I am trying to specify the noise requirements for a constant-current source that would drive a diode laser with a certain relative intensity noise PSD. Lasers typically behave as forward-biased diodes, with the emitted optical power linearly proportional to forward current above threshold:
I would expect a small change in current dI to create a small change in power dI * dL/dI, where dL/dI is the slope of this curve (with units W/A). But when I consider noise this way, the current spectral density dI will have units A/rtHz, giving dL units of W/rtHz. This is a weird unit--dL (the RIN) should be a PSD, with units 1/Hz, or dBc/Hz (without the square root).
A solution that fixes the units might be to calculate the applied power to the laser diode (I^2*R, with R the differential resistance dV/dI), but then this implies that the laser power is proportional to the applied power, not the applied current (so L vs I would be quadratic rather than a straight line). I must be missing something.
What's the proper way to calculate the intensity noise on a diode laser induced by the drive laser noise?
