In a previous implementation of PID control of the vacuum tank temperature, the four AD590 sensors attached to the CNT vacuum tank were configured with a transimpedance amplifiers to produce four voltage signals. These were then fed into 4 separate ADC channels to be averaged in digitally.  

The AD590s take any voltage in the range 4-30 V and produce a current proportional to 1uA/K (where K  is reference to absolute zero).  The current is then converted to a voltage signal with buffering opamps, this is passed to a second (unknown amplification stage) and the voltage output signal is then routed through two Dsub9 connectors to the rack for ADC. 

Typing the PBC board serial number, D980401, into this elog or DCC turned up no results.  However it looks like the board is listed under D980400. It has been modified to make use of only its initial opamp x10 gain stage plus an additional gain stage.  The additional opamps used to provide a tap off of the converted AD590 signals are glued upside down to the PBC board, making it hard to identify what is going on.  Here is what I do know, the board is suppling 11.7 V regulated voltage to each of the two pin AD590s.  Putting a fixed resistance (40.2kΩ) across this voltage source gives a current I = 11.7V/40.2kΩ=2.91uA. At the output of the board we get 2.93±0.01V signal which is 10^4 V/A gain.  

I wasn't sure on the best way to measure the noise contributed by the board. I took the dark noise in two cases, one with no load across the input of the opamp and one with the 40.2 kΩ load (to simulate the expected current for AD590 at room temperature).  I've attached the two PSDs of the output noise (stitched from multiple spans using the SR780). Data and matlab plotting script also zipped and attached.

Edit: typos Mon Oct 3 14:40:52 2016 awade