[time-nuts] Making a HP 10811 better

[ed breya]

Yes, running everything chilled would be better if low temperature 
crystal turnovers were available. It is more complicated to chill 
than heat, of course, even today, but way back when they started 
temperature stabilizing things it was nearly impossible on a small 
scale - or at least very impractical, until the advent of TECs. It's 
also trickier to do both because there's that crossover region where 
the mode has to shift and you can get dead zones in the control 
function. Doing heating or cooling only is like a Class-A amplifier - 
much simpler to keep linear.

I think the main problem with chilling though, is that you get 
condensation, depending on the humidity (or dewpoint), so you can't 
go very low in temperature to take advantage of lower noise and 
bias/leakage currents in devices, without encountering the 
condensation problem. For example, one of my projects involves making 
a very stable voltage standard that has the guts inside a TEC-chilled 
(or warmed) aluminum block. I plan to run it somewhere around 15-20 
deg C or so, or as low as I can without having condensation under 
most conditions, since the guts aren't going to be hermetically 
sealed inside (but maybe have a rechargeable dessicant). Another 
project involves building a log attoammeter with the input amplifier 
TEC-chilled as far down as possible (-40 to -50 deg C or so) to 
minimize bias current. This one has to be specially guarded, and with 
the cold block inside a hermetically vacuum sealed enclosure, as any 
condensation whatsoever - inside or out - could defeat the whole purpose.

Ed