Simple DIY generator-driven LED bike light - lots of lumens

Early this year I built a super-simple generator-driven LED light as a
test of concept. No driver board or regulator of any kind is needed.
Details are below.

It works great, generating something like 200 lumens from an old
Soubitez 6V generator. That's more than enough for very comfortable
road riding, even with my poor night vision, all from a sidewall generator!

Total cost around $47 plus generator and lots of hobby time to make the
housing; can be cut to ~$36 plus generator with a rectifier, see below.

I used the Soubitez 'cause I had one in a drawer. I'm thinking
seriously about a cheap dynohub; this is for my commuter, and I do in
fact forget to charge my battery lights from time to time.

The concept I was testing was the current regulation apparently inherent
in bike generators/dynamos. What I'm told by people who seem to know is
that such dynamos will pretty much top out at 0.5 amps, regardless of
the load they are wired to. The peak voltage rises or falls (or so I'm
told) until the 0.5 amps goes out. Drag at peak voltage (and the speed
required to reach it) varies with the voltage your load requires for 0.5
amps. (Or so I'm told)

While I haven't put a meter or oscilloscope on the thing yet
(front-wheel generators are hard to bench test w/o rollers), I get LOTS
of light with only normal drag, and I haven't managed to burn anything
out, even with some high speed downhills.

Here's the design:
* 4 Seoul Semiconductor P4 LEDs, U-bin from http://www.dealextreme.com
(in Hong Kong, about $5 each)
* Housing hand "carved" from 1" square aluminum tubing,
cheap at Home Depot. Many thanks to Allen Chapman's site
for his housing design: http://bikeled.org/
* Lenses and lens mounts from http://ledsupply.com,
two 5 degree, two 15's, about $4 per LED
* Proper heat sinking is important, I used Arctic Alumina to mount the
LED stars, $15 from coolerguys.com
You won't need the entire tube for the build, a small
consolation for the high price.

Wire two pairs of two LEDs each in series, then wire the two pairs
front-to-back as a rectifier bridge (ASCII art below). Attach to
generator and go ride!

Note that with the under-$5 purchase of a rectifier, you could run the
same design, and get nearly the same light, with only 2 LEDs in series.
That makes housing construction easier. As it is, each LED gets a
theoretical peak amperage of 0.5 amps for half the time; with a
rectifier, that would be 0.5 amps all the time. These LEDs are rated to
1.0 amps continuous duty, and with proper heat sinking I know they work
well at that rating. I've run a battery-driven, regulated version of
this light for a few years, so I know what the LEDs can take.

Circuit:

|-----||--||-----|
| |
to gen. ---| |---- to gen.
| |
|-----||--||-----|

You can manage with much cruder housings, too: Just mount the LED stars
on an aluminum bar, and glue on the lenses with silicone. Your lenses
are subject to getting banged up, though.

If you have more money than time, this isn't for you. On the other hand...

Just another data point in the battery-generator wars.

Mark J.
Happy user of both battery and generator lights, each for its own purpose.

In article ,
"Mark J." wrote:

snip

Thanks for posting this. Photos would be cool, too, if you have any you
care to put on the Web somewhere.

Happy user of both battery and generator lights, each for its own purpose.

A logical and reasonable viewpoint.

Tim McNamara wrote:
In article ,
"Mark J." wrote:

snip

Thanks for posting this. Photos would be cool, too, if you have any you
care to put on the Web somewhere.

OK, I wrote up both my battery-driven and generator-driven builds, with
photos. I've also included my personal analysis (based on energy
expenditure) as to when each is preferred:

http://home.comcast.net/~mandmlj/BikeLights/index.html

Enjoy!

Mark J.

On May 6, 12:03*am, Andreas Oehler wrote:
Wed, 05 May 2010 15:32:13 -0700, Mark J.:

OK, I wrote up both my battery-driven and generator-driven builds, with
photos. *I've also included my personal analysis (based on energy
expenditure) as to when each is preferred:

http://home.comcast.net/~mandmlj/BikeLights/index.html

The wheight comparison is a little unfair: A Shimano XT front hub weights
150g. The SON delux is 385g.

