DIY Dyno-hub LED tail light designs?

[sfCyclotourist]

I've got an old tail light housing that I'm thinking of modding with a Luxeon LED I've got lying around. I've already built a 3x Cree LED headlight running off my SON28 hub and a Schottky rectifier. Any insight on tail light designs, and how to best incorporate a tail light into my existing head light? Should I give the tail light its own rectifier, and just wire it up in parallel to the head light? Or should I wire it up directly to the headlamp, utilizing the existing rectifier?

Thanks,
-Jim G

 

[pe2er]

Here is how I made my (battery powered) Luxeon taillight:


I Use the Dx 1W LED Spotlight as a donor for the driver and LED.

Both of your rectifier solutions would work. Choose the one that is most convenient to you. Make sure that you do not go over the maximum current for your rectifier if you decide to use the same one for your front ant tail light.

Myself, I would just add a separate rectifier for the taillight at the location of the driver (the driver I use already has one built in ). That makes your taillight independent of the design of your front light. In case of a malfunctioning rectifier, you would then only loose one light.

 

[unterhausen]

seems to me that you are not going to be happy unless it's in series somehow. Parallel means you have to balance the impedance so it doesn't steal too much power from the headlight.

 

[znomit]

I've been mucking around with a few blinkies recently. The LD1100 remembers its mode when the battery is disconnected, so when the power comes on again it automatically starts going again. They run nicely off 2.4V and very very very nicely off 3.3V.

You could parallel drive off a single white led or charge a supercap(see the standlight circuit thread) to 3.5V from a string of leds and you'll get a little runtime when stopped.

Seem like perfect dyno taillights. You get blinking modes too.

Will need to test whether it still turns on after its been off long term.

 

[yellow]

in all my familys bikes, I added parallel wired front and tail lights.
2 times with an Lux V and one triple Cree, all in parallel with such one here:

the taillight runs with its ~80 mA, what else?
When wired in series with the front light, the tail has no other chance than get the same 600 mA of power
(which would be too much at the tail + wont light the front one at walking speeds)


working for years now
(and looking very "normal")

 

[sfCyclotourist]

in all my familys bikes, I added parallel wired front and tail lights.
2 times with an Lux V and one triple Cree, all in parallel with such one here:

the taillight runs with its ~80 mA, what else?
When wired in series with the front light, the tail has no other chance than get the same 600 mA of power
(which would be too much at the tail + wont light the front one at walking speeds)


working for years now
(and looking very "normal")

Thanks for that. Your design seems somewhat similar to Martin's Circuit #4, except that you're using a discreet bridge rectifier and also a voltage regulator (LP2951?) to decrease the current vs. a simple resistor. Martin's design also has the taillight wired in parallel. He notes that "the tail light is damaged should the headlight disconnect" -- I don't understand why that is -- wouldn't the resistor (R1) prevent an excess of current from reaching the LEDs in his design? Also, if one uses a Luxeon or Cree high-power LED that can handle 500-600mA (along with a separate rectifier), it should be OK, right?

He also mentions another disadvantage: "the lower brightness of the tail light" -- but I think he meant to write "the lower brightness of the head light", since the parallel tail light steals some of the current from the head light circuit...?

Thanks!
-Jim G

 

[yellow]

I can only type for my light, dunno what "Martins circutis" are
(but imho the very basic designs are around for quite some time ...)


For me, I 1st look if the dyamos have some kind of Voltage restriction (usually Z-diodes):
without any load, the AC-meter hooked to the dynamo, speed it up.
When it goes up to ~7.5 volts and stays there, the Z-diodes are present. And if the voltage goes up at higher value, there is nothing (something we want here).

With the Lux V it makes no real deal, that runs with the 7.5 V. With the triple-cree the Z-Diodes have to be killed.

Now one has a front light that runs with 7.5 V, or with some 11 Volts.
Hook the wires for the tail light in parallel to the dynamo and the 5 V Voltage regulator ensures the red led (2 parallel strings of 2 in series) get their voltage, no matter what Voltage the front system is running on.
Sure the regulator burns excess voltage into heat, but with that low current no real deal.

