Semi-DIY dynamo light build

For discussions about bikes and equipment.
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Semi-DIY dynamo light build

Post by edocaster »

As mentioned on a few other threads, I've been in the process of adapting an off-the-shelf light for dynamo use. The plan was to use it to complement or as an alternative to my Axa Pico 30. Here's the progress so far - maybe it will help give others a few ideas.

First port of call is this fantastically informative page: http://pilom.com/BicycleElectronics/DynamoCircuits.htm - which explains the basics of building a dynamo light circuit. This set me on the search for a 2x LED light head so I could wire up the LEDs in series (you can go for more than 2 LEDs, but the increased voltage requirements mean you'd have to pedal faster for the light to come on at all).

I bought a Solarstorm X2 like this: http://www.ebay.co.uk/itm/SolarStorm-X2 ... 1192072049 - the advantage of going pre-built are mount, housing and lens are all ready (although not necessarily to one's taste). Plus this one had screw access to the rear.

The original Solarstorm X2 circuit - there is some space at the back of each LED
The original Solarstorm X2 circuit - there is some space at the back of each LED


I then set about building the front half of the circuit shown two-thirds down this page: http://www.candlepowerforums.com/vb/sho ... or-dummies - without the tuning capacitors (space was the main reason for omitting those components), and I didn't need the taillight side.

Space was still too tight for the remaining components, so the bridge rectifier (four diodes which convert the AC of the dynamo to DC which the lights can use) had to go outside the light. I put those inside an in-line switch a little way along the cable.

Ordinary lamp in-line switch, filled with fiddly Schottky diodes
Ordinary lamp in-line switch, filled with fiddly Schottky diodes


The original driver in the Solarstorm X2 is of no use for dynamo usage, so I removed the one main component which was wasting space (an inductor coil). I was tempted to remove the main circuit board too, but unfortunately that gives the back plate necessary shape, so I just taped over the components.

The real problem was shoving all the parts in and closing the back again (I had unnecessarily long wire). I swear I broke something, plus my soldering is pretty awful. I'm not looking back in there until the light fails! Plus I had the wrong sizes of heatshrink for the in-line switch, so now it looks a mess of electrical tape.

The dynamo circuit - extremely tight cram! I lined most parts which could short out with electrical tape... the rest was luck!
The dynamo circuit - extremely tight cram! I lined most parts which could short out with electrical tape... the rest was luck!


Testing it today, it works! Unfortunately I couldn't fit a smoothing capacitor larger than 470uF, which is too small to do any real smoothing, so the thing still flickers under 15km/h. I think it may have flickered less until I hit a big bump, which might have knocked off the capacitor... I don't know, and I'm not opening it up!

The standlight is the real revelation. With a 1.5F supercap (just about fits in there) it's at least as good as the one on my Pico - bright for about 3 minutes, then fading for the next 10 minutes. That was one of the most expensive components though. The circuit dictates that only one light ('LED 2') will be lit by the standlight, but that's no biggie. That LED also lights up first in the circuit (I presume when there's not enough voltage for both LEDs, current flows down the diode towards the supercap, and is then shared between the supercap and LED2), so that actually helps low speed illumination.

The beam shape is so-so. It's an MTB light, after all. The Pico still has better graduated road illumination. But the DIY light has more spread. There are still a few niggles, so I'll ride it a bit more, and then diagnose problems/think of improvements.

Total cost? I'd estimate £30. The light was £14, and the supercap £8. The other components I'm only partially counting, as I still have spares of those.

Hope this helps someone. If anyone has any questions/suggestions, chime in!
Brucey
Posts: 44705
Joined: 4 Jan 2012, 6:25pm

Re: Semi-DIY dynamo light build

Post by Brucey »

very interesting....

It is a shame you have started with such a small housing to work with. I wonder if an auxiliary box isn't a bad idea here; you could keep all the electronics in one place then, and just have simple/compact lights with little more than the LEDs in. My suspicion is that if the light is flickery at speed, maybe there is a fault in the circuit as built. If you have a 'scope it is worth testing the circuit to see that it is working properly. It is worth measuring the current/volts actually going through the LEDs and comparing this with the current/volts at the generator; a lot of these circuits can be really lossy; this makes for a light that isn't as bright and/or a light that doesn't come on at low speeds.

I have in mind a design that has lower losses in the rectifier, and automatically switches from paralell to series operation with speed.

