Efficient LED circuitry for Bike Dynamo

[ACharnley]

Hi all,

I'm rethinking my bespoke LED setup on my bike in order to provide a bit of time when I stop pedalling to keep the light on. Currently I have a dynamo linked to a USB 5.25V out device, followed by a Skhotty diode so (4.8V) linked to a 0.33F supercapacitor (5.5v max) which goes into 5x 2.2V/20mA red LED's via 5 resistors.

The problem is once I stop pedalling the rear light is immediately dimmer due to the resistance of the capacitor, and it decreases quite rapidly as the voltage drops. Ideally then I need to have some circuit trickery to keep the 2.2V constant as the capacitor discharges.

I also have 5x 25mA 3.2V white LED's on the front. These consume quite a bit more power and I didn't believe it feasible to hook them up, but I'd like to!

My Idea

Is to feed the output from the capacitor/power into a small DC-DC boost, a device for which costs $0.60 on aliexpress. This will ensure voltage as low as 0.9v goes to 5v. Then I'd use two small DC-DC buck's to take the 5v down to 2.2v & 3.2v. Efficiency should be greater than the dropping resistors.

Does this sound feasible? I don't know where to start for calculating the size of the capacitor, but I know large ones like 5F/5.5v are now available. Would they have enough charge to last about a minute?

Thanks!

Andrew

 

[Joe G]

maybe a stupid question, but why not add 2 AAA rechargable batteries in line, your dynamo will keep them topped up, or a small solor panel depending on your location.

 

[Tony Stewart]

It's always best to choose batteries that match your load requirements.
For the WHite. Use 3.7V LiPo cells that drop to 3.2V, then your drop voltage and efficiency is greatly improved. If ganging several LiPo's Either put the LED's in a series string or use 0.05Ohm resistors or a length of AWG30 magnet wire in series with each Lipo in case of mismatched voltage or SoC.

A LiPo cell can be charged up to 4.2, which you can monitor with a White LED in series with a silicon diode. The 5mm LED will have ~15 Ohm internal resistance starting from ~2.9V when diode is saturated. Or for Red HB , it is about the same (15 Ohm) except from 1.8V threshold.

If you get 1W WHite LEDs, each one is around 1 Ohm from 2.85V and up
If you get 3W White LEDs, each one is around 1/3 Ohm from 2.85V and up

CR123A Lithium primaries are cheap on **broken link removed**, Amazon and are exactly 3.0V which match most power white LED's with no series R. Disposable but no extra peddle effort and low cost <1$ ea. It may or maynot drive at full brightess of a 3W LED but last longer running around 1W. ( depending on source of LED and specs for ESR from VI curve).. but again 100% efficiency

 

[Tony Stewart]

This 30deg LED using only 10mA of 20mA rating @ 20Cd on a CR123A from Panasonic rated at 1400 mA might last up to 140 hrs.

It wont light up the road, but others can see you for a long way.

You can't really beat a helmut mounted rechargable (hand type) Torch with 5 degree spot for a headlamp.

 

[audioguru]

Riding a bicycle at night in the dark? Why?
Here in Canada I have seen only a few people doing it (maybe because they have no lights and I could not see all millions of them?). Perhaps it is common in China.

I admit I rode my bicycle far from home then it started raining cats and dogs. When the rain stopped and the road dried (no fenders) then I rode home in the dark (but streetlights are bright) and I survived.

There was a guy who tried a computer fan as a generator on his bicycle. Its output current was almost nothing.

 

[ACharnley]

I get 5-15w from the dynamo, the modern ones are very good.

In this case I'm trying to not rely on batteries, unless they are usb rechargeable. The reason for this is an extensive tour next year through Afgan and some remote regions.

A one cell 18650 power bank would do it, but it would be great if I could just use capacitance instead - I only need a minutes worth of light.

 

[Tony Stewart]

Ultracaps are very interesting for portable Electric Irons and motors etc, but not really matched for LEDs
Maxwell ultracap

BCAP3000 P270 K04
$60/pc
3000F
2.7V
0.29mΩ@100A
1500Hrs @65°C 10yrs @25°C
3Wh (not practical to drain all energy)
9,300A s.c.
130A@ΔT=+15°C

Voltage Monitors for Integration $85

These are great for long life if cool , like controlling pitch motors on Windmill Props,
White LED's need more voltage. Boost Chips to 3.2V are only 80% effic. So you need to string several in series to a Buck Regulator. Efficiency improves with voltage difference.

 

[Misterbenn]

look for the Gaia HUGD-50 module this is a boost buck hold-up module.

