Voltage Reference

[branadic]

Thanks for the fish.
You are write, one factor that could cause problems, but as long as the temperature over the complete circuit board is nearly constant (and it is) this is not a problem and the currents flowing in the circuit are small. In addition this board is just for my lab experiments and not a (mobile) reference or standard cell replacement.
It could be cased to prevent air flow and therewith temperature gradiants that could causes thermovoltage effects, no question. But on the other hand you need to have the equipment to measure such small effects.
I'm not Jim Williams or some of that voltage nuts 

[branadic]

My reference is now running a few while. The reference has seen a three weeks burn-in @ 120°C, afterwards the heater was glued on top of the SMD resistors and thermally dryed.
After all that the reference showed values between 10.00179-10.00180V (34401A, 21.4°V, RH=32%) and now more than a month ago it's at 10.00182-10.00183V.
Meanwhile I found the 7.5 Digit Prema 5017 also uses the National LM399H

http://www.amplifier.cd/Test_Equipment/other/Prema-5017.html

and could verify that on a device at work. This is pretty interesting so I need to wait a further few months to watch out the drift and see if its worth for such a use.