Very simple/cheap dynamo led light

[PhxCycler]

Ktronic,

Back in Post #6 you have a very simple circuit which uses a bipolar capacitor. You then substitute 4 polarized caps in its place later on. I'm curious. Is there any advantage to the 4 polarized caps over the single bipolar other than ease of finding them at hobby shops? I really like the simplicity and size of the single bipolar circuit and I can get them very inexpensively.

Also with regard to the bipolar circuit. How did you go about determining the size of the cap? Was it trial and error, or do you have some equations tucked away somewhere that you'd like to share?

Thanks,
-Mike

 

[Martin]

PhxCycler, the 4 bipolars (you could actually replace them with 2 bipolars being biased with a resistor towards the rectifier "+" output if you give up the option of going to a voltage doubler) are purely for reasons of cost, size and availability.
If you can find bipolar capacitors at a good price, no problem using them.

How to determine the value ?
If you find a good electrical model of the dynamo, you could do PSPICE. Well, I don't have that model so I do by experiment:
- Put a reasonable start value first
- Measure the LED current (DC) while you ride the bike
- If the capacitor is too small, the current will increase as the speed increases, at one speed it will reach a rather high max value, as you go faster the current will decrease again. Typical values are 500mA at very high speed, 600..1000mA at the peak.
- If the capacitor is too large, the LED current will rise slowly with speed and remain stable from a certain speed on.
- The perfect capacitor value will produce a current that rises fast with speed, has a very slight peak at a specific speed and then drops back as speed increases. By thumb, the peak current should not be higher than 110% of the max speed current.

Kerry likes to go for a higher peak current, my approach is more conservative.

If you let me know what type of dynamo you use and how many LEDs you connect in series, I can probably recommend you a capacitor value from my wealth of measurements.

 

[ktronik]

Ktronic,

Back in Post #6 you have a very simple circuit which uses a bipolar capacitor. You then substitute 4 polarized caps in its place later on. I'm curious. Is there any advantage to the 4 polarized caps over the single bipolar other than ease of finding them at hobby shops? I really like the simplicity and size of the single bipolar circuit and I can get them very inexpensively.

Also with regard to the bipolar circuit. How did you go about determining the size of the cap? Was it trial and error, or do you have some equations tucked away somewhere that you'd like to share?

Thanks,
-Mike

the ripple current that the bi-polar cap can handle, is much less than 4 low ESR caps...for a long term usage they are under spec for what we are using it for...

give me your email & I will email you the numbers for all the different caps sizes...Martin has done all the testing & made some spread sheets, I can send you one of those.

K

 

[Martin]

Bipolar caps are usually rated for AC usage, but if their voltage is low enough and their ESR high..
The polarized ones also need to have low ESR and a rather high voltage (63V or 100V).
Why not try. You won't break more than the capacitors and diodes, the LEDs will survive.

 

[PhxCycler]

If you let me know what type of dynamo you use and how many LEDs you connect in series, I can probably recommend you a capacitor value from my wealth of measurements.

Thank you for the explanation Martin! I have a Schmidt SON hub generator and am still trying to determine if I want 3 or 6 LEDs. I'd really like to see cap values for both 3 and 6 and would ultimately like to select between 3 and 6 on the fly. I can buy bipolar caps for about 25 cents, so they are cheap. I have two light housings made with 3 Seoul U bins each.

I also built up test jig with a motor and speed controller so that I can rotate my wheel while monitoring the circuit with my meter and oscope. I do see the peaking that you describe. Interestingly, I have measured 0.96 A with 3 LEDs at ~20 mph. I thought the hub was only capable of 0.5 A max.

I have the switchable doubler circuit (Post #49) built up and my motor definitely senses the load difference when I switch between the doubler and full wave. The speed will go up and down a few mph because of the load when switching between the modes.

Thanks,
-Mike

 

[Martin]

For 6 LEDs on a bridge rectifier you use 100uF (150uF would be more conservative) and for the 3 LEDs you use 470uF (you can give 330uF a try but this could do too much peaking).
The cap values are based on my Shimano hub, the SON is probably not too far away from it. Now with your setup, you can actually do your own speed vs power curves so you really know what fits the SON.
The max current you get at the peak can be considerably higher than what is nominal. This is to be expected. And to some extend, we exploit this.

Your motor's speed control is not too well stabilized, or you would not see the speed changing with load. Wired it correctly, connected the feedback ?

 

[PhxCycler]

For 6 LEDs on a bridge rectifier you use 100uF (150uF would be more conservative) and for the 3 LEDs you use 470uF (you can give 330uF a try but this could do too much peaking).
The cap values are based on my Shimano hub, the SON is probably not too far away from it. Now with your setup, you can actually do your own speed vs power curves so you really know what fits the SON.
The max current you get at the peak can be considerably higher than what is nominal. This is to be expected. And to some extend, we exploit this.

