > > I'm a bit puzzled there seem to be a number of 3-wire platinum resistance > thermometers. I can understand using 4-wires for a Kelvin connection, but > can't understand the use of 3-wires. > > Of course, for environmental monitor in a lab, one is most unlikely to need > very high accuracy. > > Dave > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there.