[sdiy] Temperature Compensated Exponential Converter Using SSM2164

[David G. Dixon]

> > 1.  The "obvious" option:  Take 15V straight from the supply rail
> > and divide
> > it down according to the scheme as outlined before.  I'm presuming
> > that the
> > supply rail won't bounce all over the place!
> 
> WRONG - supply rails bounce all over the place.  Why do you think op-
> amp manufacturers specify PSRR?  UNLESS you have a circuit which is
> ratiometric so doesn't care about the exact value (and even then I
> still think its dumb - why let supply rail noise into your circuit?)
> never rely on the supply rails being 'constant' - they are anything
> but constant.

Point taken.

> > 2.  The "clever" option:  Drop the rail voltage with a zener diode
> > rated between 4.7 and 5.6 volts...
> 
> Ok, not so dumb as using the supply rails, and fine if you're not
> concerned about temperature stability too much. 0.01%/K = 100ppm/K.
> Not bad.

...and about all one could expect from the metal-film resistors in the
circuit, in any case.

> Better yet try something designed for this, like a bandgap or buried
> zener.  For example, the REF02C gives you +5V at 20ppm/C typical,
> max. 65ppm/C.  Bandgap ref. diodes give you similar stability, e.g.
> the ZRB500 will give you typ. 15ppm/C.

Neil, this is the first I've ever heard of these parts!

I see that Digikey carries REF02 5V references from Texas Instruments for
$3.11 for 15ppm/K, and $5.04 for 10ppm/K.  This seems a fairly small price
to pay for rock-solid tempco.

Thanks for the tip!