Summary: blue laser / coupled to optical fiber using a Melles Griot part and a (simple) custom-machined adapter flange / long run of fiber to coupler on rats head / short fiber from coupler into nut piece and brain.

Details:

A. Blue laser: 473 nm DPSS, 80mW nominal (actually ~100 mW), TTL control purchased from Laser Glow (laserglow.com) for $2689 (part number: LRS-473-AM-50-3). Note that we're currently negotiating with a vendor (Sanctity Laser) in china for a yellow laser; their prices for the 473 nm parts are probably about 25% cheaper. Our laser allows from TTL control but not output level control. See (B). The DPSS laser that we purchased is somewhat suboptimal in that it has a lot of high-frequency noise (though this probably doesn't matter) and that it really needs to be either mated to a large heat sink or actively cooled if you run it in steady state. Pulsing seems to be fine, with control up to about 1-3 kHz no still producing square wave output (beyond 3 kHz, the onset transient becomes more apparent). Purchasing from a more established vendor (e.g., JDSU, Coherent, etc.) might result in better stability and less noise, but their products are MUCH more expensive. Same goes for the 488 nm diode lasers. (488 nm is directly produced by some sort of crystal - 473 nm is frequency doubled, and the 946 nm source wavelength is produced by pumping another crystal using like a 1058 nm diode, hence "Diode-pumped solid state" or DPSS).

B. Melles griot coupler (09 LFM 001). We had a flange machined with a threaded hole the same diameter as the laser head output, and mounting bolt holes for the coupler. Note that it goes directly from the laser head into the fiber; it might be preferable to have a free space beam path that you could put a variable neutral density filter or polarizer (the beam is polarized) to controllable attenuate the output power. Also note that at the time we purchased this coupler, it was the best option we could find. Thorlabs now sells a very similar part that might be equivalent or better.

C. Fiber: We use 62.5 um core / 125 um cladding fiber. The connectors that are on the rat's head are LC connectors (which are small and plastic). (And a coupler that mates two LC connectors together.) There's a shop across the bay (Fiber Optic Cable Shop) that has made cables for us: 1 long cable with an SC connector on one end and an LC connector on the other to go from the laser to the rat's head, and many "pigtails". The pigtail is what we use to go into the brain. It has an LC connector on one and and no connector on the other end. Note that all the cables are "buffer only", meaning that they consist of glass core and cladding, and a thin protective layer of plastic insulation ("buffer"), but not the heavy orange outside protective layer that you usually see on data cables ("jacket"). The buffer is still to big to stick into brain, so we strip down to bare glass for the part that goes into the nut piece. We bought a "Micro-strip" fiber stripper from Thorlabs to remove buffer. (The buffer is 900 um in diameter.)