[volt-nuts] Voltage Reference Mini-Oven

[ws at Yahoo]

Here is another reason the Opamp is not listed as a "critical" part.

It is easy enough to remove the effect of its offset error without needing 
any additional test equipment.
With about a 100 to 1 resistor divider added across the amps inputs, one 
just needs to then adjust the opamp's zero offset so there is minimal change 
then switching one of the opamps inputs between the actual error signal and 
the divided down error signal.

Example:
Say the amp is being used as a neg gain of one with two 10K ohm resistors.
The amp's plus input is to ground and the neg input is connected to the 
summing junction of the two 10K resistors.
Add two more non-critical resistors in series, a 500K to the 10K summing 
junction and a 5 k to ground.
To zero the amps offset, adjust for min change at the output when switching 
the opamps neg input between the normal summing junction and the 100 to 1 
divider

Now this manual adjustment can be used to remove all zero offset of the 
opamp before (and while) any critical measurements are being made.

ws

************
Don't forget that you can have 1uV per C just from the opamp offset voltage. 
That makes three critical parts.

***************
>
> Using a circuit similar to
> http://www.febo.com/pipermail/volt-nuts/attachments/20110903/b7da2eca/attachment-0001.gif
>
> A 10 reference standard can be built so that there are only TWO really
> critical parts that limit its performance.
> The reference  Zener and the 5/3 gain matched resistors.
> The rest just takes careful analog design considerations.
> The way I hope to remove the resistor matching limitation is to do that 
> part
> digital.
>
> As far as more details on the Mini-Oven, nothing else to say. It is just
> three small parts "Taped" together.
> One is the device to be controlled, one is the heater and the last is the
> temperature sensor. Same as any oven, just smaller.
> I don't know how this mini-oven would apply to a standard cell.
>
> ws
>
> *******************