Sounds like a really high threshold, which is only reached by the rush from cable connecting,  if it is really high, then it will heat quickly and extinguish itself.   Turning it up CW will most likely fail the device.  it needs a high V pulse to get it there, which provides a rush of current just long enough to see it lase, but in CW this will not happen, as the device will heast up before it gets to the threshold.  try a train of very short high current pulses, vary the duty cycle.  

Nice. Thank you.

It may be that LD is damaged by static discharge. In that case surface (end mirror or clean semiconductor surface) is damaged by high intensity light. Surfaces has few times lower damage level as bulk material. As result there is scattering spot where should be mirror-like surface and cavity quality goes down. So, you cannot get normal LD operation any more. But by over-pumping (due to some static or other way) you still can get generation with bad cavity quality.

This is why special precautions should be taken working with LD - very easy to damage them by intense light due to strong static discharge current. And surge currents of power supplies at On Off kills LDs.

Thanks.  All the usual precautions were taken. Other diodes work fine with the same driver and pin out.  This particular diode is at an uncommon wavelength. The diodes are going back to the manufacturer for them to investigate what happened (I decided not to 'max-out' the current).   They only started production of these diodes at the end of 2013.

If you saw a flash as the current cable was simply connected to the diode, it would suggest that there was electrostatic discharge, which can produce a current pulse well above the max. ratings. As the laser output is red, this is probably an AlGaAs laser diode. With this material, the light output can track the current input to very high levels, as there is no saturation mechanism like that in InGaAsP laser diodes to help protect them. The laser facets are probably blown off. You may like to measure a current - light power curve, which will likely indicate from its linearity that you now have a LED, most probably with a low efficiency. The fact that you can see the light coming out would support the guess that the crater in the laser facet is scattering light.

Thank you. We did measure the PI curve and sent it to the manufacturer.

Interestingly the unplugging and plugging of the driver cable always creates the same level of flash intensity.  There doesn't seem to be any further deterioration (assuming the facets are damaged).  Also if the facets were damaged I wouldn't expect the reasonable looking mode shape that we see (admittedly difficult to assess properly when it only lasts ..microseconds(?)  Hmm, that's something we could have measured - the transient time. Too late now.  Thanks for everyone's feedback (which is what we were hoping to do with the laser!)

Hello,

looks like you have the supply with current stabilization. In that case, when the cable is unplugged, voltage is increased to max value (preset max voltage limit). However, output small capacitors (which typically are present) and cables (also small capacitance) are charged. The stored charge is discharged via LD producing short peak of current when you connect the cable of the LD to the supply . The supply instantly switched from voltage limiting to current stabilization which takes place at lower voltage level and everything starts to work "normally". But damage is already done - LD, even if it was OK before, may be damaged due to short current peak.