Originally Posted by
dhalbrook
OK, I'll admit it was a mistake to say "unequivocally". The overall drag of the Magnics compared to the Cyo/SON is significantly lower.. a simple wheelspin test confirms that easily enough. Of course, the light output is a fair bit lower as well, so it gets a lot murkier in terms of overall system efficiency. Personally I haven't performed any tests that are anywhere close to scientific, so I can only tell you my conclusions based on my observations, namely that (and yes, I understand Lux is highly dependent on beam pattern) in my estimation the Magnics are roughly 1/2 as bright as the dyno setup but have far less than 1/2 the drag.
AFAIK Dirk has investigated the actual system efficiencies. IANA electrical engineer so I can't pretend to follow it in great detail. There is some info here:
http://www.kickstarter.com/projects/...ess-bicycle-dy (under What's new in Magnic Light iC?) regarding the LED efficienty but I'd love to hear him explain it in greater detail. I can ping him and see if he can provide something more substantial if you're interested.
[MENTION=90674]dhalbrook[/MENTION]: Ok, I haven't yet taken a good look at Dirk's new Kickstarter page, but I am an EE and I'll try to assess what he says from that point of view. If you also gain some more insight from him, that would be great.
[MENTION=134161]dougmc[/MENTION]: I applaud your rigor. I also support seeing some measurements. But on the optical side I think the ideal is to measure the ability to illuminate that which is useful to see, not to measure total light output, or electrical to optical energy conversion (lumens/watt). Depending on lumens to characterize optical output gives a product positive marks for spraying light where it isn't needed or might not be wanted.
A focus on lumens rather than illumination means that the best light is one with no light distribution control system - no light guides, lenses, reflectors, Fresnels, blackouts, et cetera. LEDs without lenses do not radiate full intensity into a 2pi space and the distribution is predictable, but it places a lot of its light where it might not be wanted or needed. Clearly one can use 10 degree (or narrower) factory-lensed LEDs to approximate spotlight functions. But that isn't what some of the better generator lights are doing (i.e. Lumos and Edelux 2), they are using sophisticated light distribution controls to approach the behavior of European automotive lights. For those of us who think that's the best model for bicycle light performance, illumination needs to be the target metric.