The bike light write-up.
The inspiration for this project was some creations on this site, mixed with a few requirements and ideas of my own, but the trigger to start work was entering a charity 24hour mountain bike race and I was tasked to produce some lights for the team.
At this point I thought I would do a little more research and looked at other lights on the market. In my opinion the bike light benchmark is set by Lupine and their range of lights, The top of the range being Edison 10 HID light. This produces 980 lumen and has a 10 hour run time - but reading the specs more closely, 980 Lumen is only achieved for 6 hours, if you want the full 10 hours you need to drop to the 500lm setting. The cost £650 over $1200 USD. A lovely bit of kit but too pricey. There must be another way.
One of the reasons I chose Lupine as a benchmark was that they are one of the few who advertise the Lumen output and power consumption of their lights.
This gives the following efficiencies -
980lm/16W = 61 lm/w on high
500lm/10W = 50 lm/w on low (a lower efficiency at lower power)
Efficiencies of other Lupine Models:-
Nightmare Halogen (regulated)
550lm/25w = 22lm/W
Passibo Halogen (non-regulated)
Of interest also is the Wilma using 4*K2 producing 420 lumen at an un-advertised power which we will take to be 4W (1000ma*3.9v) = 16W(+) = 26lm/w.
Stop Press - Lupine are now using SSC/Cree and their Wilma is now 800lm!!
It did not take a genius to decide on using SSCP4 Ubin, it was also clear that to approach the output of the HID lamps multiple emitters would be needed. From studying the data sheets and some CPF testing, the key fact regarding LEDs became clear- they hate heat - simple as that. Installing a big heatsink only adresses the symptoms, the solution is to get the most efficient emitters and to drive many of them at the lowest drive current.
So what would I build?
Well,if the best lightmaker had four LEDs - I should have five. The total Vf for 5 emitters puts it outside the range of a readily available battery pack and buck type controller, I would need a voltage boosting current limiter. The fatman can only handle 16v (until the MaxFlex is sorted) which leads to the Shark driver. The shark could just handle 6 led but 5 gives a little margin on safety.
The potential output ( realisticaly assuming 95lm@350ma ) is 1050 lm at 1A an efficiency of 56 lm/w. Very pleasing but perhaps an un-necessary amount of light and a 20watt demand on a battery would make it difficult to get through the night. It also does not take advantage of the fact that LED efficiency rises as the current is reduced. Driving the LED at only 500ma raises the efficiency to around 75lumens a watt and still be producing over 600 lumen, a fine bike light and surpassing the output and efficiency of my benchmark light.
The Shape of the light
The most LED you can get behind a single lens (that I have seen for sale) is Four, Kahod and others sell a 50mm diameter lens with 4 integrated mini-lenses. If I wanted 5 I would be on my own and using some combination of single lenses. I could position these lenses (or reflectors) in any position I considered reasonable.
I could have copied Nightrider and had Leidel square lenses, but I thought that they were too expensive and I had some suitable lenses already from Best Hong Kong @$0.75 each, I got both types of lens (both listed as 15deg?) so I could mix and match to get a good balance of flood and throw. Also I reasoned that if you are putting out a huge amount of light then you would be blinded by a really tight hot spot overwhelming the side spill and destroying your peripheral vision. My aim was to get a broad soft edged flood - not impressive for the white wall hunters but more useful for riding a bike.
Its funny how one idea keeps coming back no matter how many directions you try to take the design in. For me the recurring theme was a thick backbone strip of alloy that the LEDs are mounted to. The only problem this raises is the mounting. A single central mount leaves a large distance from the mount to the end of the light which would wobble around and be very vulnerable to damage from stray branches. However, if you put a mount each end that problem goes away, and there is no reason to have the lights sitting on top of your handlebars, it can equally sit infront of the bars.
I decided that a chunky 1/4 inch back plate would hold the LEDs and mounting brackets. The 1/4 inch thickness would also allow me to tap into the sides with confidence (crucial for my design). A 1" alloy channel would form the top front and bottom of the light.
