Adding the 3.3V regulator is an excellent idea. Could even use a 7805 5V regulator too. The main reason I see this as so great of an idea is that Raghunathan has said that his LED test gadget has already been built and he was looking for something simple that could be added to it to correct for the leakage problem.

As a side comment on the leakage subject. If we calculate a typical case where the VF of the LED is say 2.2 volts and the anode side sits at 4 volts then this means that there is 4-2.2 = 1.8 volts across the series current limit resistor. Knowing that Raghu has shown us that the resistor is 2.2K this means that the leakage current is 1.8/2.2K = 0.82 mA per output. This is a total crap design on the part of NI for producing a board with this much leakage if indeed the board is truly rated for a bias supply of up to 30 volts. I say this because a general purpose I/O board like this should be designed for driving any type of output. I do understand that it falls on the user's shoulders to apply the boards in a manner that meets their needs but I still think NI has something to answer to here.

Michael Karas