Would something like the LM399 work as the voltage reference? How would I get the voltage down, and also, are there any other ADI or TI parts that I can use?
What other ADI or TI parts: references or ADCs?
In my opinion the most precise way (without using precision resistors within a temperature controlled cabinet) to divide the LM399 to usable ~5V is a 3:2 divider giving around 4.6V as reference voltage.
I have already simulated this with LTSPICE:
Input voltage is from a LM399 simulated as 6.86V with some "AC noise"
C4 is the 10 nF (but I need it doubled because the simulation uses 2 different LTC1043 instead of one)
C3 is charged to 1/3 of the input voltage and charged by C1+C2.
So the output voltage is Input voltage minus C3 voltage giving the 4.6V.
C5 is selected exactly for 1.5 times of C1...C3 to optimize settling time (and thus minimizing output ripple).
Of course the output has to be buffered by a OP-Amp to maintain stability.
It is a good idea to select C1...C3 for equal values. And C5 for a exact 1.5 fold capacity.
This will minimize startup time and ripple.
As power supply I have used 14V because I want to measure up to 10V with a second LTC1043 and because the heater current and PSRR of the LM399 will be lesser with higher power supply. Of course with specioalised cirquits you can use the LM399 from around >9V on.
With best regards
Andreas