I used Google sketchup to produce some nice sketches which got me up to a point, but there were still some last minute changes for example when I realised the Shark would not fit in the space I had left for it.
I think the following photo's are probably more descriptive than my words
Lets try this link
The worst part about trying to specify the power requirements was that I did not know how long I would need the lights, I would be riding the 24hour race in relay with 3 other riders, at least one other would be sharing the lights. I knew the sunrise and sunset times, but how does that translate into the light required to ride a bike in Forrest on a cloudy night? It was also not clear how available charging during the event would be. I could however confirm what I was trying to do would work and that my back-up batteries would work.
Max output 1A*18.6V (total Vf of 5 LED) = 18.6 Led Watts
18.6/0.85 = 21.9w battery demand after shark efficiency
Shark Max current input = 4A
Shark Max preferred current input = 3A
21.9w/3A = 7.3v = Minimum battery voltage required
ie my back-up 7.2v Tamiya batteries would need to discharge at around 3A to power the light on full power. Lap times were projected at being around an hour, there was a fighting chance of getting a full lap out of the batteries.
The decision for the primary battery pack was finaly made with the charger- the AllBattery charger for 1-4cells was 0.5Amp and the charging rate for the fast charger was 1.5A but was only available in 4cell 14.8v. So I went for the 4.4Ah. 14.8v pack the current drain is 1.48A well within specs of both the battery and driver giving a 3 hour burn time.
Well, the light was used on two bikes in the end, The result was that our night lap times were within fractions of our daylight times. I was certainly aware of charging on the downhill sections and passing better cyclists who had overtaken me on the climb-up. This was not only due to having more light, but as I got closer, my lights dwarfed theirs, and as they were attached to my wiggling handlebars, the rider infront was "strobed" until he fell off or let me past. Although not as funny as a guy who was caught on a road section, looked back at the lights, got dazzled, thought it was a car and cycled into the bushes at the side of the road. (a little unkind - but at 4am and into your 5th hour riding you arent exactly yourself)
£100 ~$200 split pretty evenly between the head unit and the batter/charger
but thats not counting one dud shark, two LEDs (ripped off gummy domes) the bunch of reflectors and lenses and current limiters I didnt use and the Epoxy and metal and the diesel to drive to the milling machine.
But it was fun
Last edited by Doh!Nut; 08-02-2007 at 09:49 AM.
Very nice work Doh!Nut !!! (A light like this really deserves a name too )
A successful project from your night lap times, part about the guy riding into the bushes is too funny (hope he wasn't hurt).
Your picasa gallery shows the Shark driver with stock potentiometer in place, have you considered a remote mounted 20K pot or Remora UI board? This would allow brightness adjustment when max output isn't needed, helping extend battery life and lumen maintenance as well. As if this weren't bright enough already, your light will be be even more incredible when the 135 lu/W P4's are released !!!
To link the images, uploading to a free host like imageshack.us works well, they provide the link code for you to cut and paste into CPF. You can also use the IMG tags too, just type [IMG] before your image URL and close with [/IMG] and it'll show the pic in your post.
Nice simple and robust looking design. I do wonder about heat dissipation, since you have 5 LED's. The aluminium bar that you have attached them to will have a nice bit of thermal mass but in the end you still need to get rid if the heat and for that you need surface area.
You should post your design up on the SingleTrackWorld forum.
Two points were well made, cooling is not as efficient as it could be (surface area/mass) and there is only one power level.
The original design was to have a much lighter backplate but have a direct metal to metal link with the handlebars via metal end-plates, effectively using the bike as a heatsink, when I realised I would have to swap the light between bikes I decided to use the plastic quick releases. Certainly there is some scope in thinnning the back plate and bonding on a proper heatsink.
Having said that the light only got hot on the long slow climb, by which I mean noticably warmer than the handle bars but no-where near uncomfortable. A quick downhill blast would bring the temperature down very quickly.
The remora is nice but if I understand its use correctly you can only select the power level on start-up and must turn it off for 5 second to re-select. What I think is needed is a high/low switch changing between two pre-set resistances in-place of the potentiometer.