VCO hacks, tweaks again

[terry michaels]

Message text written by terry michaels
>
Message text written by Joachim Verghese
>
> I'm glad you caught the 3300ppm/K vs. 3500ppm/K problem, I missed that
one
> completely.

I fear this issue is even more complicated if you opt for really
accurate compensation.

The problem is that the temp coefficient of the scale factor is a
non-linear function of temperature. 3300ppm/K is accurate only around
30 deg(C) (303K), since

(303 + 0.5) / (303 - 0.5)  =  1.003306

At 40 deg(C) (313K), which perhaps is closer to the operating temperature
of an enlosed circuit, the required compensation coefficient is

(313 + 0.5) / (313 - 0.5)  =  1.003200, i.e. 3200ppm/K

In other words, using a (linear) tempco resistor isn't the most accurate
compensation method, although it works well enough for most purposes.

-joachim


<

Hi Joachim:

I have to dispute your math.  The scale factor for the exponential function
of a silicon transistor is given as delta Vbe = kT/q log delta Ic.  (ref:
National Semiconductor Application Note 30).  K = absolute temperature,
which means the scale factor varies directly with absolute temperature, and
is a straight line relationship which reaches zero at zero temperature. 
The Tel Labs, Inc. datasheet on the Q81 resistor has a graph which shows a
straight line resistance vs. temperature change between -55C and +275C 
which, when extrapolated, passes through zero.  Therefore, the Q81 has a
linear tempco, proportional to absolute temperature.  The Q81 will give a
first order compensation for the scale factor of a silicon transistor, and
can be trimmed to the exact value needed by adding a small value zero TC
resistor in series with the Q81, as suggested by Ian Fritz.

BTW, the value of 3500 ppm given by Tel Labs is the tempco between -55C and
+275C, averaged to room temperature.  The value is misleading if you don't
know that the manufacturer calculated this way.

Terry
<

Joachim:

- - OOPS. I screwed up - - 

Upon further thought, the addition of a zero tempco resistor in series with
the Q81 will only work perfectly at one temperature, as the composite
resistor so created will no longer have a tempco that passes through zero,
and won't track the scale factor at all temperatures. However, the error is
small, and it should work pretty well within the range of normal ambient
temperatures.

Terry