> I'm using a precision reference (Analog Devices REF191) an a classic three
> pin regulator constant current source configuration
> i.e. Vout is tied to the floating ground pin via a current setting
> resistor, and the load is connected to the floating ground pin.
> In this case, the regulator forces 2.048 volts across a 2.1K ohm resistor
> (Rset), giving me a nominal 1.01 mA constant current output.  This circuit
> can be found on most three pin regulator data sheets and cook-books.
>
> The power supply is around 7-10 volts.  The headroom of the regulator is
> 1.5 volts, and the Vout is 2.5 volts.  In theory, I should have at least (7
> - 1.5 - 2.048) 3.4 volts available to drive the constant current through
> the load--which at 1 mA should in theory be as much as 3.4K Ohms.
>
> What I am seeing is this:  Regulation and constant current is VERY stable
> from a load resistance of zero to about 1.6K, at which point the current
> starts to drop radically.  If I monitor the voltage across Rset, it remains
> stable until Rset = Rload, at which point regulation ceases.
>
Sounds for all the world like you are running out of voltage
to maintain compliance, but of course this doesn't add up.
I would try and see if the regulation will hold up for a
higher input voltage (it can take up to 18volts). This might
help to show if the problem is regulation based or some
sort of self oscillation or heating etc.
BTW the A/D data sheet recommends a 1uF cap across
the set resistor, maybe that is worth a try?
Just a stupid point..the sleep pin should be tied to Vin.
_____________________________

Lance Allen
Technical Officer
Uni of Auckland
Psych Dept
New Zealand

www.psych.auckland.ac.nz

_____________________________