Finally, I have a chance to start discussing a project that has been brewing in the back of my head since July, when I purchased a broken portable laser for peanuts. My original plan was just to refurbish the beaten up host and replace the damaged diode, but as time passed, my plans grew. They’ve finally begun to settle down now and I’m well on my way toward finishing plans for a 532nm lab laser constructed pretty much completely from scratch. My goal is to make the laser as efficient as possible. Doing this is not just a matter of finding the optimal diode positions and so forth; it requires careful temperature control of the nonlinear optics.
The crystal set (which is the main reason I purchased the laser) is particularly good for three reasons: the crystals are very large, they are unbonded (that is, they weren’t glued together by the factory), and they are already rotationally aligned (they have to be oriented within some small specific relative angle range in order for good efficiency and these crystals have already been cut so that they can lie flat on a heatsink without having to worry about the orientation).
My plan is to design easy-to-machine parts that will fit together nicely while still providing a good level of flexibility for tuning. I will need to create heatsinks for the crystals, a heatsink for the diode, mounts for the optics, etc.
The laser will be driven by an LM338-based voltage regulator, which will provide the necessary 2.3A or so to the 2W 808nm C-mount pump diode. The diode and YAG crystal will both have thermoelectric coolers with thermistors embedded into their heatsinks. A microcontroller will take input from the thermistor and based on some optimal temperature, use PWM to individually control MOSFETs for the diode and crystal TECs. Furthermore, the KTP crystal will have a resistive heating element attached to it as well as a thermistor embedded into its heatsink; the microcontroller will also try to keep the crystal at its optimal temperature.
From my research, it seems that KTP likes to be very hot; sometimes around 90C. YAG, on the other hand, likes to be somewhere between 10-20C; these temperatures can vary greatly based on the chemical structure of the crystal. The microcontroller will try to get the TECs and heaters to keep the crystals at their optimum temperatures.
To determine the optimum temperatures, I might try to build an automatic temperature optimizer using the analog output of my Scientech 372 power meter. The microcontroller would read this input and watch it as it reaches a maximum.
A few simplified diagrams of the laser:
Because the MOSFETs for the TECs and heater will be generating currents on the order of several dozen amps, a lot of heat will be generated in the rear portion of the laser. I will place a small exhaust fan on the back for that reason.
The most mechanically complex portion of the laser will be the crystal mounts. In order to obtain maximum heatsinking, I need to build something with as much contact area as possible. I came up with the idea above; I’ll use springloaded screws to keep the side portions snug against the crystal as well as more springloaded screws to mount the side portions to a thermally conductive bottom. This will provide even heat sinking around the crystal. Furthermore, a thermistor will be embedded into the bottom so that the temperature reading at this location is very close to the temperature of the crystal.
A TEC needs a heatsink on its other side to dissipate heat; a resistive heater, obviously, does not. However, the lower portion of the heatsink for the KTP crystal will look very similar to that of the YAG.
A few words on the microcontroller: I will add a manual bias function. Essentially, there will be a three-way switch with states {auto,yag,ktp}. Choosing auto tells the microcontroller to ignore all biases. Choosing one of the others allows you to control the temperature bias with the potentiometer. An LCD screen will display the temperature of the thermistors on the diode, YAG, and KTP.
I’ve already begun to collect part names as well. For the AVR, I will use the Teensy development board. It’s perfect for this application.
The plan is to machine all the parts here at MIT’s Hobby Shop and build the laser itself at home this January. IAP is going to be pretty fun. 
I’m keeping a running page on my work here.
