Another 'designing dynamo lights' thread: front 1 LED + rear 2 LED in parallel

[OndrejP_SK]

Hi all,

I've read many threads about dynamo light circuits design, I already have an idea about the various possibilities, and here I'd like to put it all together and ask some questions that are still unclear to me.

I would like to design both front and rear lights.
They should be connected parallel (so that each could be replaced with a commercial light if I feel like it).

The front light:

I want to use the chinese XC-997 as base for this construction. I have a dynamo version, but it lacks standlight which I would like to add when designing my own circuit.
I want to reuse the housing, the reflector and the board with LED which is XM-L at the moment.

using single LED in the front light would clamp the voltage of the whole system to something like Vf of LED (2.9V) + Vf of rectifier bridge (0.8V) and I coule only use ca 400-450mA for the LED. However I read about a design with buck converter ( http://www.candlepowerforums.com/vb/showthread.php?334214-dynamo-light-with-single-LED and http://minisystem.blogspot.sk/2012/01/dynamo-digital-buck-converter-progress.html ) which should in principle allow increase of current flowing through the LED when the dynamo voltage is more than the 3.7V of the front light as seen by the dynamo while at the same time increase the impedance of the front light so I wouldn't clamp the voltage to 3.7V.

Is this doable with a simple integrated chip switching buck converter like LM2575/LM2576 or is all the fancy 'load matching' and PIC necessary ?

Another option is to add a couple of diodes in series with the LED to increase the Vf of the light and to rise the voltage usable for 5.5V/1F supercap.

Now that I mentioned the supercap - I am considering using the supercap as source for 555 timer to make the standlight a blinky becuase I think that when standing in the middle of a junction at traffic lights a blinky would make me more visible. But I could also use a simple current limiting resistor to drive the LED from the supercap.

I would also like to use a P-channel FET transistor to enable the standlight when the dynamo voltage drops to zero. I've seen circuit designs with PNP bipolar transistor to achieve the same goal, but could the FET be also used ? if operates only with voltage (and uA of current) so would not steal current from the LED... right ?

The rear light:

I bought an additional rear fog light for cars - it's a simple 2 part box (13x6x5cm) with one half translucent red and a reflector inside. There is plenty space for mounting almost any circuit inside of this housing.

The rear light should pull about 50-100mA.
I was going to use 2 or 3 10mm oval LEDs with Vf 2.1V and Ifm 50mA and standlight powering only 1 of the LEDs (again in blinky mode) but might consider also some higher current versions (some 350mA red leds on star from fasttech or dx). 2 standard round LEDs are easier to mount in the original reflector of the housing. placing 2 stars could be more complicated.

To rise the voltage for charging the supercap close to its maximum 5.5V I would need a total voltage of the LED string about 5.5-6V, and together with rectifier bridge the overall voltage of the rear light would be 6.2-6.8V. I could use 2 LEDs and add some silicone or schottky diodes in series to get the desired voltage.

The front + rear combination

A question of how to ballance the front / rear light current is something I'm not sure how to answer. I've read that I could use a resistor in series with the rear light to adjust how much current the rear light gets. Is there some other way ? I found a switching current regulator NCP3065 but I'm not sure it would be worth the complexity.
Or maybe LM317 in constant current mode ? (edit: probably not because the LM317 would try to reach the set current even in slow speed when there isn't enough current in the whole system)


Please I would be very grateful if you could help me clarify some the things I've got wrong or that I don't fully understand. Thank you in advance.

 

[Steve K]

shucks... that's another project that I haven't found time for.... rats.

I'll be interested to see if anyone else has tinkered with this.

 

[OndrejP_SK]

@Steve I was really hoping that you could help. I've read so many useful posts from you. Pity you haven't tried this.

I have some buck converter modules from the bay http://www.ebay.com/itm/301218533352. I wonder if I could try it in a configuration: dynamo - rectifier bridge with smoothing cap - buck converter module - current limiting resistor (1R?) - LED (XM-L) and measure the current through the LED...

Edit: tried to use the buck converter from e-bay and the current through the LED ca. 460mA was the same as when I connected the led directly to the rectifier/smoothing cap. Seems this is not the way...

 

[Edocaster]

I tried testing an MC34063-based 'car cigarette lighter socket to USB' charger running off a dynamo. The components weren't optimised for this (the charger was rated for a 12-24V input), but unloaded, it outputted 5.3V fine. Loaded, however, it sagged to 4.4V. I thought about changing some of the components using an online calculator, but before I had a chance to face that complexity, it somehow failed, taking out a USB 4-port hub I had plugged in.

It failed in the worst possible way, outputting unregulated voltage.

Did I overvolt it somehow? Who knows. So far, that's my one and only experiment with a switching regulator.

Good luck with your progress on this. I think the link you provided earlier included ways of controlling the duty cycle, to allow more current. Most of the off the shelf buck converters give you no direct control.

As for balancing front and rear light, someone once recommended a standalone constant current source regulator for the rear, like this: http://www.onsemi.com/PowerSolutions/product.do?id=NSI45020 - which seems nice and simple (I think you can parallel multiples together for more current), and less 'variable' than picking resistors.

 

[OndrejP_SK]

@Edocaster: thank you for the tip on that current regulator. It sounds like a good idea for rear light. I could use 2x or 4x20mA depending on the LED used. Unfortunately I can't buy it locally - would need to order from abroad = expensive.

Instead of load matching, I was considering just using a buck converter at fixed 50% duty cycle. But I haven't been able to find an IC buck converter contoller which would allow duty cycle control... is there such a component ? Or do I need to build a buck converter from some 555 timer + discrete components ?