Not sure if you can get this part but if not this is the type of functionality I'd try to copy.

 

[Tony Stewart]

As with all Buck Boost types, efficiency drops <=80% when Vi-Vout is low and can be as high as 98% when difference is high.

 

[Misterbenn]

Yes but currently 56% of the power is being dropped over the current limiting resistor so 80% would be an improvement.

 

[Tony Stewart]

LTC4020 supports these Buck-Boost topologies with different battery chemistry shown below.


Experience in extending the Lifetime Watt-hrs on LiPo cells, reported at Battery University, concluded that reducing DoD to 10% and reducing temperature max to 30'C by reducing C rate and ambient significantly..

Chart to follow

 

[Tony Stewart]

https://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

WHat this study shows is if you Charge a 1500mA pouch cell Mobile phone battery with 1C and drain with 1C such that 100% DoD = 3.0V and full charged is 4.2V you get 500 cycles of life or 2700 watt-hours @3.6V*1.5Ah=5.4Wh per cycle consumed in 1 hour (1C)
What this curve shows, that I made , is you can expect up to double the lifetime Watt-hours if you regulate consumption near 50%. Battery suppliers often ship at 50~60% SoC (or 50~40% DoD) to maximize shelf life for a reason. Test results courtesy of Cadex.

This comes at the expense of 4x as many charge cycles but may well be worth it if you have twice the capacity that you need in a single charge.

 

[audioguru]

Why are you thinking about using a super capacitor instead of a rechargeable battery?
A capacitor voltage begins dropping quickly the moment you begin discharging it. A battery voltage remains fairly high during its discharge then drops quickly when it is almost dead.

 

[dr pepper]

A lipo battery feeding a little inverter supplying the leds would work, the lipo being charged by the dynamo.

The more circuitry you have the less efficient it will be, however do you really need super effcicent?

My wifes pushbike has a commerical lighting system, its a hub generator which feeds a light unit that has a simple blocking oscillator voltage booster powering leds, and a pair of rechargeable coin cells, I didnt think it'd be any good, but it works well and its lasted 4 years so far.

 

[Tony Stewart]

That is correct audioguru ... to put some numbers on that.
Essentially an ultracap must discharge to 0V from 2.7 to capture all the energy, whereas a LiPo has >90% of its energy at 1C discharge rates drops < 30% of voltage from 4.2 but at lower discharge rates the drop in voltage is flatter after initial drop from 4.2 to 3.8V, dropping only to 3.6 before falling quickly to 3.0

Thus for low discharge rates , which extends life of battery from minimal temp rise, Ic=Cdv/dt
for 5Ah 7.4V 2S battery at 1/5C discharge rate = Ic*dt/dv = 1A*3600s*5h/(2*(3.9-3.6V))=30,000 Farads !! but at 7.4V (2S) or ten (10x) times that of the largest MAXWELL ultracap at 2.7V which if added in series for 3 caps would be 1/3 the capacitance. A single 5000mAh cell would be 60x the capacitance of an ultracap. cheaper but not last 10 yrs like an ultracap if kept Cool then it only lasts 3000h.

So ultimately I dont agree ultracaps are the ideal solution, yet.

Consider $60 for one 3,000F Cap 2.7V max if allowed to drop 0.7V gives up this amount of energy in watt-seconds [Ws] or Joules[J]

E=½C (Vi²-Vf²) from initial to final voltage.

assuming Vi=2.7V and Vf=2.0V, C=3000F ,

E=1500 Ws [J]

Consider $38sale to $52reg for one 2-series cell (2S)
Venom 20C 2S 5000mAh 7.4V LiPO Hard Case - UNI
[URL='https://www.amazon.com/Venom-5000mAh-7-4V-LiPO-Hard/dp/B000VUF3O2/ref=sr_1_17?ie=UTF8&qid=1446220473&sr=8-17&keywords=lipo']or similar SPC LiPo Power, Lithium Polymer (LiPo) Battery Pack; 5000mAh, 40C, 2S, 7.4

[/URL]

https://spcracingbatteries.com/index.php?route=product/product&path=59&product_id=81


Note the Thunder Power on top are very expensive $120 for a 2S and all exceeded 4500mAh , which may not represent the total population

Then for a std 18650 Cell ( almost like AA cell size )

FWIW

Tony

 

[Joe G]

me personally I would just use the smallest battery possible, feed the dynamo into them, and power the leds with the batterys. You would use less parts ($) and have much more efficiency, less problems and would most likely take up less room. And in the event that something goes amiss while on your trip, you won't be stuck looking for some special IC's solder or a solder iron to fix it......
remember "KISS"... keep it simple.

