Bicycle dynamo circuit

I dunno, but you're supposed to be riding the bike the same direction as the traffic (i.e. on the right-hand side in the USA), so headlights shouldn't be a problem in the first place.

Good Luck! Rich

The input rectifier is overcomplex; if your capacitor C3 is big enough, D1 can be left open, D2 replaced by a wire, short C2 and make C1 as big as C3. Then, it's a voltage-doubler rectifier, only ONE diode conducts at a time (half the forward drop lossses). If you ever want to use a battery, the C3 can be replaced with a lead-acid 6V, gel type.

The second terminal of the dynamo is usually hard-wired to the frame ground anyhow.

It's not clear why one wants the photosensor, a switch works for me.

The usual problem with bike lights is the long downhill that burns up the lamp; it's not clear that this circuit has any protection for the LEDs against overcurrent. But, I'm usually using a battery, so maybe your generator is enough of a limit.

Have you considered oncoming cars in the adjacent lane?

C'mon, Rich, I know you're smarter than that.

John

Amazing how complicated things can get. If my math is right, 6V 12 ohms is 3 watts.

3.25V x .7A is 2.275W x 4 = 9.1W You sure you can get from here to there? The cost/reward ratio of the photodetector is too high. It's gonna be more grief than benefit. Also think the light output from the supercaps will be too dim or too short (depending on your choice of resistor) to be useful. You don't just stop at an intersection. You slow gradually and by the time you're stopped the caps are discharged...I think the 1K is WAY too big to give useful light and your time constant will be much shorter than you planned.

If it were me, I'd use a 18650 lithium cell, they come with built-in protection if you're so inclined. Use a microcontroller to skip cycles of the AC to keep the lithium cell at a constant, safe voltage. Make lighting decisions based on the frequency of the AC and set delays for the various modes. When you ride, the light comes on. When you stop, after some time the light goes off. Don't charge at low speed if that's your intention. Many fewer parts, near constant intensity, infinite flexibility.

And if you don't already know how to program a microcontroller, you learn a marketable skill in the process.

Headlights are supposed to be aimed such that they don't blind the drivers in the opposite lane.

Thanks, Rich

Are C1 and C2 of any value?

When you activate low speed mode, a surge from C3 may burn out D7.

Forget the photodetector. Use a switch.

Check the internal resistance on the ultracaps. A rating of 5.5V 1.5F sounds like low current CMOS backup caps. You need mobile power supply ultracaps.

Place an LED power supply module where the 1K trickle resistor is, then ditch the whole low voltage bypass circuit. The power supply modules are tiny, current regulated, and can reach 90% efficiency at low voltage. When current through the LED string gets low, the LED power module will activate to keep D7 on until your caps are down to about 4V. Make sure you wire it up correctly so the module's current sense works with the string feeding D7. RECOM Power and LuxDrive modules are commonly sold online.

--
I will not see posts from Google or e-mails from Yahoo because I must
filter them as spam

Who said anything about riding on the road?

Can't it just?

People have been able to get 12W out of a hub dynamo. 3W is just a nominal figure based on using incandescent bulbs; 2.4W at the front and

0.6W at the back.

Yes, I'm beginning to come to that conclusion, especially as I realised last night that it has to work at both 3v and 13v.

Real life experience in other places (candlepowerforum) suggests it will work, but I may have resistor value wrong.

What happens when I stop for a long time, do I need another switch to prevent the cell from discharging all the way?

Quite tempted to do this anyway.

Thanks - very helpful

Nick

Oh yes, they can double the maximum current output from the dynamo. See

for deatils

Yes, I can see that could happen, especially if I accidentally switched from Hi to Lo at speed. Got any suggestions as to how I could prevent this happening?

Seems to be the consensus.

Ah, they're going to be physically bigger aren't they?

The current through the LED string almost never gets low. It's the odd nature of the dynamo that it produces approx 500mA almost as soon as the wheel turns but the available voltage rises with speed. Going up a steep hill my speed will slow to the point where there isn't enough voltage to light more than a single LEDs (about 6mph) (it's hilly here). Very interesting idea to put LED driver in place of resistor. I really like the idea of having an auto swithcing circuit based on speed/voltage. The problem is the LED string won't light at all until I'm going fast enough to power the whole string and charge the caps.

Tiny is relative, this is my mark 1 light.

