Feeler: Driver Circuit for max LED Power from Hub Dynamo

[Martin]

In the past I've been experimenting with LED driver circuits that draw the maximum power from a hub dynamo and feed this into some LEDs.
Here's a power vs speed graph that compares different configurations (3, 4 and 6 LEDs) with my latest driver circuit:

The driver circuit that can do this has been presented here and it's now confirmed that this same circuit can handle 3..6 LED systems as well as hub and bottle dynamos. Just the component values have to be adapted.
Now before I start work on a PCB, I ask around if there are people who want one and if it should be made in a certain way.
It will be an unpopulated PCB, I will not go into having modules manufactured.

So far I have discussed the project with ktronik and we said that it would have to fit into a steerer tube. So the PCB would be 1" wide but rather long, rectangular, not round.
The circuit requires 4 to 5 larger capacitors (in case of a hub dynamo, they can be bigger than the rest of the circuit) and these will not be put on the PCB but connected by wires. This way the PCB can remain small and the capacitors can be located where there is space. The PCB is estimated to take an area of around 16 sq cm (2.5 sq inch) without the external capacitors.

I wonder if the project should be done in traditional through-hole technology or rather SMT. While SMT can reduce the size and avoids the need to drill holes in the PCB, the components are more difficult to find and to handle for the hobbyist. Any opinion on this ?

The circuit does not contain any stand light / battery / supercap / charger features. It neither contains a tail light driver / controller (but no problem connecting a readily made one across the main LEDs). Who considers these features essential ?

I intend to make the wire connection to the PCB by large solder pads. I consider this smaller and less prone to corrosion than any clamp or connector. What do you think ?

An overvoltage crowbar would have to be added to prevent damage to circuit and LEDs in case that the LEDs disconnect. I also plan on adding a strain relief for the connecting cables. These things will add length to the PCB but are needed when the driver circuit is not located inside the lamp cabinet. Maybe I should put them near the end of the PC so they can be cut away ?

Let me hear what you think.

Martin

[Calina]

Hi Martin,

Would you have by any chance a power graph for 2 LEDs?
What caps would be required for 2 LEDs and a bottle dynamo?
As you point out SMT is not easy to work with for hobbyist, I would prefer through-hole technology.
Also I think a stand light is an absolute must.

[Calina]

Your title is not very informative.

I'm sure you can find something but maybe the word "Feeler" at the beginning would bring in more responses.

[Ritar Hylon]

Hello Martin!

I have been a CPF lurker for a few years, but have decided to make my first post in response to your amazing work.

For the past few years, I have tried to live a (largely) car-free life in the California Car Kingdom. My ride-of-choice initially featured a generator driven Lumotec, but the combination of low-output and no stand light caused me switch to a DIY overvolted lithium-ion powered MR-16 solution. This seemed to provide adequate light (~400 lm) but had obvious drawbacks (run-time limitations, fragility, beam pattern issues, etc.)

Based on your work (and others in this forum) it is time to reconnect my dynamo. The above circuit seems to supply ample power for an LED-based light, but ideally, I would like to keep some kind of stand-light capability. Here are some of my specific requests:

• Coupling KTRONIC's Dyno-Batt system with this autoswitching circuit would be ideal in that it would give a stand-light (assuming I can get some big caps).
• A tail light capability, would be of less importance (I can get an entire season's worth of taillight illumination from a set of primary cells).
• A through-the-hole board is a must, if I am to put it together (for a pre-assembled board, I would prefer SMT.)
• Solder pads would be desirable, for my needs.

I am eager to see what others think and what ultimately comes from this project. Thanks for the great posts!

RH

[Martin]

It's on my website, called Double1 and Double2.

Hi Martin,

Would you have by any chance a power graph for 2 LEDs?
What caps would be required for 2 LEDs and a bottle dynamo?
As you point out SMT is not easy to work with for hobbyist, I would prefer through-hole technology.
Also I think a stand light is an absolute must.

[Martin]

..

• Coupling KTRONIC's Dyno-Batt system with this autoswitching circuit would be ideal in that it would give a stand-light (assuming I can get some big caps).
• A tail light capability, would be of less importance (I can get an entire season's worth of taillight illumination from a set of primary cells).
• A through-the-hole board is a must, if I am to put it together (for a pre-assembled board, I would prefer SMT.)
• Solder pads would be desirable, for my needs.

