5x Cree XM-L dynamo light (in progress)

Hello all!

I mentioned in earlier threads that I would detail my build of a dynamo light using Pilom circuit number 9 (http://pilom.com/BicycleElectronics/DynamoCircuits.htm#DualMode) and an existing light head unit bought from eBay.

It's nowhere near ready, but the purpose of starting this thread is to ask a couple of questions (at the bottom).

My reason for attempting the light was the challenge (I'm an electronics noob), and I also had great results with a simple (but so far very reliable) 2x XM-L dynamo light based on a Solarstorm X2.

Here's the unmodded light head:


Here's the circuit. I read about how some people found voltage doubler mode redundant, but I plan to use it most of the time, hence the high LED count:


I've mostly finished with the breakout box, which was laid out like this:


A quick test connected to my dynamo wheel (driven by another bike wheel) showed the circuit works almost exactly as the Pilom site predicts. This was partly the goal for me, as I saw at least a couple of threads in bike forums where people doubted a dynamo could produce such output. I went for conservative component values, but am still getting a peak of 7.9W! By the Cree charts that's over 1000 lumens on the bench! :twothumbs



So far, it's looking really promising. Now the questions:

- Will my standlight circuit work? I've not used the last two LEDs in the string, but rather third and fourth. Does it matter (i.e. that they are one more LED raised from the ground rail), the circuit path should be to the other supercap terminal and not to ground, right?

- I found a CPF thread (this one, I think: http://www.candlepowerforums.com/vb...ht-questions&p=2168455&viewfull=1#post2168455) where it states than on switching from voltage doubler to full-wave rectifier the voltage doubler caps may have to discharge a lot into the LED string, causing a current surge up to 1.5A, even with a decent-sized smoothing cap in the way. How will this affect my standlight circuit? I've looked at the datasheet Vf curves for the schottky diode before the supercap and the Cree XM-L - is it just a case of making sure the two XM-L Vf minus the diode Vf stay below 5.5V (i.e. perhaps by using two schottky diodes in series)?

Thanks in advance!

regarding the standlight.... it seems like it should work fine. Nothing jumps out at me that looks like a problem.

regarding the other stuff... beats me. I'd have to spend some time looking at the circuit and trying to remember how it was supposed to work. Someone must have built it and made some measurements by now, right?? What does Martin say?

Looking again at the thread about current surges, that appears to be using Martin's autoswitching circuit, which has 4 capacitors contributing to the boost in full-wave rectifier mode. On switching, I think 2 of them are re-used for voltage doubling. I think it's at that point that the surge may occur if they accumulated a high voltage when running in full-wave rectifier mode.

The circuit I used keeps the boost and voltage doubler capacitors separate, so hopefully this specific problem won't occur...

Edocaster said:

Looking again at the thread about current surges, that appears to be using Martin's autoswitching circuit, which has 4 capacitors contributing to the boost in full-wave rectifier mode. On switching, I think 2 of them are re-used for voltage doubling. I think it's at that point that the surge may occur if they accumulated a high voltage when running in full-wave rectifier mode.

The circuit I used keeps the boost and voltage doubler capacitors separate, so hopefully this specific problem won't occur...

Click to expand...

I ran C10 for a winter with 5 LEDs. No problems.

Well, one problem, got high beamed a lot from cars :twothumbs

Great to see someone keeping dynamo light creativity going! So have you measured the voltage drop across the standlight part of the circuit - leds #3 & #4? Will it be approx 6.2V? I'm keen to see the outcomes...

Savvas

Znomit: Thanks - that's great news! I plan to run this largely in low mode, and will tame it in high by either spreading part of the beam with a diffuser, or using the innate non-total-grippiness of the eBay light mount to manually dip the light as required.

Savvas: Voltage across the two LEDs is approx 5.71V in full-wave rectifier mode, and 5.54V in voltage doubler mode. So, with one schottky I should get about 5.45V on the supercap (or 5.3V if using doubler mode). My worry earlier on was that the surge on switching would overvolt the supercap, but by trying to build in a safety margin by using two schottkys (schottkies?) the drawback would be lower charge level in day to day usage (especially in voltage doubler mode).

I have four supercaps ready, and will use two for 3F capacitance. The other two are spares if I smoke the first two...

Quick update, I finished the head unit today, pushing the 1800uF smoothing cap, two 1.5F parallel standlight caps and the standlight circuitry in the back. It was an absolute squeeze, and the one major disadvantage of adapting a light - i.e. being constrained by someone else's design. I'm hoping the electrical tape holds these disparate parts together well, not that there's much room for anything to move.

Testing at handspun speeds (about 5-10km/h) I was surprised that LEDs 1, 2 and 5 (from my original circuit diagram) lit easily, and glowed afterwards. Zilch from my standlight (LED 4) and the LED (LED 3) helping to support the standlight caps! I started to worry I'd screwed up the assembly. But I held out a hope that this was just the caps charging from scratch, so kept spinning. Then, after about 20 seconds, the standlight (LED 4) came on. Still nothing from LED 3.

Finally, after about two minutes of spinning, LED 3 came on. But only if I spun hard, and it would peter out first. Meanwhile, LED 4 was doing its standlight thing admirably.

