Dynamo lighting



TimC wrote:
> http://michaelcarden.net/blog/index.php?p=76
>


Nice.

Are you/he going to power the LEDs directly or run them via a current
regulator? Also, if you run them directly, why can't you use the LEDs
back to back and skip the Shottky diodes, is there a problem with
reverse biasing LEDs?

Friday
 
On 2007-06-23, Friday (aka Bruce)
was almost, but not quite, entirely unlike tea:
> TimC wrote:
>> http://michaelcarden.net/blog/index.php?p=76
>>

>
> Nice.
>
> Are you/he going to power the LEDs directly or run them via a current
> regulator? Also, if you run them directly, why can't you use the LEDs
> back to back and skip the Shottky diodes, is there a problem with
> reverse biasing LEDs?


It sounded like he'd continue the article when he actually finished
them.

Normal LEDs break down at -5V or so -- they act as a zener diode and
pass the full current available to them when reverse biased, but I
have no idea about white CREE LEDs. In the case of an unrectified
alternator system, as long as no current is being drawn, the voltage
becomes very high. So initially, when voltage is reversing, no
current flows. Then when the voltage increases in response to the
lack of current flow and as the generating cycle goes on, it will get
above -5V, and the LED will start conducting (and not emitting useful
light as a result of this) as a reverse biased diode junction.

I can't tell what will happen then with such a sinewave current supply
-- I guess you'll end with a forward biased device with a 3V drop
across it (3V typically for a white LED), and the necessary current
flow to give 3 watts (since that what these generators supply). So
your 3 watt LED will supply 3 watts of light. But for the other half
of the cycle, as the diode is reverse biased you'll be sinking the
same current, but the voltage across the diode will now be the 5V or
so that the diode drops when reverse biased. No useful light out of
this, excess energy usage, and much excess heat generation in both the
source and the sink.

You'll probably end up with something referred to in the industry as a
semidestructor, a Dark Emmitting Diode (DED), a friode or a reversible
diode. And then you'll have to gather up all the magic smoke it lost,
and stuff it back into the DED before it will become a LED again.

-- Ti "magic/more magic" mC.
 
Friday wrote:

> Are you/he going to power the LEDs directly or run them via a current
> regulator? Also, if you run them directly, why can't you use the LEDs
> back to back and skip the Shottky diodes, is there a problem with
> reverse biasing LEDs?


It depends. "normal" LEDS behave like 5v zeners in reverse, so if you're
not going to go above about 5 volts, you should be OK.
I've just looked up the spec sheets of a couple of these XLamps, and the
stated reverse voltage is 5v max.
I'm guessing that since you can get away without any regulation, it's safe
to assume that the voltage isn't going to be anything above 3.something or so,
so yes, it appears it should be OK.

--
Linux Registered User # 302622
<http://counter.li.org>
 
In article <[email protected]>,
[email protected] says...
> TimC wrote:
> > http://michaelcarden.net/blog/index.php?p=76
> >

>
> Nice.
>
> Are you/he going to power the LEDs directly or run them via a current
> regulator? Also, if you run them directly, why can't you use the LEDs
> back to back and skip the Shottky diodes, is there a problem with
> reverse biasing LEDs?
>
> Friday
>


You'd only get half the light.
The dynamo will provide the constant current, but only one LED will
light at a time.
Using the Bridge, with LEDs in series, you'll get the same current
through both LED's and thus be twice as bright.

Are the schottkies really needed?
I figured it's a current source, the voltage will sort itself out. Seems
to work good enough for me.

I have a CREE and an Luxeon in series, and the CREE leaves the Luxeon
for dead in light output. Must get around to putting in the other CREE I
have!
Oh yeah, running off a SON dynamo. Let there be light!

Ray
 
On 2007-06-23, John Tserkezis (aka Bruce)
was almost, but not quite, entirely unlike tea:
> Friday wrote:
>
>> Are you/he going to power the LEDs directly or run them via a current
>> regulator? Also, if you run them directly, why can't you use the LEDs
>> back to back and skip the Shottky diodes, is there a problem with
>> reverse biasing LEDs?

