# Here are some awesome LEDs

Discussion in 'Electronic Design' started by Doug Goncz, Feb 4, 2004.

1. ### Doug GonczGuest

Hello!

From down under, with stamps all over it, comes a package from Oatley
Electronics containing four 1 watt white LEDs and four lenses. Each LED is
about the usual size but is heat sinked to a disc about the size of an American
nickel. Quite heavy.

Each is rated 3.6 VDC, and a whopping 350 milliamps. I have a 15 V regulator on
my bicycle. Let's see, that's 2 x 7.2 or 14.4 volts for the LEDs off of 15 from
the regulator.

What are the efficiency tradeoffs between a resistor and a constant current
source? Is it better to use a constant current incandescent lamp as a source in
this application, or a semiconductor solution? I suppose the semiconductor
would be more efficient but the lossy lamp gives some light which counts toward
efficiency.

I calculate a 43 ohm, 6 watt resistor. Did I get it right?

What's the most efficient setup for these LEDs in series or even parallel, with
the given source, 15 V, capable of 5A. (It also runs an inverter)? Do I need to
evaluate the performance of the regulator at 350 milliamps?

My intuition is that these 4W of LED lighting should be the equivalent of 16W
of incandescent, assuming they do about as well as fluorescents. With the
lenses, I should have plenty of light forward with a mere twitch of the pedals.

Is the right resistor calculation 0.6 V / 0.35 A = 1.7 ohms, 1/4 watt? That is
acceptable on efficiency.

Should I further heat sink the 4 LEDs side by side on a 1/8 or 1/16 inch thick
aluminum base about 6x1 inches, forming a chassis with more structure
supporting the lenses and enclosing the system?

My physics project at NVCC:
Google Groups, then "dgoncz" and some of:
ultracapacitor bicycle fluorescent flywheel inverter

2. ### Ian StirlingGuest

They don't.
They are about the same efficiancy as halogen lights.

3. ### Syd RumpoGuest

Another particular advantage for a bicycle is that the LEDs will still
show white at low power input when an incandescent lamp will be
glowing a very inefficient dull yellow or red.

Up to a point, a bicycle dynamo behaves more like a constant current
source than a constant voltage source, so you *may* be able to
dispense with any current limiting or regulation.

Your 15V and 5A sounds either wrong or very hard to pedal, unless
that's 15V open circuit and 5A short circuit.

A 3W dynamo - which many are - would probably allow you to use three
of these LEDs in series. I'd try it, slowly at first, and with an
ammeter in series. Substituting a suitable zener might be a good idea
for a first try if you're nervous, and 11V high power types are

4. ### Doug GonczGuest

Thanks, Ian and Syd.

I asked too many questions. I will share more of what I know.

A 48T cog on the rear wheel of my bicycle drives and 11T cog on a Kollmorgen
Servo Disc motor. Analysis in Mathcad shows that should be 96T, and I can make
that soon. Any way, this is the road-coupled generator/motor. It's potential is
proportional to road speed. It can output 275 mechanical watts at rated 41V,
8A. Decent efficiency.

What I found in the analysis that is just plain stunning is that on a one
hundred foot plus descent of Apex Circle near here, about a tenth of the
available potential energy is converted to kinetic energy. Much is lost to air
friction. But some nearly 1/4 is quickly pumped into my storage medium of
choice, ultracapacitors. A couple of watt hours. There is a longer hill
available.

The Ke of the motor is 12.1 V / 1000 rpm, so it can generate around 30 volts
with a 96 tooth gear on a descent. So there's plenty of power to illuminate
scary descents. I'm working on storing it for use to power lighting. It is of
use on the long climb back up Apex Circle. In this application, it is pretty
efficient, way up there. I have a post to sci.physics.research to illuminate
the difference between positive and negative traction power and current flow.
There is a four quadrant efficiency graph I can't get to. With this graph,
which must make sense of the two contradictory quadrant, I can get a mean
efficiency. Right now, I pick traces off the energy graph and compute the
potential energy of the descent and the energy picked up by the capacitors.

Now the other motor generator is planned only for continuous lighting. It has a
Ve of 37V/1000rpm, an 8 tooth pinion, and a 48 tooth driver on the cranks. So
at a leisurely pedaling speed of 60 rpm we have 60*48/8 = 60*6 then
360*37/1000=about 12 volts for the taillight, a bright JC Whitney red LED
marker light. The low power front light isn't figured out yet. I have a four D
cell, 16 white LED flashlight coming, and the four white high power LEDs with
their nice little assymetrical condensor lenses. I have a 6V converter from TI
sample service that's a fat TO-220 package and this will provide 6V, 1.5A. I'm
sure I can make the output 9W and another 3 for losses using the pedals. I need
a 12V regulator for the taillight.

The 15V, 5A converter is for powering a 150W sine wave inverter, one or more
dimmers, and two 20 W dimmable fluorescents. That's for the scary descents and
for making maximum use of the energy in the caps. The inverter will run from 15
to 10 V At 12.75 V it beeps. Does that explain the 15V 5A 75W plus losses
figure? It's not coming from the pedals. The idle consumption of the inverter
is 6W. If I go back to my "modified square wave" inverter and use diodes to dim
the fluorescent tubes, the idle will go down and the adjustment of light level
to surroundings will be less distracting.

