For anybody other than Beaky and I who is reading this thread, I'd say just buy some Cointaker or Ablaze leds, pay attention to colour, polarity and stick 'em in - then enjoy fiddling, moving them around and playing with your toy which is why we're all here. Then have a beer to celebrate.
For Beaky and anybody else who is still following or even remotely interested in the to and fro of the techno bable and willie waving that we're both engaged in I think we've just about reached the point where we can't add any further and will just have to agree to disagree. I'll have one more attempt and await your response telling me where and why I'm wrong. I take it all in good humour, feel and trust you do too, and the only thing that is offended is my sense of good elegant circuit design!
I reckon you'd benefit from looking around and reading up on current design trends, ideas and thinking. This application note on driving leds put out by Texas Instruments puts it pretty well.
http://focus.ti.com/lit/an/slyt084/slyt084.pdf Have a quick read of the section headed "Direct Control of Current" for a good summary of why not to use voltage control and why to use current control.
The reason people have had to use the leds with resistors to stick them in pins is much like your argument that you continued to wire in parallel because of a legacy machine. Basically it's a cheap, inefficient cludge, but that doesn't mean it's good practice and certainly wouldn't be the way to go if starting with a fresh sheet of paper, either for a pin or a commercial lighting product.
Some of the arguments you've put up are either the same as what I'm saying or support those ideas, some of them you selectively and arguably wrongly interpret. A classic being your point that the Vf across leds will be different, this is precisely why you want to put them in series and regulate current.
The 724 data sheet that you quote is largely irrelevant - it's just a dc to dc regulator. A poor old design doesn't become a good design just through passage of time or number of units made. It's a case of using not the best technique or approach for the application - nowhere does the 724 data sheet mention its' use or suitability for regulating leds - that's 'cause it's not that suitable. All it does is get you a voltage that works with the combo you have, the resistor is doing the real job. Take a look at the Maxim MAX1698 chip, specifically designed for driving leds. Notice that they talk about current regulation of multiple parallel strings of series leds and describe a trivially simple and standard method of protection for the (unlikely) event of a led failing open circuit. Notice how easy it is to regulate intensity over a very wide turn down ratio. This would be an ideal candidate (simple, efficient, handles up to 5W) for the application that brought me into this discussion; namely driving, with good variable control, GI leds.
Between 2005 and now there have been enormous leaps forward in led technology and an explosion in efficient led use and it's likely these specific chips weren't around in the period 2000 to 2005 (current regulated designs were then and they remain now the most appropriate way to control led brightness) - luckily the chips are here, available now, simplifying things a bit.
For a really nice off the shelf, just install solution have a look at the LumiLeds buckpuck controllers.
Anyway, enough from me, thanks for the good humoured discussion and I'm happy to let you try swim against the tide, eventually you'll end up down river. I intend this to be my last post in this thread unless you come up with something really good, exceptionally bad or stunningly controversial. All the best, hope you leds burn long and bright, I am now going out the back to bury the soldering iron. Cheers, Firepower.