>
> It depends. "normal" LEDS behave like 5v zeners in reverse, so if you're
> not going to go above about 5 volts, you should be OK.
> I've just looked up the spec sheets of a couple of these XLamps, and the
> stated reverse voltage is 5v max.
> I'm guessing that since you can get away without any regulation, it's safe
> to assume that the voltage isn't going to be anything above 3.something or so,
> so yes, it appears it should be OK.


They don't necessarily need regulation because the hub can only put
out an amp or so, short circuited. That's of the same order that a
CREE LED wants, forward biased. Reverse biased though, the current
may be limited to the same, but the voltage may be higher, and hence
the power dissipation may be catastrophically higher. All for no more
light, and needing the hub to generate more power, stealing it from
your legs. Shottkys are easier. Of course, if you really care about
power loss, one of those funky "ideal diode" MOSFET circuits --
"synchronous rectification".

--
TimC
I don't want Perl to be beautiful -- I want you to write
beautiful programs in Perl. --Larry Wall
 
TimC wrote:

> They don't necessarily need regulation because the hub can only put
> out an amp or so, short circuited. That's of the same order that a
> CREE LED wants, forward biased. Reverse biased though, the current
> may be limited to the same, but the voltage may be higher, and hence
> the power dissipation may be catastrophically higher. All for no more
> light, and needing the hub to generate more power, stealing it from
> your legs. Shottkys are easier. Of course, if you really care about
> power loss, one of those funky "ideal diode" MOSFET circuits --
> "synchronous rectification".


Right, but the OP mentioned "some" leds. That is, more than one. If you
have two in parallel, (in opposite polarity) the forward biased LED will keep
the voltage to about 3.something volts, well below the 5v max reverse for the
other LED that's in reverse for that part of the cycle.

--
Linux Registered User # 302622
<http://counter.li.org>
 
On 2007-06-23, John Tserkezis (aka Bruce)
was almost, but not quite, entirely unlike tea:
> TimC wrote:
>
>> They don't necessarily need regulation because the hub can only put
>> out an amp or so, short circuited. That's of the same order that a
>> CREE LED wants, forward biased. Reverse biased though, the current
>> may be limited to the same, but the voltage may be higher, and hence
>> the power dissipation may be catastrophically higher. All for no more
>> light, and needing the hub to generate more power, stealing it from
>> your legs. Shottkys are easier. Of course, if you really care about
>> power loss, one of those funky "ideal diode" MOSFET circuits --
>> "synchronous rectification".

>
> Right, but the OP mentioned "some" leds. That is, more than one. If you
> have two in parallel, (in opposite polarity) the forward biased LED will keep
> the voltage to about 3.something volts, well below the 5v max reverse for the
> other LED that's in reverse for that part of the cycle.


I missed that at the time. Turning a very expensive CREE into a
diode instead of using a $0.20 part!

--
TimC
"Perhaps the truth is less interesting than the facts?"
-- Amy Weiss, RIAA's Senior Vice President of Propaganda^WCommunications
 
John Tserkezis wrote:
> TimC wrote:
>
>> They don't necessarily need regulation because the hub can only put
>> out an amp or so, short circuited. That's of the same order that a
>> CREE LED wants, forward biased. Reverse biased though, the current
>> may be limited to the same, but the voltage may be higher, and hence
>> the power dissipation may be catastrophically higher. All for no more
>> light, and needing the hub to generate more power, stealing it from
>> your legs. Shottkys are easier. Of course, if you really care about
>> power loss, one of those funky "ideal diode" MOSFET circuits --
>> "synchronous rectification".

>
> Right, but the OP mentioned "some" leds. That is, more than one. If
> you have two in parallel, (in opposite polarity) the forward biased LED
> will keep the voltage to about 3.something volts, well below the 5v max
> reverse for the other LED that's in reverse for that part of the cycle.
>


That's what I had in mind. The light output would still be the same but
coming from two LEDs instead of one, sharing the load if you like. They
ARE diodes so why not make use of that. Seems to be a waste of power
running it through the Shottkys.
 
Friday wrote:

> That's what I had in mind. The light output would still be the same but
> coming from two LEDs instead of one, sharing the load if you like. They
> ARE diodes so why not make use of that. Seems to be a waste of power
> running it through the Shottkys.


It's hard to say, you have to do your sums. It's a question of economics.
With two LEDS, you get the brightness of one, evenly distributed across two
LEDS. (with double the cost of the lighting elements).

