[time-nuts] temperature sensor

Chris Albertson albertson.chris at gmail.com
Mon Jul 21 13:38:34 EDT 2014

I got a few of these
to replace my TMP36.   The idea was to get rid of the noise by using a
sensor with a built-in digital interface.     They are spec'd for only 0.5C
accuracy but I think they are more repeatable and do much better than 0.5c
for relative measurements.   You can use the link above to buy them but you
can also get free samples.    Another advantage of the digital ones is you
can install several into the heat sink and all of them only take one pin so
you can add any number of these and still use the tiny 8-pin uP.

Calibrate any sensor with boiling water.  Either just assume it is 100C or
look up the local air pressure and compute the boiling point.
Or, I just happened to have a good laboratory glass thermometer, the kind
that is about two feet long.  Place it and the sensor in the same liquid
bath and cal is "close enough" for anyone who is not a "temperature nut".

I have not yet had time to use these in a real application.  but it looks
like they work.  Again you have to couple them well to whatever you are
measuring.  Do not conduct the "wrong" heat into any of these devices via
the lead wires.

On Mon, Jul 21, 2014 at 9:02 AM, Javier Herrero <jherrero at hvsistemas.es>

> Hello,
> There are two types of NTC thermistors that are (very) frequently used in
> spacecrafts, the Measurement Specialties (traditionally YSI, Yellow Spring
> Instruments) 44907 and 44908, that are 10k @ 25ºC NTC:
> http://meas-spec.com/downloads/44908.pdf and http://meas-spec.com/
> downloads/44907.pdf
> Both are specified in error against temperature, being ±0.1ºC for the
> 44908 and ±0.2ºC for the 44907, in the 0-70ºC range.
> I suppose that these ones are obscenely expensive, but the "civilian"
> versions are the 44006RC and 44031RC, with the same specified tolerances,
> and the same calibration curve, and readily available (e.g. from Farnell).
> I suspect that the difference between the space qualified ones and the
> others are quality control and traceability, and perhaps (only perhaps) a
> different low-outgassing epoxy coating, but probably the sensor material is
> the same in all of them. The advantage is that you get a quite good intial
> tolerance, so calibration can be simplified, since you only need to
> calibrate the resistance measurement device with known resistances to
> obtain a ±0.1ºC error from the sensors.
> The qualification mandated by NASA for these sensors refers mainly to
> MIL-PRF-23648 standard, that describes a load life test, consisting in
> application of maximum power specified for the themsitor during 1000h,
> intermittently (30min. on, 30min. off) and checking at 250, 500, 750 and
> 1000h that the thermistor remains in the specified tolerance in the
> applicable specification sheet, so I suppose that these thermistors are
> quite good wrt long-term stability.
> In applications where more stability is required, also Pt thermistors are
> used in spacecrafts (mainly Pt1000 and Pt2000, also others), but the
> problem is that depending on the wire run, they shall be measured using
> 4-wire techniques, and also initial calibration is not so tight (when it is
> so tight, they tend to be also very expensive) and not so easy (a Pt1000
> has a rough variation of 3ohm per ºC).
> Regards,
> Javier
> On 21/07/2014 16:12, Attila Kinali wrote:
>> On Mon, 21 Jul 2014 04:39:51 -0700
>> Alexander Pummer <alexpcs at ieee.org> wrote:
>>  NTC are not that very stable, they are amorphous material winch could
>>> recrystallize slowly and therefore change it's electrical behavior ,
>>> PT100 style is more reliable since it is pure metal
>> How long is the time constant for NTCs?
>> I guess, it wouldn't matter for most of the measurements we do,
>> as NTCs need to be "calibrated" before precision measurements
>> anyways. Unless one measures over several months, or years.
>> But on this timescales, i wouldn't really trust an off the shelf
>> PT100 either. Not unless i measure its stability
>> For use in GPSDOs and OCXOs, i guess it doesn't really matter,
>> as long as the NTC stays within spec. There an external loop
>> corrects for the variation/drift of the measurement.
>> While we are at it: what is a good way to calibrate/characterize
>> temperature sensors that is available to hobbyists?
>>                         Attila Kinali
>>  _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/
> mailman/listinfo/time-nuts
> and follow the instructions there.


Chris Albertson
Redondo Beach, California

More information about the time-nuts mailing list