Just room for 2 LEDs, a regulator, smoothing cap and switch,

Thanks very much for taking the time to think about this.

The problem as I understand it, with a voltage doubler is that it halves the current and sinc my dynamo produces about 500mA (without the boost caps) I can't afford to half that as the LED doesn't produce much light before it gets 350mA. Once I get up to speed the voltage doubler works and people have build circuits that switch from bridge rectifier to voltage doubler at speed.

Not on mine,

Everybody seems to think the same!

Yes, as far as I'm aware nobody has managed to get more than about 1 amp out of one of the dynamos and the LED is ok up to 1.5A

Thanks

I dunno - I guess I just ass-u-me-d that that was what you meant by mentioning "headlights."

Thanks, Rich

Oh yeah, fair point!

Did it occur to you that if you can see the oncoming lights that it means light is coming into your eyes? Sure, I know that most of the light is aimed down their own lane. But not all of it goes there.

You're welcome, John

Memory backup supercaps ain't gonna be much good. 4.5F of real low esr supercaps is gonna be huge and expensive. yes?

Possibilities are endless. I'd have it dim some seconds after you stop, then turn off some time later. Modern processors can sleep on half a microamp and wake up on the first rev of the generator. Same reason there's no hardware on/off swtich on your calculator, tv remote, etc.

For me, the biggest obstacle to the project would be the worry that someone would steal it.

I'm a microchip guy. Startup effort can be daunting. Not hard, just complicated getting all the hardware and software together and working.

I picked up one of these at a TI seminar

Retails for $20 for processor, programmer, development system/compiler etc. Looks like a relatively painless alternative. This thread has motivated me to take it out of the drawer and see what it can do.

I'm a big fan of the minimalist approach. I have zero experience with hub generators, but depending on the number of poles/frequencies involved, there's an old motorcycle trick that might apply. Put the leds in parallel, but two of them backwards so they run alternatively on opposite half-cycles of the ac input. Or maybe two in series parallel with the other two backwards, depending of generator output characteristics. Put an inductor in series. At low speed, the inductor looks like a short. As the speed increases and the frequencies go up, the inductor impedance goes up and limits the current. Depending on the characteristics of the generator output this can greatly reduce the change in intensity with speed. But the devil is in the details. You may find that the inductor required weighs 5 pounds. Do the math. The generator leakage inductance may already provide what you need because you need the same function for the incandescent.

Use a 20mA led with a battery for the low intensity (please don't run over me) light when you stop. You can use some small NiMh and charge 'em at low rate from the generator if you like. Just keep the current low and replace 'em every few years when the abuse takes its toll.

Yes, I have often considered making the effort to get into PIC stuff but have always been daunted by the steep learning curve.

snip

The problem with hub dynamos in bicycles is that they don't spin very fast. This is particularly true in my case because I've bought the latest, lightest hub which is really designed for 20" wheels and I'm using it in a 29" wheel. It only works at all because LEDs are so much more efficient than incandescents.

Unless you go down the rectified, smoothed voltage route, the flicker is unbearable even though the losses are greater this way.

My hub is inherently limited to somewhere between 500mA and 1A.

Rather than do the classic bridge -> cap -> load, instead try bridge -> switcher -> load, perhaps with battery standby for minimal lighting when you stop?

I've been making various LED drivers with the NCP3063 chip (much better than the 33063/34063 chips it supercedes) recently and it works down to low voltage, as well as up to 40V. On the output side I've run parallel strings of LEDs with current mirrors (>200 LEDs with external switch MOSFET) and even a voltage multiplier to run a single string of 50 LEDs with >100V.

There's a lot you can do with simple electronics before you go play with microcontrollers. Which is also a place I play too, but not yet with the LEDs. Yes, steep learning curve. I have the PIC chips as well as some TI 430 whatsits that cost only USD4.30 each on an intro offer at the start of the year to play with.

Switchers can extract the maximum available power, transforming the load to suit the dynamo, within large limits.

Grant.

Good suggestion, Grant.

The LED traffic signals I worked on work similarly. The input is a bridge feeding a flyback switcher whose secondary gets rectified and applied to a series/parallel string of LEDs. Feedback is LED current converted to volts for the switcher. Proper tailoring of the feedback (low pass) gives a power factor correction.

This was done with a current-mode switcher chip rather than a true power factor controller.

Cheers, John