Just tinkering with something 638nm RED 8W and bright as hell.

[Photonbeam]

Indeed, 100um will deliver a nice round very very huge divergence red pyramid.

And what this would cost does not bare even thinking about.

I was thinking for a few years about tapered fibers and they do indeed exist but they also do not go to a very small size, maybe 400um down to 200um. The cost for one was about 2,000 dollars and not a lot to gain from it.

No free lunch. I experimented with tapered fibers in the '70's. You can decrease the fiber size but the trade off is change in NA = greater divergence - trying to collect and collimate the output, so you're back where you started.

[steve-o]

"You cannot change the laws of physics"

[andythemechanic]

"You cannot change the laws of physics"

Sad but true.

[planters]

There is one way that does not defy physical laws and does work, but to a limited extent. Wavelength division multiplexing can allow a higher brightness than even the best packing and stacking. Because of the large delta lambda /delta T with these red diodes, if you cool/ heat a diode approx 20C the wavelength shift is about5-6nm and this will allow you to overlap gently converging beams in a prism. This is done with IR diodes with a more elegant approach to produce higher brightness out of a fiber, but this cruder method works and I have done it. You need to hold the diode temp to better than 1C to maintain alignment and the set up is bulky, but it works and is inexpensive.

[steve-o]

There is one way that does not defy physical laws and does work, but to a limited extent. Wavelength division multiplexing can allow a higher brightness than even the best packing and stacking. Because of the large delta lambda /delta T with these red diodes, if you cool/ heat a diode approx 20C the wavelength shift is about5-6nm and this will allow you to overlap gently converging beams in a prism. This is done with IR diodes with a more elegant approach to produce higher brightness out of a fiber, but this cruder method works and I have done it. You need to hold the diode temp to better than 1C to maintain alignment and the set up is bulky, but it works and is inexpensive.

Then you could use this method with a 637nm diode and a 642 nm diode (No cooling/ heating required) .. ?

[planters]

Yes. But you are limited to just a doubling. The way the IR diodes do it is the cavity includes a wavelength selective reflector that shifts the output a couple of nm as you step down the bar.

Andythemechanic,

So, you work @ Dilas. "The Diode Laser Company" Their products are really impressive. I am working on this neat "Z" fold DPSS and I was hoping to move up from F mount free space diodes to fiber coupled modules. And so... I was wondering...you know

[Soulstorm]

There is one way that does not defy physical laws and does work, but to a limited extent. Wavelength division multiplexing can allow a higher brightness than even the best packing and stacking. Because of the large delta lambda /delta T with these red diodes, if you cool/ heat a diode approx 20C the wavelength shift is about5-6nm and this will allow you to overlap gently converging beams in a prism. This is done with IR diodes with a more elegant approach to produce higher brightness out of a fiber, but this cruder method works and I have done it. You need to hold the diode temp to better than 1C to maintain alignment and the set up is bulky, but it works and is inexpensive.

Know of someone who did this prism trick, with nine diodes.. They said it was about 18m of beams just to align it and an absolute nightmare. Bright as hell but very impractical.

[planters]

I did this with three beams and it was bulky, but pretty easy. The practicality...well. I think that the idea that a bar of closely positioned diodes can be a source of bright red might be viable, but the only way to even equal the mm/mrad numbers of a well knife edged, multi-diode set up is if each emitter in the bar is individually addressed. This would have to be done at the micro scale and the tolerances would be minute. I will be astounded to see if this has happened, but it would have the advantage of a monolithic structure and less prone to miss-alignment.

[Soulstorm]

Nice work. A much more monolithic solid state solution would be brilliant.

What about mechanical mating with a microlens array? Some CMOS cameras have them in single digit micron range, so it should be feasible. But as you say, tolerances would be crazy. None the less, if it can be done with even moderate levels of waste, it's probably viable in the long run.

[planters]

I doubt it would work. At a competitive level of alignment, lets say (+/-) 1 mrad , for the individual beams, then with a large number of emitters in the bar this should ADD 1.4 mrad to the blur circle in the far field. I can't work out the centration requirement for a micron scale aspheric collimator, for example, that would produce this level of angular deviation. But, if we look at a similarly small (say 10um) reflective surface that acts as a mini-knife edge then; 1mrad x 1/2 for reflection x 5um radius =2.5nm mirror edge positioning. This would have to be measured and then set by the robot during the manufacture as is done with FAC lenses and then maintained by an organic adhesive once it is cured. Oh boy!