But right now I'm building a simpler (no standlight), cheaper front light which is intended to make the most of the feeble output of an SA dynohub. It will have a (different) low loss rectifier and will be built round some LEDs I have found which have a low forward voltage of ~2.45V @ 100mA. A nominal 1.8W will be used thus;

0.18W -rear light
0.21W- rectifier
1.41W - front lamp LEDs

I expect to be showing a (flickery) light front and rear even at a walking pace.

Target cost of the build is about £6-10. There can be as few as nine components in the circuit, including the switch and the rear light parts.

cheers
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Re: Semi-DIY dynamo light build

Post by edocaster »

I have no scope, but there are no major symptoms of anything amiss, electrically. The flickering seems to be typical unsmoothed rectified current, and peters out by about 15 km/h - when the pulses would be passing about 50Hz.

The typical calculation for a smoothing capacitor is C = I*t/ΔV (where t is the time and ΔV is the voltage drop you want to bridge). So, allowing a generous 1V drop over the 5.7V the two CREE XM-Ls expect at 500mA, gives t as a miniscule 0.00094s! The wheel would have to be doing 38 revolutions per second to do true smoothing! That's why the typical smoothing capacitor is 10 or so times larger. My 470uF is probably doing very little. I guess it's more critical if you're trying to smooth current for a charger, for example (even though a regulator will probably be able to deal with the last of the ripples). I can live with flicker at low speed, especially in the city.

I did consider a project box, but really wanted to keep things simple and small.

I need to try out the light again tomorrow, but the brightness is more or less as I would expect of up to 500mA going to the LEDs. Equivalent to between the low and medium setting on the original light when run off batteries. This tallies with measurements I've seen people give for the Solarstorm X2 for (battery) current on low (figures between 150 and 400mA) and medium (figures between 700 and 900mA).
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Re: Semi-DIY dynamo light build

Post by edocaster »

Hmmm... looks like I was a little too enthusiastic about this. While everything seemed to work like it should, brightness was way below what I was expecting, and less than some single LED torches which I know have a battery current of under 500mA. So today I took the dynamo light apart to measure current to the LEDs. The reading was a disappointing 50-60mA.

I then set about removing as many components as I could, including the standlight, and just driving the two LEDs in series off the rectifier. Result? About 100mA. Very disappointing.

While my multimeter could be a bit off with unsmoothed current, it's still very clear that output is way below what it should be (presumably 500mA at the LEDs is achievable at reasonable speeds? I was hand cranking a second bike upside-down to drive this bike's front wheel to about 30km/h). Does anyone know what could be the likely culprit?

I've also tried driving the whole light from 4xAAA. The voltage wouldn't be enough to light both LEDs, so only one LED (LED 2 - the one connected to the standlight) lights, but current is low there too (about 70mA). Although I presume that's a result of passing through two diodes and a resister to get to LED 2, with the resistor limiting current. I'm not sure what to diagnose next...
Brucey
Posts: 44705
Joined: 4 Jan 2012, 6:25pm

Re: Semi-DIY dynamo light build

Post by Brucey »

if you try just a single LED off the rectifier, no smoothing cap this will tell you if there is a problem somewhere else or not.

This arrangement should easily produce 500mA plus into the LED. However without some form of regulation the LED could see a touch more at speed and fail.

If if doesn't work there are only a few places it could be going wrong. If it does work then you can reintroduce parts one at a time and see where the issue might be that way.

cheers
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~
MikeF
Posts: 4347
Joined: 11 Nov 2012, 9:24am
Location: On the borders of the four South East Counties

Re: Semi-DIY dynamo light build

Post by MikeF »

Thanks for posting the results of your experimental light. Not sure why you are getting such a low current through the LED. As Brucey suggests, try a basic circuit. According to the datasheet the Cree XML LED has a maximum forward current of 3Amps so you are unlikely to destroy it by current from the dynamo. Heat is the biggest problem as regards destruction, but it looks like the LEDs have adequate cooling.
You say your soldering is not too good. Have you made good enough connections? eg not a 'dry' connection meaning where the solder has not flowed enough to make good contact?