Also the led version runs on under 100 mA, while any normal tail light steals much more from the front,
AND - when the tail were in series with the front light not to "steal" power from it - it were brutally too bright AND the light would need the bike to go relatively quick to light up.


So, I have a very "easy to make" and small solution of front and tail, both hooked to the 2 exits of the dynamos in parallel. No hassling with more wires from tail to back.




PS: my system (the triple-cree) works with walking speed already, with a very simple and small MOSFET-Rectifier solution (coming from juergen.h at www.metb-news.de)

there surely are much more sophisticated "voltage doubler" or "switching" circuits.
I also remember someone in here measured a few that were better than the MOSFET.
From all I tested the Mosfet gave the most light at low speed. This and its small physical size is why I choose this one.

 

[yellow]

didn't notice that You linked martins Circuit ...
there is quite some difference between #4 and my solution.

#4 is a "single" circuit which has front and tail together, my idea leaves more options.

At first only a SINGLE led at front is a joke!
Any dynamo can power up to 3 powerleds, so why use a single one?

And "my" tail version with the regulator runs on the voltage of the front system, no matter how many led it uses.
Should the front light fail, should have no effect on the tail light
(in martins version the tail - wired antiparallel - would get the full power of the dyno and that would kill the led)



To get concrete:

this is my front system:

abut a triple CREE, not SSC as in parts list

the pics still show the original "switching" circuit, that did not work good and was removed to what is shown in 1st pic.
1 led floating free (necessary because of the Mosfet), 1 ElKo for the remaining led and one led with that Gold Cap (for a very basic "Stop light". Runs for 1/2 a minute or so)

here is the MOSFET and LuxV in "quick and easy" original looking bike light

(did this for family and friends because I had some LuxV and it is very quick to do and looks "original")


both the triple and the LuxV are hooked to the 2 dynamo contacts,
and all the bikes have the rear circuit mounted into the big red reflector and this is als hooked to the 2 dynamo contacts, parallel to the front wires.



PS: I also add a resistor after the regulator to get current down, if needed. Depends on the individual led.
The 4 led w. 30-50 mA each are more than bright enough

 

[Steve K]

I've got an old tail light housing that I'm thinking of modding with a Luxeon LED I've got lying around. I've already built a 3x Cree LED headlight running off my SON28 hub and a Schottky rectifier. Any insight on tail light designs, and how to best incorporate a tail light into my existing head light? Should I give the tail light its own rectifier, and just wire it up in parallel to the head light? Or should I wire it up directly to the headlamp, utilizing the existing rectifier?

Thanks,
-Jim G

hi Jim,

I think you've seen my design before. My preference is a taillight wired in series with the headlight. I use ten 5mm red leds in parallel, with a plain diode in parallel, but reversed anode to cathode. The leds carry the positive half of the AC current, and the plain diode carries the negative half.

Steve K.

 

[yellow]

:thinking:
with another led instead of that plain diode,
wouldnt these give light also?

So You are waisting one half of the AC with that diode?

Maybe I have a wrong pic in my head (?)

 

[Steve K]

:thinking:
with another led instead of that plain diode,
wouldnt these give light also?

So You are waisting one half of the AC with that diode?

Maybe I have a wrong pic in my head (?)

I think you have a good understanding of what I described. To get the most light, you would want to use two sets of ten 5mm leds. One set would conduct for the positive half of the AC, and the other set would conduct for the negative half of the AC.

I use ten leds, with a schottky diode in reverse parallel. This gives me very low voltage drop on half of the AC cycle, which important to me because my hub dynamo turns slowly as I pedal up the 14% grade on my commute home each night.
Keeping the taillight voltage low helps me keep current flowing through the led headlight, since the dynamo doesn't produce much voltage at low speeds.

Ten 5mm leds do produce quite a bit of light compared to most battery powered taillights, so I never felt a need to use 20 leds. It is a huge improvement over the B&M taillight that I used before... it only had one 5mm led. That might be okay in Germany where people are in the habit of seeing bicycles, but it didn't seem to be sufficient here in the middle of the USA.

Steve K.

 

[sfCyclotourist]

I think you have a good understanding of what I described. To get the most light, you would want to use two sets of ten 5mm leds. One set would conduct for the positive half of the AC, and the other set would conduct for the negative half of the AC.

