Re: how is a laser bar assembly like this colimate
Forget lenses, even rod lenses (except, perhaps, for the slow axis).
Use two cylinder mirrors to collimate multi-emitter, multi-mode diode bars, stacks and arrays.
And if you are going to pump a crystal with it, you can use two cylinder mirrors to focus the fast axis into the cavity from the side, along with a rod lens that you place at the distance which gives you a slow axis beam width equivalent to the length of the rod you are pumping. Not that you really need any collimation- given proper cavity design- to come within a hair of a diffraction limited beam, at least not with an actual rod (some places will say 'rod' when they mean 'bar').
For fiber couplling, a common technique is to use a gradient index lens to focus a collimated beam into the fiber, or to use a microscope plan objective for the same purpose. The latter has the benefit of letting you test the setup with a far safer red laser before you actually use the CS mounted high power pump diode (which needs excellent cooling for Nd:YVO[sub]4[/sub] pumping, like a heat pipe based CPU cooler with a fan, or precise cooling for Nd:YAG, which absorbs a narrower spectral band).
As others have said, a CS mount is dangerous. No second chances. The smell of bacon may be the only clue something went wrong, unless you get it into your eyes, in which case the pop and the blindness will be a reliable indicator a few seconds before the bacon smell. Diffuse reflections are not safe, either. Just remember the goggles, use pilot beams, and make damn sure you know what you're doing. Also, a lid to prevent light leakage and accidental reflections is worthwhile, though the full CHDR guidelines are worth a read. Don't pass on anything powered by a diode bar to a hobbyist, unless you suspect they're too stupid to avoid trying to build one for themselves without the requisite skills and knowledge (harm reduction). A sufficiently skilled hobbyist- one that can treat such a diode properly- will either know what they need to make one for themselves, or will acquire that knowledge (as you are now attempting to do).
Fiber delivery is more expensive than free-space, but worthwhile, as it allows the system components to be sealed, without losing the ability to work with them to get the results you want. Grin lensed ferrules are available at a reasonable price, and will provide you with a 1.8mmØ collimated beam. For burning (those goggles are, again, a must!), a half-ball lens can be optically glued to the grin lens to focus the beam. The sapphire ones will have a shorter focal length than the glass ones, but are more expensive.
The average 20W pump diode can be used for fine soldering electronics, CNC milling of PMMA, acrylics, and a number of other applications, in addition to the obvious (pumping a crystal). It can not cut metal, since the focal spot will never be small enough, but it can probably sinter some metals (consult a table over melting points, thermal conductivity and so forth, along with a graph showing absorptivity at 808nm).
Needless to say, do not power one up around people (or pets) who aren't laser savvy.
Really neat devices, though, just like cars, rifles and guns.
Final piece of advice...
The diodes from snoctony are an expensive place to start. New diode bars are available from half the cost he is selling them for. Used ones are even less expensive, and suitable for testing to see if you have got the rig set up properly, although you will have to realign when you switch to another bar.
Hope this helps.
PS: If you're knowledgeable about high voltage electronics, you can pick up a 3µm thick mylar foil, aluminize or gold plate both sides of the foil, mount it, measure the wave front, and then set up a wire mesh behind the foil to match the wavefront. Apply a high voltage differential (500-4kVdc) to the mesh and the back coating of the foil to deform it into the correct shape for collimating the beam. With enough work, you can improve beam quality by an order of magnitude. The telescope crowd can give pointers on this, though they usually deform it with gas for a spherical contour.
PPS: Should you need more than 40W CW, or 100W QCW, you can join the collimated beams with a reflective diffraction grating, so long as it's a type that can withstand the power level in question. Use the full area of the grating, and compress the beam afterwards if it's too big for your purposes. Also, llook into how phase masks and zone plates work for more ideas. Wedges are your friends.
