[time-nuts] temperature sensor

Alex Pummer alex at pcscons.com
Sun Jul 20 14:56:42 EDT 2014

1) Axtal makes a direct temperature sensing crystal, see Bern Neubig's 
former note, but since I grew up by learning how do you make one iron 
wheel from wood, I tried and used the transistor PN junction method many 
times, it works and it is not a which craft to calculate the tree 
resistors  for the bias  the only issue is that the resistors are also 
temperature dependent, but you could go around it by using two resistors 
one carbon film and one metal film, by selecting the proper values you 
could get -- really!  -- zero temperature coefficient [ carbon film has 
negative tempered coefficient , metal film has positive...] although 
every tranansistor is different, the slope of the "curve " , which is 
very linear is contant cca 2.2 mV/C -- that is on  the visit card of the 
silicon --  for much larger range than the crystals, the only larger 
disadvantage, that from the crystal you get a high level signal with 
changing frequency [ and it is tempting to belive that it is easier to 
handle since it is "digital"  and from the transistor you get a DC 
changing DC which is analog. Of course you have to calibrate one point 
of the DC level at known temperature.
2) yes crystal cutting is a kind of black magic, until you do not know 
what you are doing, but if you cut a few thousand pieces it becomes very 
logical, and fun. Ask Bernd... he did it for very long time...
C) Ask Bernd he knows it better, he is still making his living with it..

On 7/20/2014 7:55 AM, Tom Van Baak wrote:
> Hi Alex,
> Thanks for this level of detail. Fascinating. Is the fundamental physics behind the quartz angle-of-cut well understood, or does this fall into advanced alchemy and industrial magic?
> I understand about the time constant now. Yes, on the order of a few seconds makes sense. Would it be possible to have other mounting techniques that improve environmental contact with the crystal?
> Do you know of any commercial quartz crystals (say, in the $1 to $10 range) that have been optimized for large tempco at room temperature? Or optimized for linearity over a large range (e.g., -40 to +40 C)? I was able to test one once, a 5x7mm XO, but I don't know any more about it other than it came from Switzerland.
> Thanks,
> /tvb
>    ----- Original Message -----
>    From: Alex Pummer
>    Sent: Saturday, July 19, 2014 8:22 PM
>    Subject: Re: temperature sensor
>    close to the inflection point -- which is dependent of the cut, and for ordinary crystal not made for ovenized operation or for temperature sensing,  between +20C° to +28C° -- the frequency versus temperature function for the first 8C° to 15C° bellow and above the inflection point -- the linearity could be as good as 0.03% . The steepness of the slope for certain cutting angle [35° 12"] is almost zero [less than 0.1ppm/C°. By increasing the cutting angle the slope  becomes negative, by reaching 30° 30" it is approx  3.35ppm/C°.
>    Going the other way at, 30°05" the slope is +1.0ppm/C°.
>    The thermal time constant of an ordinary quartz is in the range of seconds -- up to 10 sec -- since the quartz is in  vacuum -- to keep the mechanical friction to the air out, and the Q high -- the only thermal conduction between the outside world and the crystal are the very thin -- 0.08mm or less -- wires which providing electrical contact
>    That is how I remember as Jean Hoerny and me -- yes that Hoerny one of the traitors -- made the first French quartz clock at LIP in Besançon, back in the past century,
>    it is enough number there, or should I look for my old note book?, there was a note; we did not needed to grind the quartz to a precise frequency, we measured it and set the divider, that made the production very economical, how much? that remains the secret of LIP.
>    73
>    Alex
>    ----- Original Message -----
>    From: "Alex Pummer" <alex at pcscons.com>
>    Sent: Saturday, July 19, 2014 6:16 PM
>    Subject: Re: temperature sensor
>    temperature sensing with crystal is very accurate, but unless the
>    crystal was made for that application --  has a very large time constant
>    73
>    Alex
>    On 7/19/2014 4:45 AM, Attila Kinali wrote:
> On Wed, 25 Jun 2014 14:21:49 +0200
> "Bernd Neubig" <BNeubig at t-online.de> wrote:
> the time-nut approach for temperature measurement would be to use
> a temperature sensor crystal - like the good old Hewlett-Packard guys
> did many years ago. If you do not look for ultra-linearity of the frequency
> vs. temp response, there are several possible types of crystal cuts
> possible. The simplest one is the Y-cut or the slightly rotated Y+5° cut,
> which has a slope of about 90 to 95 ppm/K @ room temperature.
> Smaller sensor crystals are tuning-fork type crystals, which
> come in the same small cylindrical package as normal watch crystals.
> For further reading I have attached an application note for such a
> crystal from AXTAL.
> Do you have any data on the temperature resistance from case to crystal?
> The PT100 and NTC sensors have the nice property of having a very good
> thermal coupling between the sensor element and the case. But i suspect
> that temperature sensor crystals have a very small area that couples
> the crystal to the case (in order to get a high enough Q for the oscillator
> to work), which in turn limits the speed at which the sensor reacts to
> temperature changes.
> 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.

More information about the time-nuts mailing list