Oh... um, send me a note offline and I can help you design a LM3404HV that will take up to 72VDC and provide a stable 700mA output. National Semiconductor has some nice simulation tools to deal with it. Did you try using those and were still unable to get it to work well? I'd also mention that the current varying between 650 and 750mA depending on input voltage isn't *that* bad of a scenario. Generally an end user can't tell the difference anyway, especially with the dramatically different LED efficiency bins you can buy. I went with the most efficient LEDs because I like nice things =)
Slashdotters are always critical, especially of MIT alums. I'm used to failing at things, whereas most slashdotters don't bother doing something unless they're guaranteed success. *shrug* Guess who's more likely to actually make something useful in the long run? God knows I learn an awful lot more from taking on a lot of very hard challenges and doing my best, succeed or fail, than any armchair engineer. Despite being a physicist and material scientist working on catalysts for renewable energy, I had no problem jumping into and being wildly successful with designing ultra-high frequency digitally controlled PID feedback loops (literally a 5MHz response time), building these and similar LED lights, and so on. I just didn't have the pleasure of taking classes on it, or working at a company doing electronics design so I had to teach it all to myself over the last four years. I think that regardless of what armchair engineers might think, teaching yourself something in your spare time and doing it well enough to produce useful things is not bringing shame on the institvte. The funniest comments are from the people bashing my EE skills. Usually, the solutions they lay out are not workable in a reasonable real world system, once you think it all through. It's sort of like those hilarious comments where someone brings up "what if gravity doesn't work" in a thread about GR. Seriously? Or mentioning basic statistical analysis and sample bias as the "reason" why a 3 year study is wrong. Gee, who do you think knows more about statistical analysis... I can't imagine they tried to account for things that you learn about in high school math...
Anyway, I am serious about collaborating... at the least, my site has some publicity now, would be happy to help sell things, and would be happy to be a sounding board for new ideas as well. I personally checked out the existing hobbyist offerings at sparkfun and also found them to be overpriced and annoying to use. However, there is definitely a place for discrete controllers and chained rebel star boards. It's just not in a quasi-professional lighting fixture that I'm (actually) expecting to make a lot more money from selling to clubs than to hobbyists.
You know what would be awesome -- check out the new 40W RGBA LEDs from ledengin. That's what I plan to use in my next fixture, after I make back enough money (hopefully) from this to design new stuff. I want to put three of them in each light, and use the same power electronics. The hard part is fitting those huge optics and dealing with heat sinking... I might end up having to use active cooling if I want it in the same 5" cylinder form factor (which I do!). But, if I can solve that, which I think I can, I have a 3000 lumen fixture in a tiny size! Awesomeness. Naturally, too expensive for slashdot to approve; surely they could do it for $40 due to my idiocy, but whatever... daddy needs new brake lights for his car...
Email: neltnerb@saikoled.com