Shottkys I'm guessing are significantly cheaper than one LED, so, with a
small hit on brightness, you get away with one LED.

The question is, is the hit on brightness worth the cost saved?

--
Linux Registered User # 302622
<http://counter.li.org>
 
On 2007-06-23, John Tserkezis (aka Bruce)
was almost, but not quite, entirely unlike tea:
> Friday wrote:
>
>> That's what I had in mind. The light output would still be the same but
>> coming from two LEDs instead of one, sharing the load if you like. They
>> ARE diodes so why not make use of that. Seems to be a waste of power
>> running it through the Shottkys.

>
> It's hard to say, you have to do your sums. It's a question of economics.
> With two LEDS, you get the brightness of one, evenly distributed across two
> LEDS. (with double the cost of the lighting elements).
>
> Shottkys I'm guessing are significantly cheaper than one LED, so, with a
> small hit on brightness, you get away with one LED.
>
> The question is, is the hit on brightness worth the cost saved?


But you don't get a hit on brightness, because there is the same
current flowing through both shottky and LED if they are in series.
Voltage across the both of them will go up by 0.2V or so, because of
the new drop across the shottky, but who cares? You will not feel the
tiny increase in resistance of the hub.

--
TimC
"Legacy (adj): an uncomplimentary computer-industry epithet that
means 'it works'." -- Anthony DeBoer in ASR
 
If I were knocking one up, I'd probably use bridge-connected power MOSFETs rather than diodes. That would reduce the forward drop to stuff-all, so the light comes on at lower speed.

A pair of CREE or Luxeon-IIIs in series would probably be a very good match for a hub dynamo.
 
In article <[email protected]>,
[email protected] says...
>
> If I were knocking one up, I'd probably use bridge-connected power
> MOSFETs rather than diodes. That would reduce the forward drop to
> stuff-all, so the light comes on at lower speed.
>
> A pair of CREE or Luxeon-IIIs in series would probably be a very good
> match for a hub dynamo.
>
>

I've been using a SON hub driving two Luxeon-III in series for almost
two years now.
I have a conventional bridge rectifier, no fancy synchronous rectifiers
or schottky diodes.

Anything above walking pace has the lights flickering, ie walking with
your bike.
At 10km/h you have usable light, and you have easily reached full drive
by about 20km/h.

Keep It Simple Simon

I have only just started to migrate to CREE's, I replaced one Luxeon a
while back, and the difference is astounding.

The biggest problem really is lack of good cutoff optics, you really do
feel for oncoming traffic!

To increase complexity, I reckon a PICAXE could be used to intelligently
used to short one LED out when stopped / low speed, and have that driven
by a battery that is across the two LED's at high speed.

I personally think chasing the nth degree of rectifier efficiency is
wasted effort with a dynamo, may as well eat another biscuit, as it
really makes 2/3rds of stuff all difference.

Ray
 
Ray wrote:
> Keep It Simple Simon


I thought I pretty much did that, with the system I knocked up for my
commuter bike last year:

http://www.fieldlines.com/story/2006/6/20/83754/2658

Essentially it's a small box containing 6xAAA NiMh batteries, a
regulator/rectifier to charge them, and a 555-based circuit to provide
flashing pulses for my front and rear "conspicuity" lights. The
headlight is just a cheapie $10 krypton-bulbed unit from K-Mart. The
biggest bonuses for me over the past year have been that I haven't had
to remove and charge any batteries, and all 3 lights are switched off
by flicking 1 toggle switch.

It's satisfactory for my needs, anyway. ;-)


BTH
 
"TimC" <[email protected]> wrote in message
news:[email protected]...
> On 2007-06-23, John Tserkezis (aka Bruce)
>> Right, but the OP mentioned "some" leds. That is, more than one. If you
>> have two in parallel, (in opposite polarity) the forward biased LED will keep
>> the voltage to about 3.something volts, well below the 5v max reverse for the
>> other LED that's in reverse for that part of the cycle.

>
> I missed that at the time. Turning a very expensive CREE into a
> diode instead of using a $0.20 part!


Sometimes you use the more expensive solution to get greater efficiency, increased output, or more
light for less (pedalling) effort.