I use ten leds, with a schottky diode in reverse parallel. This gives me very low voltage drop on half of the AC cycle, which important to me because my hub dynamo turns slowly as I pedal up the 14% grade on my commute home each night.
Keeping the taillight voltage low helps me keep current flowing through the led headlight, since the dynamo doesn't produce much voltage at low speeds.

Ten 5mm leds do produce quite a bit of light compared to most battery powered taillights, so I never felt a need to use 20 leds. It is a huge improvement over the B&M taillight that I used before... it only had one 5mm led. That might be okay in Germany where people are in the habit of seeing bicycles, but it didn't seem to be sufficient here in the middle of the USA.

Steve K.

Steve, why not connect your tail light to your headlight after its bridge rectifier? That way, the tail light won't flicker, and you don't need the reversed diode...?

I'm trying to figure out a design that doesn't flicker, since brevet rules require a non-flashing tail light.

-Jim G

 

[yellow]

the "low speed problem" is, why I wire my front and tail in parallel.

With the rear light in series, one adds something like "another" white led, and this - with the Mosfet + 3 Cree light - makes a difference of "no light at walking speed" to "enough light to get out of the bike storage room in total dark"

 

[syc]

I think the reason he doesn't put it in series after the rectifier is because it would introduce a voltage drop on the entire AC cycle, instead of only half of it, and he's concerned that the lights would drop out entirely on slow climbs.

I think for your purposes, you might try modifying the lm317 standlight circuit in the giant standlight thread, and instead of feeding the lm317's output to a supercap, send it to your taillight led. You'll get the rectified output, but it will be in parallel and only skim off a small amount of the current.

This circuit:

But replace the goldcap with your red led, and cutout the "discharge rate resistor" entirely so that you don't bleed off any power into the lower LED on the right side.

Hopefully one of the EE type folks will tell us if that will work or not...

Steve, why not connect your tail light to your headlight after its bridge rectifier? That way, the tail light won't flicker, and you don't need the reversed diode...?

I'm trying to figure out a design that doesn't flicker, since brevet rules require a non-flashing tail light.

-Jim G

 

[Steve K]

Steve, why not connect your tail light to your headlight after its bridge rectifier? That way, the tail light won't flicker, and you don't need the reversed diode...?

I'm trying to figure out a design that doesn't flicker, since brevet rules require a non-flashing tail light.

-Jim G

the quick answer is that the headlight and taillight were designed to be interchangeable with conventional lighting.

the longer answer is that this would make the headlight use four wires, and I just don't want to have that many wires going in and out. Plus, the taillight is its own rectifier... there's no gain from using the headlight's rectifier.

The flicker is an issue that is separate from where rectification takes place. Flicker only occurs over a small range of low speeds, and not at speeds where you'll spend 99% of your time. Doesn't seem like it ought to be considered a flashing light. My assumption is that they want to avoid annoying other riders with steady flashing lights.

A cap could be added to reduce flicker, but that requires a diode to isolate the cap and taillight from the AC waveform. My taillight does include a AA nicad and boost converter to drive four 5mm leds when the dynamo's AC voltage is gone. Maybe that would avoid the flickering light concern?

Steve K.

 

[csaadaam]

the pics still show the original "switching" circuit, that did not work good and was removed to what is shown in 1st pic.
1 led floating free (necessary because of the Mosfet), 1 ElKo for the remaining led and one led with that Gold Cap (for a very basic "Stop light". Runs for 1/2 a minute or so)

Hi all!
Sorry for bumping an old thread, but i'm thinking about building a hubdynamo-powered light system on my bike, just like this one.
I have a few questions though. Shouldn't be there more heatsink for this light? about 2W of power should generate some heat, i think.

I was thinking of using only 2 LEDs, XR-E R2 types. One with the optic like this one, but one with a diffused optics, for lighting the road just in front of me.
I was thinking that a 25mm wide heatsink, of 60mm length should be good. The 2 stars would be glued on one end, slightly bent in between (5-10°), and only the part that is outside the lamp housing would have 4-5 blades.
This way the far (5-10m) part would have a concentrated light, and the near (<5m) part would be lit more diffused.
I'm using my bike mostly in the city, so I'm not going for hundreds of meters of sight

 

[jdp298]

I use Two 1W LEDs up front in old dynamo housings. Heatsink not required. So long as you don't back the LEDs directly onto something easily damaged, like a capacitor, open space around it ought to be enough. Kind of depends on how powerful the LED is I suppose.