Your motor's speed control is not too well stabilized, or you would not see the speed changing with load. Wired it correctly, connected the feedback ?

So I'm assuming that these cap values are for the bipolar circuit? Is there a 1:1 correlation between the capacitor value used for the single bipolar cap and the caps used in your doubler circuit?

Also, my motor setup has no feedback (except me, of course). I found a cheap pump motor and motor speed controller on eBay. I mounted a lawn mower wheel on the shaft and run it against my hub wheel. It actually works pretty well, but the motor gets pretty warm a slow speeds. I don't think it was ever meant to be run like this.

 

[Martin]

Yes, simple. C1=C2=C3=C4 = Cbipolar

If you can bolt a section with a smaller diameter to the current motor wheel, this will help with low-speed measurements.

 

[PhxCycler]

Yes, simple. C1=C2=C3=C4 = Cbipolar

Well, that's certainly easy enough! :thumbsup:

Now I'm just awaiting my assortment of bipolars to arrive. I'll play around with the combinations and see which I prefer. I've looked at the spreadsheets that Ktronic sent me and so I have a good idea of where to start. You guys have done an incredible amount of analysis on this setup. Thanks for sharing all of it!

 

[PhxCycler]

I've played around with various values of bipolar caps and have found a combination that I like. I'm using the very simple circuit with the 4 diodes, single bipolar cap, and a 2200uf smoothing capacitor.

I like the performance of a 69 uf bipolar (47uf and 22uf in parallel). I'm able to drive 3 LEDs just above 1A at 23 mph. I also tested the 6 LED setup and can achieve a maximum of 620 mA at 23 mph. This is a roadie application for my Schmidt SON hub.

I measured the current thru the LEDs at various speeds then converted it to lumens out of the LEDs. I didn't correct for optical losses, etc. I just used the Lumen vs. Current curve for the Seoul P4 LEDs. Here are the results. I rode a few miles last night with the 3 LED setup and it was awesome. I hope to do a real night ride soon where I will test the 6 LED setup. I have a number of optics that I'd like to try.

 

[Leow]

Phxcycler,

If you're going for a switchable 3/6 LED setup wouldn't it be more efficient to forgo the bipolar boost capacitor ?

I've bought enough capacitors to build Martin's boosted manual switched/doubler cct but I'm having 2nd thoughts.

Anyone done any measurements on power input(loss) vs lumens?

If hub dynamos are 60-70% efficient and you are getting 9 Watts of LED power (.9A @ 10v) you will be using upwards of 15 watts + cct losses at pedalling speed.
It's a different matter going downhill at 25+mph where MAX Brightness is paramount and efficiency is secondary - good time to switch on another bank of LEDs.

LeoW

 

[PhxCycler]

If you're going for a switchable 3/6 LED setup wouldn't it be more efficient to forgo the bipolar boost capacitor ?

Not sure about the efficiency. The value selected for it has a marked effect on how the above curves act. Have you tried removing it all together? What kind of response did you get?

Anyone done any measurements on power input(loss) vs lumens?

I haven't done any input measurements. I think that would be very interesting results to see. Unfortunately, my meter doesn't measure AC current, only DC.

If hub dynamos are 60-70% efficient and you are getting 9 Watts of LED power (.9A @ 10v) you will be using upwards of 15 watts + cct losses at pedalling speed.
It's a different matter going downhill at 25+mph where MAX Brightness is paramount and efficiency is secondary - good time to switch on another bank of LEDs.
LeoW

My thoughts exactly. This is precisely why I'm trying to come up with a usable 3/6 switching arrangement. Unfortunetely, I'm seeing large spikes during the switch (when using resistors instead of LEDs) and I'm concerned it will blow the LEDs out.

Another aspect of all this is the duration I need to use these lights. I often ride multiday events where I may need lights for 8-12 hours per night for 3 days straight. Battery lights are a royal pain in this situation. I've also found that my night riding speed is effected by the amount of light I have available. So, while I do understand that the power comes from my legs, I also know that I ride faster even with the extra power requirements. I've been running dual 3W halogens (6W total) for years and want more light.

 

[ktronik]

all theory aside...

I have been riding the dyno system for almost 2 years now... I race it in night MTB events...

I am still posting some of the fastest lap times...it does not seem to slow me down @ all...

just the 3 LEDs is bags of light!!!

for less drag I ran it in voltage DB mode...

K

 

[Leow]

all theory aside...

.........

for less drag I ran it in voltage DB mode...