5 AA-cells with a battery-holder weight 160-200g - which is nearly the
weight difference between the XT hub and the SON delux. The 5 AA cells
deliver about 15-16 Wh in the best case. This is 4 hours powering a LED
headlight with 4 Watt - comparable to a SON delux powering an Edelux. With
a battery light it is essential to always have a set of spare batteries at
hand - expecially if you want to use it close to their nominal capacity.
With a spare set of AA-cells you have already to carry more wheight than
the hub dynamo solution.

Andreas - working for the SON-manufacturer...

That's an interesting article Mark has there, and all the better for
his lamps being shown in the prototype stage rather than their
finished perfection.

I was just working through all this when your post came up, Andreas. I
don't think it's two sets of batteries Mark needs but in fact three.
He talks of through-the-night 24hr rides. Nowhere is the night only 4
hours long, so he needs two sets of batteries plus a spare set. The
only way the battery light will seem "superior" is if he turns it to a
lower output for at least part of the night, in which case he might
get away with two sets of batteries. Two sets of batteries aboard is
the weight of a hub dynamo or more already; maybe in LiPo the
batteries might have a small weight advantage or at least stand on an
equal footing.

I have a 6V battery bottle for some halogen lights I no longer use,
and I wouldn't care to lug three of those around. A hub dynamo is in
fact pretty conveniently disposed weight from both a centre of gravity
and an acceleration viewpoint...

Depending on the weather and the roads, a sidewall generator might be
as good for Mark's purpose as a hub dynamo, and it could be
substantially the lightest option here.

Andre Jute
Neither a SON nor a Shimano employee, though I have hub dynamos from
both, and find both excellent, as I do the AXA HR sidewall generator I
have on another bike.

Andreas Oehler wrote:
Wed, 05 May 2010 15:32:13 -0700, Mark J.:

OK, I wrote up both my battery-driven and generator-driven builds, with
photos. I've also included my personal analysis (based on energy
expenditure) as to when each is preferred:

http://home.comcast.net/~mandmlj/BikeLights/index.html

The wheight comparison is a little unfair: A Shimano XT front hub weights
150g. The SON delux is 385g.

Fair enough on the SON weight; I was using what info I could glean from
the web. I'll try to correct that tomorrow on my web page. I'm less
certain about your XT weight; whether a skewer is included is unclear on
many web listings. Does your SON weight figure include a skewer? I was
using the weight of my own front hub, an older Campy Chorus, per
weight-weenie listings on the web. I believe the 125g weight is w/o skewer.

5 AA-cells with a battery-holder weight 160-200g - which is nearly the
weight difference between the XT hub and the SON delux.

NiCD, NiMH or Li-ion? I'm using Li-ion, which have clear weight
advantages over the others. There are some dangers, too (readers may
remember MacBooks catching fire) so one should get battery packs with
built-in safety circuits (thermal and over current protection).

The figures you cite, ~15 Wh for 180 g is 83 Wh/kg. My battery pack
(again, a fairly generic Li-ion) is giving ~44 Wh for ~350g, or 125
Wh/kg; my figures are based on actual experience and measurements, for
what that's worth.

The 5 AA cells
deliver about 15-16 Wh in the best case. This is 4 hours powering a LED
headlight with 4 Watt - comparable to a SON delux powering an Edelux.

OK, so for 235g of Li-ion (assuming you could find a battery pack at
exactly that weight), that's (0.235 kg)(125 Wh/kg)/(4 W) gives 7.3 hours
of runtime, much less than the 16 I figured, but still a lot.

By the way, my generator drag figures are based on ~3W output and 75%
efficiency. With 4W output and actual efficiency, I expect the energy
input required will be higher than my web page says. What are the
measured efficiency figures for a SON for typical riding at 4W output?

With
a battery light it is essential to always have a set of spare batteries at
hand

We'll have to disagree on the "spare set of batteries" statement. I
will admit that when I started PBP over 20 years ago, spare bulbs were
required, and possibly spare batteries. Does anybody know if they still
do that?