If I read minisystem's measurements graph correctly, the current increase should be observable as soon as the voltage from dynamo rises above the Vf of LED. In minisystem's case it was 5V at speed of 20kph (why at 20? shouldn't the dynamo output 6V/500mA at 15kph?).
And if I use only 1 LED with Vf=3V would I get increase in current from lower speed ?



For reference I'm adding a link to an older thread which preceded the evelopment of minisystem's circuit design here http://www.candlepowerforums.com/vb/showthread.php?316815-peak-power-trackers-for-bike-dynamos

 

[Steve K]

Instead of load matching, I was considering just using a buck converter at fixed 50% duty cycle. But I haven't been able to find an IC buck converter contoller which would allow duty cycle control... is there such a component ? Or do I need to build a buck converter from some 555 timer + discrete components ?

no.. no one makes a buck regulator that doesn't regulate. There might be a way to hack a regulator controller IC, but just going with a 555 would be a bit more simple. It does require that you can select the proper components, though.

I'm also thinking that it may be necessary to inhibit the converter when the input voltage is below some fixed level... but it's been a while since I've thought about this stuff, and I think I was still having some problems in my simulations.

 

[OndrejP_SK]

Steve: Thank you for even thinking about this I'm a software engineer with one semester of electrical circuits, but this is over my head. I' barely able to wire a 555 timer or similar IC when I have the specsheet, but designing from scratch and selecting the right components I feel like

 

[OndrejP_SK]

In the meantime I have sketched whaI think could be my rear light. It should be switchable from constant light to blinky when I feel like it (in city I prefer blinky to attract attention, however on country roads steady light is supposed to be better for estimating distance). Standlight in blinky only - both to save power and to attract attention even when the light output might be very low.

I wonder what would happen if the front light failed and all the dynamo output would go to the rear light ? what kind of protection should I add ? also does the current limiting using resistors look usable ?

 

[Steve K]

that seems plausible. You'll need to add parts to the 555 circuits to get them to oscillate, though.

You might also look into using a cmos version of the 555 for the standlight circuit.

In the days of incandescent dynamo lights, it was much more common to worry about what happens if all of the dynamo output goes to the taillight. A common fix is to put a power zener across the dynamo terminals somewhere. A 5 watt zener is probably appropriate, and will need some heatsinking.

I'd recommend breadboarding this before building it. Depending on the LED, it might not end up being very bright.

 

[OndrejP_SK]

Thanks steve. I know i need the timing components for the 555 but i omitted them for the sake of readability. I was planning to use the cmos version for less voltage drop - i just used whatever 555 i could find in the eagle library

 

[Steve K]

Not to detract from your work at all, but when I look at the circuit, it seems like there ought to be a way to design it with a single 555. It would probably involve using a BJT or mosfet on the low side of the LED string to do the pwm'ing. The 555 would drive the BJT or mosfet, of course. The 555 would have to be powered from the supercap, but I don't think a cmos 555 would draw much power.

The next trick would be to get rid of the LM317, and just use the voltage drop across the red LEDs to charge the supercap. This might conflict with the pwm'ing of the LEDs, though... turning off the current through the LED string would result in high voltage at the supercap.

just some thoughts..

 

[Edocaster]

Instead of load matching, I was considering just using a buck converter at fixed 50% duty cycle. But I haven't been able to find an IC buck converter contoller which would allow duty cycle control... is there such a component ? Or do I need to build a buck converter from some 555 timer + discrete components ?

If I read minisystem's measurements graph correctly, the current increase should be observable as soon as the voltage from dynamo rises above the Vf of LED. In minisystem's case it was 5V at speed of 20kph (why at 20? shouldn't the dynamo output 6V/500mA at 15kph?).
And if I use only 1 LED with Vf=3V would I get increase in current from lower speed ?



For reference I'm adding a link to an older thread which preceded the evelopment of minisystem's circuit design here http://www.candlepowerforums.com/vb/showthread.php?316815-peak-power-trackers-for-bike-dynamos

One approach might be to force the system to work at a higher voltage. This seems to be what the German 'forumslader' charger uses: http://www.forumslader.de/fileadmin.../Download/UniladerUSB/Schaltplan_Si-Diode.jpg

...the 12V linear regulator means that the 5v switching buck converter downstream always has an excess of power to play with and convert to higher current, rather than dragging the voltage down itself. Plus, the switching regulator no longer has to deal with a wildly varying input.

I suspect if I had read all this earlier I would have avoided the unnecessary meltdown of the cigarette lighter socket charger mentioned above.

 

[OndrejP_SK]

Edocaster but at what speed does the dynamo reach 12V ? it doesn't make sense to me.

 

[Edocaster]

The forumslader 12V system seems similar to the Pilom system from circuit 8 onwards on this page: http://pilom.com/BicycleElectronics/DynamoCircuits.htm - namely, there is a switch between full-wave rectifier mode and voltage doubler mode.

If you're working out how fast you'd need to reach 12V on full-wave rectifier mode (i.e. the 'normal' mode for most people who are rectifying bike dynamos), the diagrams under circuit 7 on that link should give a clue: 12V is somewhere between 3 and 4 LEDs on those charts, so the answer is about 20km/h.

If you want to ride slower, then voltage doubler mode can give you 12V from a little over 10km/h. Half the amperage, but what counts is you are getting 12V.

My own experiments with a 5x LED light back this up - I need at least 13.3V to light up my string at all, and both modes work almost exactly as the Pilom page predicts.