K

Any one know the theoretical/practical efficiency of a simple doubler cct ?

75% - 85% - 95% ?

Offers !

I either want to switchout 2 LEDs or use the doubler cct to turn down the bank of 3.

Phxcycler most of the guys I know who do multiday Audax's used to use twin Halogens but have switched to twin 3W Leds (Solidlights)
http://www.solidlights.co.uk/products/1203d.php
These switch the 2nd LED automatically on with increasing speed but they aren't any brighter than my current Lux V / simple rectifier.

LeoW

 

[Martin]

The voltage doubler and the full wave rectifier have their relevant losses in the diodes' Vfwd. This is 0.3 to 0.4V per diode.
A dynamo is pretty much a current source, so the voltage across the components is what matters:

Bridge Rectifier
3 LEDs = 3 x 3.4V = 10.2V
2 Diodes 1N5818 = 2 x 0.35V = 0.7V
=> 7% losses / 93% efficiency

Doubler
3 LEDs = 3 x 3.4V = 10.2V => 5.1V at doubler input
1 Diode 1N5818 = 0.35V
=> 7% losses / 93% efficiency

I assume you're using capacitors that are sufficiently large not to heat up and burn precious energy.

Bottom line: If you have been happy with the bridge rectifier, you can tolerate the doubler just as well.

 

[Leow]

Thanks Martin,
I was worried when you mentioned that the capacitors can heat up (lost energy). But if they have enough "headroom" they shouldn't heat up ...

LeoW

 

[Martin]

Re: MOSFET-rectifier beats all

any switch has to work automatic - are there circuits for that?

as to rectifier:
we use one made from two IRF7319 Mosfets around here.
Makes the three leds already give a useable light at speeds, when normal rectifier, Schottky-rectifier or voltage doubler start to give a slight glow.

K1 is input
K2 output
I dremeled some boards and the Mosfets work great.
There is only one thing to care for: one of the leds (or a diode) must not be bridged with a capacitor - else the mosfet-thing does not work

Eventually I have measured the MOSFET-rectifier and compared it to a plain schottky bridge-rectifier and one with tuning capacitors. I also compared it to my doubler / bridge-rectifier autoswitch.

Test conditions:

Shimano DH-3D71 hub generator

• MOSFET circuit: Smoothing cap of 2200uF across two of 3 LEDs, as the rectifier doesn't tolerate a capacitor right at its output. This arrangement drives the remaining LED with AC, reverse voltage reaching -5V which is exactly the absolute max rating of a Cree XR-E. Current and voltage I measured on the 2 LEDs that have the smoothing capacitor across, then added 50% to include the third (AC-driven) LED.

• Schottky rectifier: 4x 1N5818 and 2200uF smoothing cap, connecting to a string of 3 LEDs.

• Schottky rectifier with 470uF tuning cap: 4x 470uF tuning caps, wired for an effective 470uF total capacity, 4x 1N5818 and 2200uF smoothing cap, connecting to a string of 3 LEDs.

• Doubler / bridge rectifier autoswitch: Combines a 2x 470uF doubler and a 1x 470uF bridge rectifier. Selects the best of the two.

See the power curves (0..50 km/h and 0..15 km/h):

Observation:

• Doubler (autoswitch) superior at low speed. At 5km/h, it delivers 1W vs 0.15 W of the competing systems.

• Anything with tuning caps superior at mid speed.

• Curves merging at high speed

There's a small difference b/w the MOSFET rectifier and the schottky rectifier while the other two circuits outperform these two at low and mid speed.
The one LED of the MOSFET rectifier that doesn't have the smoothing capacitor across, flickers heavily.

My conclusion on the MOSFET rectifier: While it offers a little bit more power over a basic schottky rectifier at low speed, I don't think it's worth the extra effort and the strong flicker of one LED.

 

[Martin]

Just uploaded my revamped page on dynamo LED drivers.
I think you'll like it (let me know if I'm wrong)
Martin

 

[ktronik]

WOW what a body of work!!! One day I do hope the world gives you the thanks you deserve... :goodjob:

You really have bought dynamo systems into the modern age... I would not consider using any other system now that I have tasted 3 LEDs on the more basic system...

Can't wait to hook up the 6 LED version!!!!


Best

k

 

[Martin]

Thanks Kerry, the world actually does since I put some Google ads !

Did you notice, in the name of miniaturization I have drawn a circuit that eliminates 2 out of 4 tuning capacitors without sacrificing performance ? It's manual mode selection only, using a standard 3p2t switch. It won't get smaller or simpler, with today's components.

Now I'm waiting for Cutter to come up with a 4..6 LED Rebel light engine to go into my sugar caster..