- expecially if you want to use it close to their nominal capacity.

Running the battery until dead will shorten its life, yes, at least for
most chemistries. Modern Li-ion battery packs tend to have a cutoff
circuit to avoid this problem. I bench-test my light+battery
combinations when I buy/build them to make sure I know how many hours of
light they will give, and figure in a margin for battery deterioration.
Getting caught with a dead light in the middle of nowhere is not
something I'm willing to risk. I'm really getting the 44Wh I cite above.

With a spare set of AA-cells you have already to carry more wheight than
the hub dynamo solution.

There is no question I'm comparing the SON solution to a highly
optimized battery solution. I penciled out lots of battery-and-LED
combinations before building my battery-light for a 2008 flèche.

Over-the-counter solutions will vary. It seems many retail lights are
using less-than-optimally-efficient (read: a year or two behind the
cutting edge) LEDs, though that of course doesn't really bear on
battery-vs-generator.

For my purposes, I still believe the battery solution is better for
infrequent ultra rides, the generator for impromptu or frequent daily usage.


Andreas - working for the SON-manufacturer...

Thanks for your correction about SON weight.

Mark J.

On May 4, 7:29*pm, "Mark J." wrote:
Early this year I built a super-simple generator-driven LED light as a
test of concept. *No driver board or regulator of any kind is needed.
Details are below.

It works great, generating something like 200 lumens from an old
Soubitez 6V generator. *...

I'd be interested in beamshots.

I built a simpler one-LED generator headlight with a plain vanilla
rectifier feeding the LED. Since the LED was donated by a friend (VP
for research of a firm that makes high-output LED products) I don't
know its exact identity, but it's clear it puts out far, far more
lumens than a standard halogen bulb. Not 250 lumens, but I don't need
anywhere near that, provided it's properly focused.

But the problem, as usual, is optics. The "oval" LED optic unit I
bought for it isn't properly focused - it just doesn't have the
"throw" I need. It gives lots of semi-useless "fill" when used in
addition to my standard halogen headlamp, but it doesn't throw light
far enough down the road.

I wish someone would produce an LED lens that mimics the beam of a
standard generator headlamp, but perhaps wider - that is, a good
rectangular patch, at least one lane wide, with cutoff above the
horizon, and with highest intensity just below the horizon (i.e.
furthest down the road). As a bonus, some extra brightness to the
side and above the horizon would be nice. This would work both to
better light sharp turns in the dark, and to attract attention from
drivers waiting at side-street stop signs....

.... not that I've ever had a problem with that. My bog-standard
halogen lamps cause motorists to wait far longer for my night passage
than they do for my day passage. They frequently wait as long as 15
seconds until I pass, when they could have made their move with plenty
of safety before I was anywhere near.

- Frank Krygowski

Frank Krygowski wrote:
On May 4, 7:29 pm, "Mark J." wrote:
Early this year I built a super-simple generator-driven LED light as a
test of concept. No driver board or regulator of any kind is needed.
Details are below.

It works great, generating something like 200 lumens from an old
Soubitez 6V generator. ...

I'd be interested in beamshots.

I would too, but in short, I don't know how to do it meaningfully.

First, since the generator is front-wheel driven, and I don't own a set
of rollers currently, I can't easily run this in a stationary setting.
I think it's important to have a consistent backdrop when comparing
lights; it's what I see on all the various web beamshot compilations.
Secondly, the only cameras I have don't allow me both f-stop and
shutter-speed control, which really need to be consistent from one shot
to another for a comparison to be meaningful.

I'm really open to suggestions on this front; if it's only the /shape/
you were curious about, that's simple: it's symmetrically round, with
all the issues you raise below.

I built a simpler one-LED generator headlight with a plain vanilla
rectifier feeding the LED. Since the LED was donated by a friend (VP
for research of a firm that makes high-output LED products) I don't
know its exact identity, but it's clear it puts out far, far more
lumens than a standard halogen bulb. Not 250 lumens, but I don't need
anywhere near that, provided it's properly focused.