Simply connecting the LED to the dynamo may not be the most efficient (assuming that is in fact a
suitable configuration). Will it get almost all the possible energy out of the AC cycle of the
dynamo, or can you do better by feeding it into a switchmode regulator which uses the parts of the
cycle that are below 3V? A clever circuit may duplicate some components and pick up both the +ve
and -ve cycles with very little loss.

It sounds like an impedance matching problem, where to get the most energy out of a source, you want
the load impedance to match the source impedance (this assumes a purely resistive load). I have
read (and Suzy probably has a spice model of Shimano Hub dynamo to prove it :) that a hub dynamo is
essentially a constant current device, and that you can hook up two 6V incandescent globes in series
to the dynamo. If that is the case, then you potentially have 6W available to you, not 3W.

Maybe we want a 3 phase hub dynamo, similar to a car alternator, which makes the filtering problem
much easier after rectification. Probably overkill.
 
In aus.bicycle on Sun, 24 Jun 2007 10:55:54 GMT
Russell Lang <[email protected]> wrote:
> read (and Suzy probably has a spice model of Shimano Hub dynamo to prove it :) that a hub dynamo is
> essentially a constant current device, and that you can hook up two 6V incandescent globes in series
> to the dynamo. If that is the case, then you potentially have 6W available to you, not 3W.


Yes you can. My Lumotec has 3 sets of connection points - one to the
dynamo, one to the rear light, and one to a 2nd front light.

I haven't bothered with a 2nd light, the Lumotec does a good enough
job for me, but some folk use a Lumotec and an E6 or two E6s.

Zebee
 
"Ray" wrote:
> Anything above walking pace has the lights flickering, ie walking with
> your bike.
> At 10km/h you have usable light, and you have easily reached full drive
> by about 20km/h.
>
> Keep It Simple Simon


Yes, I'd agree with that. I have no understanding of electronics - most of
the chat in this thread is like a foreign language to me. But that
description of your light sounds awfully like my B&M D'Lumotec Oval
http://www.bumm.de/index-e.html

I don't know why you'd go about making up a complex light when something
this good is about. With a hub dynamo you'd have enough output to run two of
these. I think they use a 3W Luxeon LED - not as bright as the Cree I know,
but damn good for any town and country road use.

--
Cheers
Peter

~~~ ~ _@
~~ ~ _- \,
~~ (*)/ (*)
 
Someone has done some designs for powering LEDs from dynamos.
http://www.pilom.com/BicycleElectronics/DynamoCircuits.htm

I didn't realise you that such simple circuits would work due to the current limiting of the dynamo.
I've got a nominally 3W LED light module from Jaycar that is no longer available, but it works with
12V AC input. I've also got a 20 year old Sanyo Dynapower roller dynamo. Both were used on my old
lights, but not together.
http://www.ghostgum.com.au/misc/BikeLight.htm

The dynamo likes to put out about 0.5A, and the voltages of about 4V open circuit at low speeds,
rising to about 18V open circuit at higher speeds. So I just tried connecting the dynamo directly
to the LED module. At low speeds it glows readily, but flickers because the frequency is too low
for the filtering capacitor inside the module, or to put it another way the capacitor is too small
for the low frequency from the dynamo. As you speed up, the flickering gets faster, until you reach
a critical speed when the flickering stops and you get a steady light. It then ran at 0.55A AC and
7.2V AC, or about 4W (assuming in phase sinusoid waveforms). So the module could "fixed" by
replacing increasing the capacitor size, but since the module is sealed that isn't going to happen.
If you speed up even faster, you get to a point where the LED module won't take any more current,
and the voltage suddenly jumps from 7.3V up to about 20V. I didn't look at the waveforms but I
suspect that the current waveform ceases to be sinusoidal.

This has giving me an idea. Try the dynamo full wave rectifier circuit mentioned on the link above,
labelled "Manual Switching between Voltage Doubler and Bridge Rectifier", connected to 3 XR-E LEDs
and see what happens. At low speeds this isn't going to work too well, so a simple alternative
might be to short out 2 of the LEDs if the voltage is too low. Hooking up a super cap to feed one
of the LEDs may allow it to keep running dimly for tens of seconds while stationary at an
intersection. The current limiting of the dynamo may keep the LED current within specs.

If it all works well, then replacing the Sanyo dyanmo with a Shimano hub dynamo may be worthwhile.