I kind of worked out what was going on: I think the 'glow' from LEDs 1, 2 and 5 are from the smoothing cap, which also gets the voltage doubler caps feeding into it. While the standlight supercaps were empty, LEDs 3 and 4 wouldn't come on until the supercaps filled up a bit and the voltage rose across them.

As for LED 3 coming on last, and being first to fizzle out... I reckon that once the voltage across the string drops below what is required for the whole string, the current takes the easier path via the schottky and 68 ohm resistor instead of LED 3. The smoothing and doubler caps give this easier path quite some life even at very low speed.

While the effect of LED 3 coming on last is somewhat ugly, it's irrelevant at speed. I think now I just need to focus on the mechanical side of attaching the breakout box reliably.

Here's a picture of the light head just before shoving the components into it:

All my measurements of space in the back were fairly rough, so I wasn't certain the 1800uF smoothing cap could fit. I actually wasted the first smoothing cap cutting the leads too short.

The weak mechanical link is probably the standlight caps, for which I'm relying on tape, a blob of Shoe Goo and copious solder to keep the legs together. The final touch was some foam padding, more tape (far more than the above picture), putting in some heatsink compound and sealing the cable entry with silicone sealant.

Over the weekend I took the light out for a spin - the days are long now so I had to wait almost until midnight. All I can say is... OMG this thing is bright! Hell, I must have been one of the brightest things on the road, and I had to angle it down quite a bit when testing full-wave rectifier mode. I've not tested them side-by-side, but it may even be brighter than my Magicshine-style Cree XM-L 2s2p battery light, albeit with less of a hotspot. Switching at around 25km/h, the difference between modes is noticeable.

The beam isn't perfect - it's a bit too wide, but nothing I can't live with. I'll make the above horizon glare manageable with diffuser lenses or just angle the thing down.

The standlight circuit seems to charge just fine at about the peak riding speed for whichever mode the light is in, and the standlight performance is akin to a standard AA bike blinky. When the standlight caps are full, losses in the circuit beyond my initial measurements should only be via current flowing in the 68 ohm resistor and the internal resistance of the standlight caps as they are topped up. Applying a 0.7 efficiency factor to account for that and optics losses, I'd estimate the light pushes 700 lumens on high and 350 lumens on low.

More testing to come...

cool!! always good to see a project come together!

The mechanical packaging of a light can be quite a challenge to do well, even if it doesn't have the glamor of the fancy electronics.

I think the amount of light should be plenty. Once you hit a certain level, all the extra light does is degrade your night vision. The beam angle is something you may want to experiment with in the future. I've ended up with some pretty narrow beams after starting with wider beams. The wider beams just put too much light on the road near me, degrading my night vision and making it harder to see down the road. YVMV (your vision may vary)

Also good to hear that the standlight is working well. That might be one area where commercial bike lights don't compare with homemade lights.

So... beamshots or some footage of the light?? sort of like this??
https://www.flickr.com/photos/kurtsj00/5920297177/in/set-72157621965148305

Not many opportunities to really use the light, as the days are so long. Here's some footage in a fairly constricted space: http://youtu.be/4vFBrJcPb40

It doesn't really show full-wave rectifier mode (which was on after the 1:04 mark) in any meaningful way, as at these speeds it's either less than or, at best, equal to voltage doubler mode. Still, you can see the standlight. I'll work on adding some form of horizontal diffuser later.

Camera settings were ISO1600, 1/30 shutter speed, and f1.7 lens.

Maybe I'll find a way for a higher speed beamshot later.

looks okay from here. I think there are major challenges in trying to demonstrate a headlight on video. Maybe the best method would be to do a side by side comparison with some known standard, but we are a long way from having a standard.

Have fun with the light, and start counting the days until we go back to standard time and you'll have plenty of darkness to enjoy the light.

I took some beamshots using the MTBR forums 'original' beamshot standard settings of ISO100, f4.0, 6 seconds. The comparison puts the DIY dynamo light against my eBay 'Magicshine-style' single Cree XM-L T6 battery light - it's a fairly generic light which thousands of people have, although I'm sure consistency varies hugely, so some people will have brighter or dimmer versions, etc. I've seen measurements for these lights asserting between 600 and 800 lumens.

6 seconds is a long time! So these shots look way brighter than real life. They are, however, useful to compare against similar beamshots all over the web.

The dynamo light was driven at roughly 25-30km/h, and the battery light had its 2s2p battery fresh off the charger. The wall is roughly 8m from the camera.

First up, the 'Magicshine-style' battery light on high:



Then medium:


And low:


Now, the DIY 5x XM-L dynamo light on full-wave rectifier:


And voltage doubler:


...I think the dynamo light on FWR has the edge on the battery light. In truth, the battery light's hotspot is brighter (and tighter), but it's hard to see in these pictures as both lights overexpose there. But, overall, the dynamo light has far more spill and overall I think it is outputting more light. VD mode is somewhere in between the medium and low modes of the battery light - which is pretty much perfect for everyday usage.

The Magicshine clone was the brightest light I had, but now it looks like this dynamo light is the brightest! I'm pleased with how this has turned out (another thanks to Martin for maintaining the Pilom website, and the folks here on CPF for coming up with ideas such as the standlight circuit!) - I can't really ask for more (not yet anyway... :devil: ).