Mysterious Dimming Laser Diode.

[The_Doctor]

Samsung SLD6351826. Index guided MQW diode, 5 mW, 635 nm, 5.6 mm package

The Strangeness is this: it dims when gently overdriven and temperature is allowed to rise, even just a few degrees. Freezer spray restores output, as does allowing the mount to cool, and on reapplying current, original brightness is seen. Basically if you push too far, it does not destroy its facet. It just goes dim. Why?

(Or in keeping with my avatar: Differentials? Go...)

(If you really DO push it too far, it permanently and politely declines to output its best ever again, but this dimming thing is different, reversible, and has a far bigger correlation with temperature than it should have.)

[Occularis]

Hmm... crystal/facet getting misaligned when it gets warm?
Are you using batteries or a constant power source?
Is it fitted in a module with a push-button for activation, if so, have you checked the assembly for bad solder joints? (same goes for any driver and LD-connection btw, check solder joints)

Just a few ideas.

[ElektroFreak]

Sounds like the wavelength might be shifting towards the IR due to the temp change.. although typically temp. related wavelength changes are nowhere near that pronounced. Have you checked output with a meter or visually?

Also, I've had some diodes that are intended to be pulsed act that way.

[The_Doctor]

Current's good, it's under constant control, so any slightly poor joint would not reduce that current. If it was really bad it would go open circuit so I know it's good. Also, I've seen this in other diodes from the batch, at many different times and circumstances. (And my solder joints rarely fail).

Misaligned chip? Nice idea, but doubtful. It's all one piece, though I suspect it might be insulated electrically from the metal heatsink, with enough thermal resistance too that it is noticeable. The dimming will occur fast, as soon as current is pushed beyond a peak very slightly above nominal maximum spec. I don't know if that's a fast temperature rise due to poor thermal coupling, or is a direct result of overcurrent. (Haven't tried to monitor that change with a photodiode and fast switched change in current).

No metering. I have a LaserCheck but to use that you have to know the wavelength for each measurement, and that will be changing too, so it's hopeless trying to keep up (and probably pointless given how dramatically the brightness dims with a very modest temperature increase). I also have a Scientech head and a simple circuit I made for it, or I can even measure directly in mW and convert, as I calibrated it from a steady power at known wavelength and the LaserCheck. It's NOT very sensitive though. It can read 3 mW or so but not with enough accuracy for this. About all it can do, which I might try in a day or so, is to determine whether wavelength or power shift is the dominant cause of apparent brightness change, but I'm fairly sure it's power, based on several years of careful physical observations of these (and similar) small diodes.

One other thing: visible specularity remains sharp at all times, even when a diode has been pushed to the point of permanent decline in output.

I can't get more precise observations, as I don't have a hundred-grand lab. The one thing I can do that would be worth doing is to see if a change in current alone, or temperature, is the main cause. I can also send a sample or two to someone with a hundred-grand lab, if they're curious enough to test it. Or any kind of lab.

[krazer]

I have observed this on the shorter wavelength diodes, they really do prefer to be run cold. My understanding is that due to the fact that they are 'right on the edge' of the bandwidth of the lasing medium, its possible to heat it enough that you kind of run out of gain and the power drops considerably. My issue with that explanation is that heating should lower the wavelength, which should make the diode even happier

[The_Doctor]

I'll buy that. Phil Hobbs said something similar on alt.lasers too. It fits, the thing LOOKED like it was on the edge of a region of well-defined behaviour, like it was simply happier (more efficient by far) being cold. Remember that the change in wavelength with temperature would be apparently far smaller than the change in efficiency for gain.

[heroic]

You might be able to see the line shift if you have a rigidly mounted diffraction grating (or compact disc...) Set it all up, mark the position of the line on a sheet of paper, and then push it towards dimming; if the spot moves off the mark, the frequency is changing.