RH

Welcome to CPF Ritar and proud to meet a California-based commuter. If you do 80% of your trips by bike, your savings on gas beat the very latest engine technology by miles.
And thanks a lot for your feedback !

For standlight, there are a number of unanswered questions: How many minutes of light at what minimum power will be asked from the standlight ?
What reservoir should be used ? SuperCaps are nice, but real expensive for a big one. LiIon would be much more affordable if both longer runtime and more power is requested at the same time. For a to-be-seen light, no issue though.
Then, I ask myself what minimum power I should leave to the headlight while charging the reservoir. 1 W OK ? Or maybe use a switch to let the user decide if he wants to put full power into the reservoir or into the headlight or 50-50 or 70-30% or 30-70% ? I try to avoid switches if can, or we will soon be 100% busy with our bicycle dashboards during a ride..

If a dynamo tail light is desired, one could connect a readily made one (here's my favorite for this purpose) across the LED string. However, you said it, a battery-powered tail light can last really long so that it is often preferable not to run 1.5m of cable to the tail of the bike. Now that I can get battery tail lights that turn on automatically if it's dark and the bike is moving, there's not much incentive in wiring a taillight to the dynamo system.

[ktronik]

I will have some time to build up a 6 LED version of the basic circuit for a 'confirm' test. I will drop the caps down to 100uf in fullwave & see how much power we get...

I have a long weekend... so into the workshop I go...

K

[ktronik]

Hi Martin,

Would you have by any chance a power graph for 2 LEDs?
What caps would be required for 2 LEDs and a bottle dynamo?

Try a 47uf & 100uf.

[n4zou]

Hello Martin!

I have been a CPF lurker for a few years, but have decided to make my first post in response to your amazing work.

For the past few years, I have tried to live a (largely) car-free life in the California Car Kingdom. My ride-of-choice initially featured a generator driven Lumotec, but the combination of low-output and no stand light caused me switch to a DIY overvolted lithium-ion powered MR-16 solution. This seemed to provide adequate light (~400 lm) but had obvious drawbacks (run-time limitations, fragility, beam pattern issues, etc.)

Based on your work (and others in this forum) it is time to reconnect my dynamo. The above circuit seems to supply ample power for an LED-based light, but ideally, I would like to keep some kind of stand-light capability. Here are some of my specific requests:

• Coupling KTRONIC's Dyno-Batt system with this autoswitching circuit would be ideal in that it would give a stand-light (assuming I can get some big caps).
• A tail light capability, would be of less importance (I can get an entire season's worth of taillight illumination from a set of primary cells).
• A through-the-hole board is a must, if I am to put it together (for a pre-assembled board, I would prefer SMT.)
• Solder pads would be desirable, for my needs.

I am eager to see what others think and what ultimately comes from this project. Thanks for the great posts!

RH

I ride a considerable number of miles on an MUP at night as a Trail Watch volunteer. My job is to help people with mechanical trouble stuck on the trail after dark and look for nefarious types looking to cause trouble. Here is the circuit I am currently using.

It's not as powerful or bright as Martin's circuit but it provides light at all times for both the head and taillights. I have over 200 nighttime patrol miles and have found I have a tendency to overcharge my battery pack. I will be adding monitoring circuits soon and start disengaging the dynamo from the tire more than I've been doing and add a few more LEDs' as clearance lights to decrease the 120mA charging current feeding the batteries when the dynamo is engaged and I am running over 10 MPH.
I am still in the design and test stage as can be seen in the photo below.

This example circuit could be used to automatically disconnect the batteries if your type of riding could discharge the battery pack due to slow and/or stop and go riding in city riding.

The driver chip would simply disconnect the batteries from the LED when batteries are depleted and your stopped. As soon as your dynamo starts providing 1.2 volts the LED driver would automatically turn on the LED headlight once again. The chip shown is only an example; any commercially produced LED driver circuit for 1-watt LED's would do the same thing.

[Martin]

Let me hear your preference for the standlight:
What would be the minimum power and runtime for it ?
How fast and when would its reservoir be charged ?
Are user controls needed and what params are to be controlled ?

[znomit]

Many times I have found that an independent front light is a good idea (especially if Im the one doing the wiring!!!). With this system a front blinky would supplement it better than a standlight.