 

[audioguru]

The few bikes with lights at night that I see usually have dimmed LEDs because their battery is running down.
Like the electric bikes that are crawling along the road with a battery that is almost dead.

 

[Tony Stewart]

As long as the generator does not overcharge the battery, thats fine and doesn't load down the cycler with drag and a shunt regulator and the generator is somewhat matched to the battery voltage which is somewhat matched to the LED voltage with 1s or 2S or 1s10p or 2s10P or whatever....... then and only then does the KISS principle apply. But then you have to live with dim LEDs with a current limiting resistor and a low voltage battery down 0.5V.

But if it were me I would use a CC regulator buck style with a 50mV current shunt sense resistor for low dropout and 95% efficiency.

 

[Diver300]

These lights:-
https://www.sjscycles.co.uk/brompton-rear-dynamo-standlight-led-inc-reflector-prod2832/
have a supercapacitor inside them, and they charge up in less than a minute from a dynamo, and run for 10 minutes or so after the bike stops. The light doesn't seem to dim when the bike stops, as long as it has run for a minutes or so before. When left for a long time, the light slowly fades out over several minutes.
The corresponding front light:-
https://www.sjscycles.co.uk/brompto...-shimano-c-w-bracket-and-main-loom-prod31204/
also stays on when stationary, but dims when the bike stops. It seems to have two brightness levels, and I guess that the lower brightness when stopped is done to make the supercapacitor last longer.

Both lights come on as soon as the bike starts to move. There is an on-off switch on the front light, with an output to the rear light so that the one switch does both lights, but I just leave the switch on.

I've not opened either up, but I guess that the capacitor runs at a higher voltage than the LEDs, and some current regulator means that the same brightness is achieved over a wide range of capacitor voltage, so that the light doesn't dim as the voltage reduces. Obviously there is some point where the voltage will be too low to keep the brightness, but until that, the brightness is constant.

I wouldn't want a battery backup when the supercapacitor one works so well. I have a Cateye Stadium on the other bike. It is an HID light, run from NiMh batteries, only. It is much bigger, I need to remember to recharge it, the beam pattern is far worse, and it is only a bit brighter, so I am thinking of changing to a dynamo system. Although I could design and build something, the commercial devices do just what I want so I would get one of them.

 

[Diver300]

I think that your circuit would be massively improved with constant current circuits feeding the LEDs. The LEDs are 2.2 V, and the supercapacitor get to around 5 V, so there is quite a lot of energy to be obtained between 5 V and 2.2 V, if something more sophisticated than a resistor is used.

Something like the first circuit on https://www.eecs.tufts.edu/~dsculley/tutorial/opamps/opamps7.html

The voltage reference would have to be separate from the supply, and the op-amp would have to work to low voltages. There would have to be one of these circuits for each LED.

Alternatively, there are boost circuits for LEDs that work very well, and are more efficient. That would put all the LEDs in series, so only one boost circuit would be needed.

A quick search found this https://www.linear.com/product/LT1937

If you have a 5 F capacitor, charged to 4.8 V, to discharge to 2.5 V the energy available is 41.975 Joules. The LEDs take 5 * 0.02 * 2.2 = 0.22 W, so it will take them 191 seconds to use the 41.975 J of energy. If the switch mode converter is 84% efficient like it says it is, that reduces the time to 160 seconds, so still over 2.5 minutes, which is probably all you need.

I haven't read the data sheet for the boost converter, and there may be limitations that mean the design isn't easy.

I chose 2.5 V as the boost converter won't go lower. There is not much difference in energy available if you go to a lower voltage. The energy is 1/2CV^2, so there is little energy at low voltages. Your post mentioned going down to 0.9 V. Well taking the 5 F capacitor down to 0.9 V would give you 55.575 J instead of the 41.975 J if you go to 2.5 V, which is a significant advantage, but probably a lot of work. However, if you can ditch the Schottky diode, and charge the capacitor to 5.25 V, that would give you an additional 11 J of energy.

If you are using non-rechargeable batteries, there is a lot to be gained by using them down to as low a voltage as possible, to avoid buying batteries more often. However, in this case you are paying for conversion circuitry and design time, when you might be better buying a larger supercapacitor to store more energy.

As I said in the previous post, buy the whole thing. It is much easier. However, if you do want to build something, I think that a boost converter is the way to go. There is no point in boosting to a higher voltage and then using a buck regulator to reduce the voltage, when you can put the LEDs in series for a higher voltage, and simply arrange the boost converter to give out the LED current you want, rather than an intermediate voltage that will need converting again.