Useful information here on output from a Shimano hub http://www.mikrocontroller.net/attachment/14850/BikeLight.pdf. They conclude that the generator has a winding resistance of 4 ohms and an inductance 120mH which is in parallel with a resistance of ~55ohms. I note that my 3N80 also has a DC resistance of 4 ohms and appears to have 14 pole pairs, so maybe the fundamental electrical innards of all Shimano hubs are little different.
This person has also found similar results http://parttimetinkerer.wordpress.com/

This YouTube video provides useful info on the output of a hub http://www.youtube.com/watch?v=pjuJmwMGzxQ#t=185. It appears the maximum rectified current is 540 milliamps, so this is the most you will get unless you add compensation/tuning circuitry. The result of 540 mA is not surprising as the hub is rated at 6v 3W which is 500mA. The voltage is more or less dependent on the connections to the hub, so if you connect your Cree LED directly across the rectified output, it will have a voltage of ~2.5 to 3v.

And another interesting video herehttp://www.youtube.com/watch?v=inn0mOAUvxw although my Russian is extremely limited. Very simple idea of hand cranking to test results. Both these videos have used a bicycle speedo presumably set for a 700 wheel or 26" wheel.
"It takes a genius to spot the obvious" - my old physics master.
I don't peddle bikes.
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Re: Semi-DIY dynamo light build

Post by edocaster »

Thanks for the suggestions. I'd seen the first video, but not the second.

Trying with one LED directly to the rectifier I get... up to 170mA. I'm beginning to think the problem may be in my rectifier which, helpfully, I've covered in epoxy and swaddled with shrink wrap. I guess I'll have to chuck that away and start with a new rectifier. Do my results seem consistent with wiring the rectifier incorrectly? I'm sure I'd checked a few times, but if, say, I'd got one Schottky the wrong way around...presumably that would make a half-wave rectifier only? And my multimeter only measures DC current, so would a reading of one-third of what I'd expect tally with that?
Brucey
Posts: 44705
Joined: 4 Jan 2012, 6:25pm

Re: Semi-DIY dynamo light build

Post by Brucey »

yeah, it sounds like you have built a half-wave rectifier there. I was somewhat suspicious of this when you reported the flashing before; it sounded a lot worse than normal the way you described it.

If it is really wrong, you may have made the rectifier a dead short on one polarity; this will make the whole affair rather draggy and may expain the current reading you obtain; even this is lower than you might expect.

cheers
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Re: Semi-DIY dynamo light build

Post by edocaster »

OK, here's the update - and it's an annoying one.

So I picked apart the switch with the rectifier in it. The path of the diodes is a mess, but (after stripping away umpteen lawyers of heatshrink) it looks like the diodes are oriented in the right direction. Then I start probing with my multimeter to look for conductivity, and get paths which shouldn't be there.

Just to be clear - for diodes there should be low resistance in one direction, and extremely high resistance in the other, right?

Well, I'm getting two sides of a diode conducting in both directions with equally low resistance.

Curiosity piqued, I check out the remaining 5 diodes I bought at the same time which I hadn't already used. And, sure enough, one out of the five conducts freely in both directions!

I'm massively peeved, both at the eBay seller/manufacturer, and myself for not checking beforehand. It probably would have been easier to buy a pre-packaged rectifier than relying on four diodes. I'm now trying to work out whether this ended up damaging anything in the circuit. I'm guessing if it's just one dud diode, no harm, but if it was two or more, that could be a different story?
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Re: Semi-DIY dynamo light build

Post by edocaster »

Looks like there were two busted diodes in the circuit, both at opposite ends of the diode bridge. So, as I understand it, on one half of the cycle the current flowed through the pair of non-functioning diodes, then to the rest of the circuit as per usual, and then on the other half of the cycle the current went through the working diodes, and then straight back to the hub through the other two diodes. WIth any luck, that was the path of least resistance (except for my legs!), so hopefully the rest of the circuit was spared.
MikeF
Posts: 4347
Joined: 11 Nov 2012, 9:24am
Location: On the borders of the four South East Counties

Re: Semi-DIY dynamo light build

Post by MikeF »

Sounds as though you should discard all of those diodes. Even those that appear to be working as a blocking diode may not be working correctly even in the forward direction.