[bbaker22]

Martin-
I'd be interested. Definitely not smt for my shaky hands. Solder pads sound fine for the wire connections. Standlight would be nice, but not essential.

baker

[Ritar Hylon]

Let me hear your preference for the standlight:
What would be the minimum power and runtime for it ?
How fast and when would its reservoir be charged ?
Are user controls needed and what params are to be controlled ?

• I would want a minimum of 100lm from the light which would be about 1 Watt from a Q5.
• The longer, the better on runtimes, of course, but I suppose I would want an absolute minimum of 5-10 minutes.
• Maybe 1/10 of the available power could be sent to the reservoir once speed exceeds say 15kph.
• An on/off switch would be a minimum control for me.
• If one could add "fancy" tricks like multiple output levels, that would be nice-to-have (steal a controller board from a Fenix?)

[rideatnight]

Martin, I was trying to figure out a way to get a decent standlight by using a Yellow LED, so the runtime would be several minutes. Does this circuit make any sense?



The idea was to use the transistor as a switch, that would allow current to flow to the supercap and the yellow LED while the dynamo is running. When current isn't flowing through the crees, the transistor closes off the circuit and the yellow led runs off of the supercap (with the voltage being the voltage drop across the cree).

http://picasaweb.google.com/cphill2/DIYLights

I've already built my shimano dynamo powered dual cree light, using your medium powered circuit (with 470uf caps in series, 4700uF cap in parallel). With the Q5 bin, it's incredibly bright, enough that I almost want to dim it sometimes. I added a 7.5V zener diode in parallel with the big cap. At speeds above 40 kph (25mph) the hub provides a noticable notchy drag. Is this just back EMF, and has anybody else experienced this?

[Martin]

Martin, I was trying to figure out a way to get a decent standlight by using a Yellow LED, so the runtime would be several minutes. Does this circuit make any sense?


The idea was to use the transistor as a switch, that would allow current to flow to the supercap and the yellow LED while the dynamo is running. When current isn't flowing through the crees, the transistor closes off the circuit and the yellow led runs off of the supercap (with the voltage being the voltage drop across the cree).

http://picasaweb.google.com/cphill2/DIYLights

I've already built my shimano dynamo powered dual cree light, using your medium powered circuit (with 470uf caps in series, 4700uF cap in parallel). With the Q5 bin, it's incredibly bright, enough that I almost want to dim it sometimes. I added a 7.5V zener diode in parallel with the big cap. At speeds above 40 kph (25mph) the hub provides a noticable notchy drag. Is this just back EMF, and has anybody else experienced this?

Welcome to CPF, rideatnight.
Sorry, your circuit doesn't make sense. As the LEDs are powered from the dynamo, the transistor's base is negative vs the emitter so the supercap will never ever charge. Give it another try, then test it !
If you want a nicely optimized solution, simply take a commercial dynamo tail light with standlight function and look how they do it. They employ between 2 and 4 bipolar small signal transistors.

When your hub is part of a wheel and being used on a real bike, you shouldn't get that notchy drag. If no loose spokes, no lose axle and no shaky fork legs, the mass absorbs the notches. If you put the hub into a lace or stand drill though, the notches are very obvious and can be really loud at a specific speed and load condition.
I assume you put the Zener to protect the LEDs in case they disconnect. It needs to be a big Zener, capable of handling 5 W. If you put the Zener in order to limit the brightness, that's not a good idea.

[John Phillips]

Hi Martin,

What caps would be required for 2 LEDs and a bottle dynamo?

May I take the liberty to offer an answer to part of the question? I have been playing with this exact set-up. For an 8-pole bottle dynamo driving two LEDs in series I suggest the following capacitances for the new full-wave doubler circuit:

C1, C3 = 100 uF
C2, C4 = 33 uF

2200 uF (2.2 mF [millifarads]) for C5, the main filter capacitor, is plenty for a bottle dynamo with its higher ripple frequency. You shouldn't notice any flicker at all.

These values will make effective series capacitances of 50 uF for the voltage doubler mode and 200 uF for the normal bridge rectifier mode. This approximates the capacitor values in the Manual Switching between Voltage Doubler and Bridge Rectifier circuit at Martin's Web page http://www.pilom.com/BicycleElectron...moCircuits.htm

I have found these values seem to work well with two LEDs and an 8-pole bottle dynamo, giving a smooth, non-peaky power graph.

Cheers,
John.

[Martin]

John, the values you suggest are critical.
Because this matter doesn't exactly fit this thread, I have put the answer in the thread that mentioned the manual switching circuit before.