It seems the current state of the art is something like 100 lumens/watt
for white LEDs (but that was true two years ago, so I might be out of
date). Even though these are manufacturer's claims, they are likely
enough to be challenged if erroneous that I'll give 'em /some/ credence.

Figures I've seen on the web (of no particular authority) suggest
halogens run around 30 lumens/watt. I don't know if that's when run at
spec, or "overdriven" as bike lights are commonly designed to do; I've
also heard that gives a 10% efficiency advantage. I can't tell if these
figures are to be trusted.

But in short, yes, you would expect the LED to be far brighter, watt for
watt.

At 1/2 amp (well, 470 mA), my LEDs are drawing about 1.55 watts each.
If I believe the Seoul Semiconductor data sheets, 470 mA should give
about 127 lumens per LED, or 254 lumens total. That's 82 lumens/watt, a
bit disappointing numerically, but the practical result is very satisfying.

I'll repeat that I find the blue-white light far more effective in
cycling practice than the comparatively yellowish light halogens have
given me (and yes, I do remember when halogens seemed so much "whiter"
than regular bulbs).

But the problem, as usual, is optics. The "oval" LED optic unit I
bought for it isn't properly focused - it just doesn't have the
"throw" I need. It gives lots of semi-useless "fill" when used in
addition to my standard halogen headlamp, but it doesn't throw light
far enough down the road.

The combination of a 5 degree & a 15 degree lens I've used seem a
reasonable compromise in practice, at least in the 250 lumen range. I
haven't tried two 5 degree lenses on the road, but from backyard tests,
I thought I would want more spill than two 5 degree lenses provide.

I wish someone would produce an LED lens that mimics the beam of a
standard generator headlamp, but perhaps wider - that is, a good
rectangular patch, at least one lane wide, with cutoff above the
horizon, and with highest intensity just below the horizon (i.e.
furthest down the road).

Yep, that would be nice.

As a bonus, some extra brightness to the
side and above the horizon would be nice. This would work both to
better light sharp turns in the dark, and to attract attention from
drivers waiting at side-street stop signs....

... not that I've ever had a problem with that. My bog-standard
halogen lamps cause motorists to wait far longer for my night passage
than they do for my day passage. They frequently wait as long as 15
seconds until I pass, when they could have made their move with plenty
of safety before I was anywhere near.

Yeah, until we have the efficiency of better optics, we can just clobber
'em with pure power.

Mark J.

bob prohaska's usenet account wrote:
Mark J. wrote:
Here's the design:
* 4 Seoul Semiconductor P4 LEDs, U-bin from http://www.dealextreme.com
(in Hong Kong, about $5 each)
* Housing hand "carved" from 1" square aluminum tubing,
cheap at Home Depot. Many thanks to Allen Chapman's site
for his housing design: http://bikeled.org/
* Lenses and lens mounts from http://ledsupply.com,
two 5 degree, two 15's, about $4 per LED
* Proper heat sinking is important, I used Arctic Alumina to mount the
LED stars, $15 from coolerguys.com
You won't need the entire tube for the build, a small
consolation for the high price.

Wire two pairs of two LEDs each in series, then wire the two pairs
front-to-back as a rectifier bridge (ASCII art below). Attach to
generator and go ride!


At $5 per LED plus $4 per lens, it's almost a wash and much simpler
to use Luxeon Star-O one watt LEDs with integrated optics.
http://www.theledlight.com/starswith_optics.html

My first thought is that these are rated (per the datasheets on the
link) at only 350mA maximum, so they could get toasted with a 0.5 A
generator.

The setup illustrated at
www.zefox.net/~bob/bicycle/
requires no separate optics and heats insignificantly under the
hardest conditions this rider can maintain.

My second thought is that, ratings schmatings, it appears to be working
for you. With a four-LED rectifier bridge, each LED sees 0.5 A only 50%
of the time, and the phase cycling's pretty fast, so I guess it averages
out.

A nice quick-and-dirty construction, by the way. Being able to /screw/
your stars onto the aluminum backplane means any cheap thermal grease
will do, and the expensive thermal adhesive isn't needed (well, I
assume). Good call.