I'm not convinced that schottky diodes are needed at all in the rectifier part. What concerns me is that 1N5819 diodes have a reverse maximum voltage of 40v which is lower than could be produced by a dyno hub. I cannot see why, for example, the common 1N4003 - 1N4007 series can't be used, or any small bridge rectifier with a 100+ reverse volts and a 1A forward current ratings. I know that the forward voltage drop is slightly more than with a schottky, but it's only 0.4 volts at 1amp rating. Having enough volts is not a problem at this point in the circuit as far as I can see and frequency is not an issue either. Maybe I've missed something as I haven't carried out any practical tests?
"It takes a genius to spot the obvious" - my old physics master.
I don't peddle bikes.
Brucey
Posts: 44705
Joined: 4 Jan 2012, 6:25pm

Re: Semi-DIY dynamo light build

Post by Brucey »

it is easy enough to test the diodes (in both directions) using a battery and a load (a bulb or something that will draw ~0.5A). I think that (the right) Schottky diodes are worth persevering with because you could easily lose another 0.25W in the rectifier vs a standard silicon one. This would be enough to run a very effective rear light.

Schottky diodes are strange wee beasties; the forward voltage drop varies greatly with the exact diode design (but you can measure it under load in the test I suggest above). The other peculiarity is the reverse leakage current; this can be high enough that a multimeter can be 'fooled' into giving a reading that would suggest good conduction in both directions. Hence to be sure of them a proper test is required.

If you are very fussy you can select those diodes which have the lowest forward voltage; they will vary even within a single batch.

cheers
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~
edocaster
Posts: 475
Joined: 10 Apr 2013, 10:43pm

Re: Semi-DIY dynamo light build

Post by edocaster »

OK - thanks for all the excellent advice!

I double checked the remaining five diodes, and the one which the multimeter indicated was borked also behaved the same way in a real circuit. Back in the bin.

I then cobbled together a bridge rectifier with the other diodes and a smoothing cap (this time 2200uF). Also chucked in a 10k bleeding resistor across this, to hopefully lessen the chance of any capacitor 'accidents' frying the LEDs when testing - will cost me 1 or 2 mA, but better safe than sorry. While I was going to order an integrated bridge rectifier (largely as it's easier to solder) as I've had decent results with one of those on a USB charger, I figured I had nothing to lose with the diodes I had.

The result for one LED: 500mA

And then the whole circuit (two LEDs and standlight): 490mA just at the limit of how fast I could spin the wheel directly by hand (about 16km/h). I'm pretty satisfied now, and the brightness was far greater than before! Standlight still works as it should. I'm less interested in further efficiency savings now, compared to finding a good enclosure and robust cable solution. Will probably have to resort to a small project box, thanks to the 2200uF cap.
MikeF
Posts: 4347
Joined: 11 Nov 2012, 9:24am
Location: On the borders of the four South East Counties

Re: Semi-DIY dynamo light build

Post by MikeF »

Well done. Glad you got it working. Your current of around 500mA seems to agree with what others find for a "simple" connection to Shimano hubs, and as mentioned previously this ties in with the 6V 3W spec. The voltage, however, depends on the circuit connected to it once the wheel has reached a certain speed. Do you still find the light with the standby connection lights first?
"It takes a genius to spot the obvious" - my old physics master.
I don't peddle bikes.
User avatar
[XAP]Bob
Posts: 19801
Joined: 26 Sep 2008, 4:12pm

Re: Semi-DIY dynamo light build

Post by [XAP]Bob »

edocaster wrote:OK - thanks for all the excellent advice!

I double checked the remaining five diodes, and the one which the multimeter indicated was borked also behaved the same way in a real circuit. Back in the bin.

I then cobbled together a bridge rectifier with the other diodes and a smoothing cap (this time 2200uF). Also chucked in a 10k bleeding resistor across this, to hopefully lessen the chance of any capacitor 'accidents' frying the LEDs when testing - will cost me 1 or 2 mA, but better safe than sorry. While I was going to order an integrated bridge rectifier (largely as it's easier to solder) as I've had decent results with one of those on a USB charger, I figured I had nothing to lose with the diodes I had.

The result for one LED: 500mA

And then the whole circuit (two LEDs and standlight): 490mA just at the limit of how fast I could spin the wheel directly by hand (about 16km/h). I'm pretty satisfied now, and the brightness was far greater than before! Standlight still works as it should. I'm less interested in further efficiency savings now, compared to finding a good enclosure and robust cable solution. Will probably have to resort to a small project box, thanks to the 2200uF cap.


You could leave the lights in their enclosure and run a simple pair of wires to them from an external electronics box. I have my box (not quite the same) mounted on the dynamo fixing (hub dynamo doesn't need it)...
A shortcut has to be a challenge, otherwise it would just be the way. No situation is so dire that panic cannot make it worse.
There are two kinds of people in this world: those can extrapolate from incomplete data.
Post Reply