For the stand light feature, I intend the following:
Charge a string of supercaps to the voltage of the LED string or below and limiting the charge current. User can set the limit by component value. The charge action may be limited to bridge rectifier mode only.
Power the LED string from the supercaps using a step-up circuit.
The step-up is LED-current controlled, with the current set to a level less than what is driven by the dynamo so that the step-up only engages if the LED string receives very little current from the dynamo. The current is user defined by component value.
The step-up will discharge the supercap to near zero after the power from the dynamo ceases. If LiIon is desired instead of the supercaps, the step-up would have to look a little different as it needs an undervoltage cutout.
A manual switch is optional, it could do this: Standlight on, Standlight off, Max power standlight burst

As you ride this circuit, the following happens: When starting the ride, the supercap slowly charges from zero and the light is on at the same time. When you then slow down to a rather low speed or stop completely, the step-up pumps energy into the LEDs so that the light level doesn't drop below a certain minimum. When you go slow or stop for a longer time, the charge on the supercap depletes and you have less light / no light.

I'd like to hear some opinions on the preferred standlight reservoir. Supercaps or LiIon ?

Unless everyone shouts that this standlight thing is absolutely needed, I will do the first version of this PCB without.

[dom]

Hi Martin
Can you make the standlight blinking -maybe use less power and have smaller caps?
I find i need it at some intersections -and the small blinky i have runs out of batteries too quickly (button cells)
I wouldn't mind one with either the Supercaps or LiIon

Cheers
Dom

[thomas.hood]

I'd like to hear some opinions on the preferred standlight reservoir. Supercaps or LiIon ?

Can you suggest advantages/ disadvantages?

Aren't LiIons liable to explode if mistreated?
Which would lose least energy to internal resistance on charge/ discharge?

Unless everyone shouts that this standlight thing is absolutely needed, I will do the first version of this PCB without.

I would consider a standlight essential: I don't see how you can turn across traffic at a junction without one. This is a time I feel very vulnerable and would like the brightest light possible. I think 3 minutes would be ample.

My aim would be to have a 3 Cree string at front, 4 red at back (to give a similar Vf for each string), powered by a SON.

I want an uber-'be seen' light, with both sets blinking, so these front/back pair could either blink simultaneously or simply alternate. I would imagine the near continuous Vf load of the second scenario would be better from an efficiency point of view - is this the case?

I'll be commuting to and from work in the dark soon, so keep up the amazing work!

Thanks,

Tom

[PhxCycler]

Hi Martin,

I'm new to the forum and have been reading with interest the development of the LED driver for a generator.
I own a SON hub, but am a roadie. I do ride 24+ hour events quite frequently and am always looking for ways to improve my lighting. I've ordered all the components to build up one of those bad boys and give it a try. Anyway, to your questions.

• The circuit board size sounds good to me, but on a road bike the steerer tube is not available for any housing.
• I'm definitely interested in one of the boards.
• Through hole technology would be my preference.
• I have no need nor desire for a tail light control nor a stand light. I always have a cheapo Cateye EL500 LED light for backup and use that for a stand light when rarely needed.
• I do like your idea of having some of the board available to be cut away for "optional" circuitry.

Many thanks for your efforts and your willingness to share all this with the rest of us!

-Mike

[rideatnight]

I put the Zener in there because I was under the impression that I could overvolt the LEDs at high speeds (40+ kmph). The notchy feeling I get is definately related to the electronics, because it goes away when I turn off the light. I'm tempted to try to remove the Zener, and see if that isn't the cause, but I'm worried about burning out my LEDs when going over 50 Kmph. It's only a 1W 7.5V zener, figuring that it was last ditch protection for the crees keeping the voltage drop across each limited to 3.75V. If the current stays around 500ma, then the LED voltages should never reach 3.75V.

[Calina]

For a stand light I'd prefer super caps to batteries.
Contrary to batteries, they will charge well in sub zero temperature. There is no worries to overcharge them or kill them by running them too low which is always a problem with accumulators and they will live much longer in any weather.

I guess a current of about 100 (125) milli amps would be enough and would provide about 35 to 45 lm of light with a Q5. Three minutes of light seems reasonable (5 even better) but I haven't calculated what cap size would be needed for that.

