[sdiy] resistor temperature coefficient measurement

[patchell]

    I am somewhat surprised at this experment myself...although from a
different perspective...

    I have done a lot of work with low TC type stuff in the past, and the one
thing I noted about freeze spray (if that is what was used) is that it will
tend to cause what I call thermal shock.  I would generally use it only to
find potentially sensitive parts.

    But the big problem I used to have was not so much with the tempco of the
parts, but rather the Thermal EMF's that were generated by the parts...back
when I used to do this work, wirewound resistors are what you used if you
wanted Low TC and precision (I used to use a lot of Jordan J-120 0.01%
resistors).  These had a built in thermal couple where the resistance wire was

bonded to the wire lead.  As long as the part had no temperature gradients so
that both ends were generating the same voltage, everything was OK, but if you

took a soldering iron and heated up just one end of the resistor, things would

really go hay-wire...so to speak (the circuits I was working with were doing
things at DC, but down below a microvolt).  Vishay Bulk Metal resistors are
better choices these days, they were just coming out, or I should say, making
themselves known, at about the time I moved on to other things...

    Also, you have to remember that when a tolerence is specified, you need to

know where on the bell curve you are...is that a 1 sigma or 3 sigma number or
what?  I have never been able to figure that question out.

Ian Fritz wrote:

> Hi Martin --
>
> Interesting result. Thanks for the report.
>
> I usually assume that if a component has a symmetrical +/- tolerance that
> the material composition has components with opposing properties that are
> made to balance.  I imagine you are right that the +/- 50 ppm is meant to
> cover anomalies in production and that most of the time the results are
> much closer.
>
> You might try to estimate your temperature excursion by cooling a carbon
> resistor (or thermistor) of the same physical size.  But even if your
> cooling was closer to 10 degrees the stability is still better than 1 ppm.
>
>    Ian
>
> At 05:42 AM 2/14/2003, Czech Martin wrote:
> >Yesterday I thought I could look into the (unwanted)
> >temperature dependency of some resistors.
> >I took some metal film types out of the box (speced to
> >+- 50 ppm) and measured the resistance at room temperature
> >and with coolant spray. To my surprise the resistance
> >did not change in the first 5 digits, but the temperature
> >dropped certainly by 30 degrees C.
> >
> >Perhaps the +-50ppm spec is the maximum deviation
> >that some production lots may have on a bad day
> >and the average is much better than this?
> >
> >Any experiences or comments?
> >
> >I will do this again with more time.
> >
> >m.c.

--
 -Jim
------------------------------------------------
* Visit:http://www.silcom.com/~patchell/
*-----------------------------------------------
*I'm sure glad Merry Christmas comes just once a year
* -Yogi Yorgensen
------------------------------------------------