A later version using
six Star-O's in their unseparated "as shipped" arrangement works
admirably and makes a very neat package.

I'm guessing that later version might be rated for a higher amperage
also, just making it better.

Mark J.

Mark J. wrote:

Early this year I built a super-simple generator-driven LED light as a
test of concept. *No driver board or regulator of any kind is needed.

I've made a couple of generator lights with salvaged Luxeon III
emitters wired reverse-parallel like yours. I used the crudest
possible mount-- I bent brackets out of aluminum plate (like reflector
brackets with no holes for mounting reflectors) and glued bare
emitters to the faces of the brackets. I restrained and sealed the
wires with flowable silicone sealant. The lights are inconspicuous,
robust, and mostly out of harm's way.

Both systems use old Sanyo BB generators. One (mine) uses two white
Luxeons with a piece of 2-conductor wire to carry both legs of the
power circuit. The other one uses two white and two red Luxeons, with
a white and a red in series on each side. (White Luxeons can tolerate
reverse voltage, but reds can't, so the white ones protect the red
ones.) A 2-conductor wire connects the white LEDs in front to the red
ones in back, and the frame of the bike closes the circuit. .

The approximately 120 degree light pattern from bare emitters is not
that effective for seeing a long way down the road, but it's fantastic
for being seen. The broad flood beams make pools of light on the
street surface that greatly enhance the bikes' visibility. And the
lights come up to full brightness at just above walking speed.

The two red Luxeons on the rear of my friend's bike constitute the
most intense bicycle taillight I've ever seen.

Chalo

Andreas Oehler wrote:
Wed, 05 May 2010 16:54:35 -0700, Mark J.:

Andreas Oehler wrote:
Wed, 05 May 2010 15:32:13 -0700, Mark J.:

OK, I wrote up both my battery-driven and generator-driven builds, with
photos. I've also included my personal analysis (based on energy
expenditure) as to when each is preferred:

http://home.comcast.net/~mandmlj/BikeLights/index.html
The wheight comparison is a little unfair: A Shimano XT front hub weights
150g. The SON delux is 385g.
Fair enough on the SON weight; I was using what info I could glean from
the web. I'll try to correct that tomorrow on my web page. I'm less
certain about your XT weight; whether a skewer is included is unclear on
many web listings. Does your SON weight figure include a skewer?

Both are weighted without skewers.

I've now used the lighter weight SON delux rather than the SON 28 on my
web page comparison.
http://home.comcast.net/~mandmlj/BikeLights/index.html

I still get ~7 hours of light (at 4W) with no drag by carrying 245g of
batteries rather than 245g of extra weight in a dynohub.

5 AA-cells with a battery-holder weight 160-200g - which is nearly the
weight difference between the XT hub and the SON delux.
NiCD, NiMH or Li-ion?

AA-cells all weight nearly the same - if they are alkaline, NiCd or NiMH.
I just considered NiMH, because this is what most people use.

I'm using Li-ion, which have clear weight
advantages over the others.

True. But charging multi-cell li-Ion packs is not trivial. How much
utility cyclists have appropriate equipment?

I'm not really talking about utility cyclists. Remember, I just built a
generator-driven light for my commuting uses, featured on my web page.
That was where this thread began. The battery light is for event rides
(ultra, randonnees, etc).

As to charging Li-ion packs, I plug it into my smart charger and wait
for the light to turn green. That's taking advantage of a lot of safety
circuitry in the charger and battery pack, circuitry I wouldn't want to
be without, but it is commonly available today.


The figures you cite, ~15 Wh for 180 g is 83 Wh/kg. My battery pack
(again, a fairly generic Li-ion) is giving ~44 Wh for ~350g, or 125
Wh/kg; my figures are based on actual experience and measurements, for
what that's worth.

That's true - but no solution for the average end-user. Both battery
technologies also suffer from great losses in capacity when used at low
temperature. At below freezing usable capacity can be less than half the
rated values.