In city traffic with constant stop and go, if too much current is required, it is possible that the stand light would be useless in only a few seconds. On country roads it is less of a necessity but still nice to have as a safety feature. Blinking? I don't know; sometimes it is more annoying than useful.

As for the rear light, I use a blinky on 2 AAAs, I fell safe and they last at least a whole season.

[ktronik]

.[LIST][*]The circuit board size sounds good to me, but on a road bike the steerer tube is not available for any housing.

-Mike

Can you tell me why??

do you have a 1" stem or a NON-ahead type headset??

K

[PhxCycler]

Can you tell me why??

do you have a 1" stem or a NON-ahead type headset??

K

Hi Ktronik,

I have a 1 1/8 in. threadless steerer on 3 different road bikes. In every case the tube is used as the clamping mechanism for the stem. There is a set bolt inside the steerer into which the locking cap screws. None of this is removable. I also have a mtn bke and the steerer is wide open as you describe. I've never seen a road bike with an open steer tube.

That said, the size of the board really doesn't matter to me. I think the size specified is good. That way it will fit inside a steerer when available. I'll just make the eletronics fit somewhere (like a small seat bag attached to the stem ).

-Mike

[thomas.hood]

I commute year-in-year-out by bike, and as evidenced by my posting posting here take an interest in lights people use. As it's been lights-on commuting for the past few weeks I've been taking an interest in what people in London use.

In the garage of the building in which I work there are ~220 bikes (big company...) of which there are precisely 2 dynamo powered bikes. Owned by a Swiss and a Dutch guy respectively.

The only other place I have seen dynamos are are SON hubs which are exclusively the preserve of the Audax bunch: accordingly at least 80% of which are steel bikes... in other words 1" steerers.

Any other UK riders concur?

Tom

(I must declare an interest having 3 bikes in my family that would benefit from this set-up, all with 1" steerers)

[thomas.hood]

Hi Ktronik,
removable. I also have a mtn bke and the steerer is wide open as you describe. I've never seen a road bike with an open steer tube.
-Mike

I've never seen a mtn bike with the steerer-tube closed. We are talking about the bottom end, right?

Almost all road bikes with plastic-carbon-fibre forks are closed at the bottom, but many have drain hole which would be sufficient to get wires into (assuming the top-cap anchor is removable).

1-1/8" Plastic-carbon-fibre forks have thick steerer tubes -- what is the ID of one? Is it so different to a 1"

I've seen very few steel forks that are closed at the bottom.


Tom

Addendum:
I've just measured the ID of a 1"-steel steertube and it's marginally _larger_ than the ID of the 1-1/8" steer tube on my MTB! (~22.4mm)

[Martin]

I put the Zener in there because I was under the impression that I could overvolt the LEDs at high speeds (40+ kmph).

LEDs are never "overvolted" in forward direction, their critical parameter is the current and from a dynamo this would stay around 500mA with a conservative tuning capacitor or no tuning capacitor. If you think the LEDs need protection, add a fuse in series with the circuit.

The notchy feeling I get is definately related to the electronics, because it goes away when I turn off the light. I'm tempted to try to remove the Zener, and see if that isn't the cause, but I'm worried about burning out my LEDs when going over 50 Kmph. It's only a 1W 7.5V zener, figuring that it was last ditch protection for the crees keeping the voltage drop across each limited to 3.75V. If the current stays around 500ma, then the LED voltages should never reach 3.75V.

Shunt the tuning capacitor with a wire and see if the notchyness is reduced.

[rideatnight]

Thanks, I'll give that a shot. I appreciate having someone to bounce this off of.

[ktronik]

Hi Mike,

We have the same system...so you got no probs...

How it done like this...

1: replace the 'star nut' (yes removable) with a 'freestyle' star nut in the same size... this star nut is ment for a cable to pass through the fork to the front brake... plug the bottom & you are sealed tight...

if you run a carbon fork then best to keep the carbon type star nut...won't work on carbon systems, so if you runnig a carbon fork you won't want to heavy up the bike anyway...

so all good...

K

[PhxCycler]

Hi Mike,
We have the same system...so you got no probs...
How it done like this...

1: replace the 'star nut' (yes removable) with a 'freestyle' star nut in the same size... this star nut is ment for a cable to pass through the fork to the front brake... plug the bottom & you are sealed tight...
K

K,

Not quite the same systems. My star nuts are not removable. They are fixed inside the tube. Doesn't matter tho, I'll put the electronics elsewhere.

-Mike