I've what you'd call a solution for your problem. I run from Kingston through Richmond Park to Whitehall some days a week (not all). What I've got seems to solve all the problems. The only time I need illumination over visibility to others is Richmond Park, where the cars can't go when it's dark. It's also where I've even had a couple of complaints about dazzling lights. From 1W LEDs at that! I took it as a complement then angled the lights down a little more.

The circuit I use 3 times over is pretty simple. Bridge rectifier, zener to clamp the voltage for the 5.5v supercaps, then a 3.3v LDO driving the LED with the right resistor. At an averagely brisk walking pace it flickers, but that goes away by 6mph. And at the rear there's a B&M standlight on the rack and a superbright 3mm red LED in the old mudguard light housing, which is an excellent diffuser.

Their positioning matches that of cars and scooters, so to be seen, the lights are vertically where people expect them. Helmet lights are all good, but I often need a second look to spot them, as I imagine drivers do. The B&M is going to get replaced by a scooter/motorbike rear housing with a half-watt red LED. the housing ought to diffuse it enough to give me both good spectral contrast (brightness), but also angular resolution (wide patch of light). This is second thing I notice; there are some seriously bright lights out there, but the best ones, to me, are easy to spot by their size of illumination.

For interest's sake, most TVs run at 800x600 pixels, or similar. Computer screens much higher, generally, because you are closer. The angular resolution of the eye is finite, so if your light only fills 1 pixel (or equivalent in the eye), then you really aren't helping yourself like you could.

 

[steveo_mcg]

Hi all!
Sorry for bumping an old thread, but i'm thinking about building a hubdynamo-powered light system on my bike, just like this one.
I have a few questions though. Shouldn't be there more heatsink for this light? about 2W of power should generate some heat, i think.

I was thinking of using only 2 LEDs, XR-E R2 types. One with the optic like this one, but one with a diffused optics, for lighting the road just in front of me.
I was thinking that a 25mm wide heatsink, of 60mm length should be good. The 2 stars would be glued on one end, slightly bent in between (5-10°), and only the part that is outside the lamp housing would have 4-5 blades.
This way the far (5-10m) part would have a concentrated light, and the near (<5m) part would be lit more diffused.
I'm using my bike mostly in the city, so I'm not going for hundreds of meters of sight

The little bit of alu is probably enough its only two watts of heat probably enough to lower its life but not by much. I can't remember what the formula to represent area to heat dispersal, but on a moving bike you don't need as much as you would in a flash light.

 

[steveo_mcg]

What do you run on the back? I've had comments on my 350ma basically bare red cree as being "some light" it was a friend of mine on a 'bent so she was perhaps a little low and getting it square in the eyes...

My question is would 500ma on the back be too bright?

 

[BrianMc]

Is 500 mA into an XP-E red or read-orange, too bright? Depends on what you do with the light. In a narrow focussed beam? I would guess so, and it would not help much outside that cone of blindness. Spread out for good side visibility and visibility from truck cabs? You are talking about 70 lumens, twice a Turbo, 1/3 the level of a Red Zone 4 on max, a bit more than a Red Zone 4 on its max steady output. Excellent at night spread out. (I run the RedZones on high steady as leg markers.) They can be sen in the day but they don't stand out in sunlight. I also run two 100 lumen wide angle (near hemisphere) Luxeon based lights and they show in the day just fine.

For the errand bike (ready to roll, no batteries), I am awaiting 3 such LEDs which I plan to have in a 1S3P array in series with a 3S1P XP-G headlight driven with Martin's circuit. One Led straight back, one each to each rear quarter. Using Flare lenses 15 x 90 FWHM. Should be about 100 lumens in a 270 degree by 20 degree part doughnut of light. Like Homer took a bite. That should be close enough to a Red Zone 4 at max 200 lumen setting into a hemispherical output. My aim was to come close to my twin luxeon battery lights.

Should prove interesting, I think.

BrianMc