I don't know the thermal time constant of my Servo Track armature but I know
the coupling. It's an ironless armature PM disc motor. Fits nicely mid frame.
But will it handle the downhill pulse? We'll have to put in a Freon and see if
it boils.

It really blew my mind getting ahold of the Runge-Kutta solver. I leaned a lot
about what can be expressed in differential form and what cannot. The variables
in my RK model are position along the track, speed, charge on the cap, and
voltage on the cap. I guess charge isn't needed as it does not appear in the
other three.

Altitude is an external function used in computing the force balance. It's the
"survey" of altitudes along the position line, by linear interpolation. Slope
refers to altitude. Rider power is constant assuming perfect gearing. Wind drag
is computed iteratively over a few simulations The torque at the motor provides
the other element of force. All these sum, then divide by mass to produce
acceleration, the first derivate of speed, which itself is the first derivative
of position, thereby defining position, that is, progress along the course.

The literature that came with the four LEDs does say heat sinking extends their
life. The right resistor really is 15V-14.4V = 0.6V, divided by 0.35 ma max is
1.7 ohms min., about 1/4W added to four watts plus is an acceptable efficiency.
We'll just have to see how hot the heat sink gets, and how quickly. I'm
putting an ammeter port on the chassis to monitor current. What I'd like is a
1/4W resistor with a heat sink interface so I can measure the total heat up of
the heat sink, including heat from the resistor. A milled slot between the two
center LEDs might provide good coupling with heat sink paste. I need some 1 1/2
inch OD clear plastic tube for my D cells so the spacing between LEDs can be
just more than that. The tubing segments will support the lenses at a fixed
distance from the LEDs and allow light not lensed to escape to some use as a
side marker for oncoming traffic.

A 3W dynamo provides 6V. I don't think that power level is more than a marker
light on a bike, not suitable for navigation. I don't think they are safe.

There's a new, small, brushless/senseless Sanyo hub motor coming out that would
get the motor mount out of my frame and let me build a cleaner system. It's
rated 200W. They don't say if that's mechanical output or battery consumption.

In some of the simulations, with a precharge as if ten cycles had been
completed, the equipped bicycle wins out by hundreds of feet over a few laps.
We'll see what the actual bike is capable of.

I'm looking for two things in a bicycle headlight: output and an adjustable
beam pattern to adapt to changing conditions.

I've got a four watt six volt fluorescent ready for the front (remember my post
about the Avon Skin Spec UV light conversions? They didn't work out. It has to
be button started. You can't just have it come on line whenever power is
available. Too bad. Nice reflector.) The little 4W light has esentially no
refector. It's a first gear, difficult terrain light.

..>From: Syd Rumpo
Yes, I'd to try like a non-dropout regulator.

I'm building a parallel/series relay PCB to allow energy capture at low speeds,
too, such as just pedaling on flat terrain in first gear, when the rider has
capacity to spare.
Yes, but this is by choking the output of an AC dynamo with its own inductance.
I can do better.

My physics project at NVCC:
Google Groups, then "dgoncz" and some of:
ultracapacitor bicycle fluorescent flywheel inverter

5. ### N. ThorntonGuest

Hi

I dont know quite what your 15v supply is, but the odds are its not
that brilliantly stabilised. Running the LEDs with a 0.6v drop R will
give you wild current variations and probably kill the LEDs.

Better to series parallel the LEDs to 7.2v and use a series dropper
for them. A bulb would be best, 8v 700mA, or in the real world 6v
700mA plus a 2v R drop. Use 2x 2v drops, one in each series chain for
good current sharing.

But with a bulb driving the LEDs you have no advantage: when the bulb

Best way is a switched mode constant currant source, but much more
complex. Just use a halogen light.

Regards, NT

6. ### Chris HodgesGuest

If these are the ones I think they are then the efficiency is better
than halognes (though nowhere near as good as fluorescents), and much
better than "ordinary" incandescents. For a torch (bike lamp) type
application you may well end up with noticeably more light for the
same battery consumption than halogens. Not a factor of 4 though.

Chris

7. ### Syd RumpoGuest

On 05 Feb 2004 13:27:43 GMT, ( Doug Goncz )
wrote:

[snipped]
Strewth. Nurse!

8. ### Ian StirlingGuest

Not even a factor of 1.5.
Krypton torch bulbs are surprisingly good lumen-wise.
They produce a lot of light by trading off life for light.

Halogen bulbs are around 15lm/W, the white luxeon star LEDs are at
the most 23lm/W. (IIRC) and down to around 13lm/W.

9. ### Jim YanikGuest

( Doug Goncz ) wrote in

How much did you pay for them?

10. ### Jim YanikGuest

I checked their site myself,and their price is \$14 AUS.and they're
currently out of stock.

BTW,I believe 350ma is a MAX rating(not to exceed),their nominal op level
is 300ma.