A trio of 5 mW laser modules arrived with a bunch of other surplus gear after an end-of-year sale:
It runs on 5 V at 20 mA, determined by the 91 Ω SMD resistor soldered across the terminals at the back of the PCB. That suggests the laser diode itself runs at about 3.2 V: 5 V – 0.020 A * 91 Ω.
The brass case connects to the red (positive) wire, so you must insulate the laser module from the usual grounded metal chassis.
Two of the three lasers arrived badly defocused, but a twist of the brass barrel broke the sealing glue and a bit more twiddling found the sweet spot.
Running one of these from an Arduino would be just like the UV LED: redefine a bit in the shift register bitfield and drive the laser with a MOSFET switch.
I’d be tempted to bypass the SMD resistor and run it from an LM317-style current regulator hitched directly to the raw battery; I’m pretty sure I have some LM317 regulators in TO-92 packages. The sense resistor would be 62.5 Ω = 1.25 V / 0.02 A, dissipating 25 mW = 1.25 V * 20 mA. From a freshly charged 7.2 V Li-ion battery at 8.5 V, the regulator would dissipate something like 80 mW =(8.5 – 1.25 – 3.2 V) * 20 mA.
Or just add more series resistance and ignore the brightness variation?
The Smell of Molten Projects in the Morning 
Actually, you must be really careful running them that way (simple current limiter). There are many ways to destroy a laser diode, voltage spikes on startup is one. The interesting part is the why they fail. A laser by definition has the intercavity beam (the Amplification part of the acronym) which can be thousands of times brighter than the sun. Realize, the beam that shines out is doing so between the very atoms of the mirrored surface of the diode structure. So a rapid voltage pulse causes a flash inside the diode to be so bright it burns it from the inside out. This can be really obvious as a failing diode becomes dimmer given the same input current and voltage, since you have modules with adjustable/removeable lenses, either unfocus or remove the lens for the widest beam and examine the output for dark stripes in the line width. This indicates damaged portions of the crystal sides, blocking the light, which then keeps the heat in causing more damage until it self destructs.
Anyway, the best info and no lie, the first pages I ever visited on the internet back in 1996 ( I know I was late in the game) http://www.k3pgp.org/Notebook/Lasersam/laserdps.htm#dpstoc
Also, some thoughts on laser operation. We charge a medium (crystal in this case) with energy. Atoms randomly drop electron valence shells and shoot photons in random directions. Only the ones that are dead parallel to the mirrors are the desired ones and they hit other atoms causing the photons to be released in phase ( part of coherence). What this means though is that we produce light brighter than the sun and throw most of it away as heat, which is why you’ll see the abysmal efficiency of any laser. For example, my 3.5 Watt whitelight mixed gas Spectra Physics 168 takes in around 12 kW of 3 phase 208 V and requires water cooling for the previously stated 3.5 W out. It’s an “on demand” water heater with a laser attached.
That’s an interesting diagnostic; the unfocused beam from these is a rectangular bar, which I interpreted as an image of the end of the crystal, and a quick check shows solid illumination across the bar. So at least one laser is in good shape…
The line and load transients for an LM317 regulator are in the microsecond range with no load-side capacitor, which is well below the millisecond-scale thermal response of the die. Adding the cap pretty much eliminates the transients: sounds like I should go with a well-bypassed current limiter!
[Update: Browsing that FAQ, it looks like the thermal response is irrelevant: optical power damages the mirrors long before the semiconductor die heats up.]
Now, that sounds like a perfect justification for a home laser!
Right, like Jetguy says, it’s optical power that does the damage – which makes it an incredibly fast damage mechanism. A tiny spike, containing very little total power, will be ruthlessly converted into light, and the light ruins the facets, and thereby the laser – considering the total volume of the lasing area is microscopic, the light intensity inside it is astronomical, even under normal working conditions. I used to work at a high speed fiber optics lab, pushing 3GHz signals through fiber (back in the 1980s, when this was considered fast). The tiny lasers we used were amazingly sensitive to damage from voltage spikes, and we couldn’t just hang capacitors on them with impunity, as that would ruin our bandwidth.
I’ve been using LED drivers (unsurprisingly) to run mid- to high-power laser diodes with reasonable success. I just put a lot of output capacitance on them. It does seem to be overkill for a 5mW one, though, but it’s done nicely for the 250mW one I chuck up in the Sherline. That one came with a 317-based supply, but I swapped in this one because the dim pin made interfacing really easy.
Paper cutter?
Oh, and other notes. So I knew it was theoretically possible but actually killed a 250mW diode a while back through reflection. I was trying to align the 168 and was stupidly using a diode laser, not a HeNe. It was even through set of mirrors at 45s, but when the OC (Optic Clear or beam end mirror, not the HR High Reflector) aligned, it returned the beam right into the diode and killed it dead from optical overload. So shining 2 lasers at each other or hitting a mirror at dead 90 can kill one too and fast. Again, I had heard of this and thought it would be dificult to reach the alignment and precision required, but it only takes a pass for microseconds through that perfect window of angle to kill it dead. Prices have dropped to the point you’re not out quite so much money for an accident, but you are always ticked off when it does happen, even on a cheap $3 pointer. HeNes are outdated, but in some cases, a lot more abuse proof.
Sounds like one of those tricks you see in the movies, where the protagonist reflects a death ray with his sunglasses and the bad guys explode.
That’s about what these are, so I can write one or two off as an learning experience. You folks definitely help me to not start from really dumb!