I've worked my way up to doing 300k in the rain (well, much of a 300k),
but I won't be doing any brevets below freezing any time soon. Again,
for utility cycling, I'm now using a generator myself. But even at 50%
capacity, I've still got 3.5 hours of free light for the weight of a
dynohub.


The 5 AA cells
deliver about 15-16 Wh in the best case. This is 4 hours powering a LED
headlight with 4 Watt - comparable to a SON delux powering an Edelux.
OK, so for 235g of Li-ion (assuming you could find a battery pack at
exactly that weight), that's (0.235 kg)(125 Wh/kg)/(4 W) gives 7.3 hours
of runtime, much less than the 16 I figured, but still a lot.

By the way, my generator drag figures are based on ~3W output and 75%
efficiency. With 4W output and actual efficiency, I expect the energy
input required will be higher than my web page says. What are the
measured efficiency figures for a SON for typical riding at 4W output?

The higher the output power - the higher the efficiency. 60-68% are
realistic values for medium speeds and 3-4 Watt output. But this includes
bearing and sealing losses which are also present at high quality hubs
(like the Shimano XT) without generator.

My web comparison now uses 70% efficiency for the SON. I forgot to
figure the drag losses of a regular hub comparing to the SON /switched
on/, but that only amounts to about 4% of the energy requirements of
driving the SON.

With
a battery light it is essential to always have a set of spare batteries at
hand
We'll have to disagree on the "spare set of batteries" statement. I
will admit that when I started PBP over 20 years ago, spare bulbs were
required, and possibly spare batteries. Does anybody know if they still
do that?

What will you do, if your estimate of battery runtime was wrong (maybe
because temperatures where lower than expected) or you have to ride longer
in the night because your time planning was wrong?

The answer is not to cut things too close. When I did my flèche, I had
enough battery to run the light continuously from sundown to sunup.
That's enough light. It only took 12oz. Add a margin if it's gonna be
arctic out there. I usually add very comfortable margins and then don't
worry about it.

But let's be honest and admit that things fail. Tires can fail so
dramatically that even the tire boot in my bag won't suffice. Cables
break. I can overestimate my ability to hammer at a high pace, and then
"blow up", or underestimate the amount of food I need. It happens, but
rarely. If one is careful with one's batteries, let-downs on their
account will likewise be rare.

But finally, again, for utility cycling, where I /have/ had battery
failures due to temperature, or just forgetting to charge the darn
thing, generators are a nice solution, and my web page says so.


- expecially if you want to use it close to their nominal capacity.

That would be a mistake - margins are needed.


Running the battery until dead will shorten its life, yes, at least for
most chemistries. Modern Li-ion battery packs tend to have a cutoff
circuit to avoid this problem. I bench-test my light+battery
combinations when I buy/build them to make sure I know how many hours of
light they will give, and figure in a margin for battery deterioration.
Getting caught with a dead light in the middle of nowhere is not
something I'm willing to risk.

You risk an accident at a fast downhill or will voluntary stop your ride,
when you find out the batteries are empty during the (unusual cold) night?

Modern lighting systems, even my home-brew bodge, have low-battery
warnings built in. Having one's light suddenly quit is an avoidable
mistake in most cases. I have had it happen when a connector suddenly
came loose (a problem independent of energy source), and it was
exciting! It was also avoidable.


I'm really getting the 44Wh I cite above.

With a spare set of AA-cells you have already to carry more wheight than
the hub dynamo solution.
There is no question I'm comparing the SON solution to a highly
optimized battery solution. I penciled out lots of battery-and-LED
combinations before building my battery-light for a 2008 flèche.

Over-the-counter solutions will vary. It seems many retail lights are
using less-than-optimally-efficient (read: a year or two behind the
cutting edge) LEDs, though that of course doesn't really bear on
battery-vs-generator.

For my purposes, I still believe the battery solution is better for
infrequent ultra rides, the generator for impromptu or frequent daily usage.

I don't deny the advantage of highly optimised battery solutions for
single "ultra rides", but I won't even count PBP in and don't see much
advantage in commuting or traveling.

Yes, for commuting a generator is appealing. See the bottom of my web page.

Mark J.