My thoughts on relays:

     Suppressing the flyback:

     Just do it, put a diode AT THE COIL.  Keep the leads/traces that run
     to the coil away from the rest of your system.  Run them as a pair.
     Twist them if you can.  (Cut the loop area !!!)  I've tried lots of
     other ways - this one works.



     Faraday shield:

     Good idea if you handle the extra capacitance from the contacts to
     your shield point.



     Reeds vs. Electromechanical:

     I've used Coto/Wabash (sp?) electromechanical relays in circuits that
     work with 100 nV, 50 nA, and 6 Gohm source resistances - it can be
     done.  I've used Omron G6E relays in circuits that sense 10nA signals.



     Latching vs. single side stable:

     Latching is better for reducing noise.



     Places to look for help:

     Keithley publishes a book called "Low Level Measurements" that most
     Keithley distributors will give you for free - it's a great place to
     start.  (Keithley makes electrometers and relay matrix boxes for
     making fA type measurements).  It's a good little book.

     Point your surfboard to www.emiguru.com/bibliog.htm and look up Henry
     Ott's book, read it over and over.


     Hope it helps.

     Jon Bertrand
     jonb@cirris.com