Ultimate dynamo light - make a wish

[Peufeu]

Yo guys.

I'm fed up with my battery powered lights : the battery is always dead at the wrong time, it's heavy (especially the helmet light one)... I want a light that's always ready no matter what, even below 0°C when batteries give up, or after a few days of camping with no power to charge anything.

My GF got a new bike with a Shimano Hub Dynamo, and a really crappy front white LED which puts out a tiny dim patch of "light" on the road ahead. I replaced that shame with a quick fix from the parts bin (a Seoul P4) which gives a nice usable flood, but of course when stopped she's in the dark. And now I want a dynamo too.

So I've taken on the project of designing the ultimate dynamo light for the bikes... ok ok that may seem a bit overambitious, but why not ?

Anyway I'd like you guys to give me your wishlist so I can make sure I don't forget anything. When it's done, I'd like to make a kit or something, or at least share the design, cause that might be useful.

Here's my wishlist :

• front : powerful white headlight
  
  I've selected Cree XP-G LEDs, very efficient, I have some with optics, nice usable beams. I plan on using 3 LEDs, each at a slightly different orientation, for a better beam. For instance a flood pointing a bit down + a spot pointing almost straight, etc. This is a nice combination and it minimizes the amount of light lost in the sky or blinding drivers on the other side.
  
  
• rear : really visible red
  
  I've got a bag of hi-brightness red LEDs from nichia, they're really good. I don't want to use a 1W or 3W LED because it's harder to get a good beam, with many small LEDs I can aim them where needed (rear / sides etc).
  
  
• automatic operation
  
  When you ride the bike the lights light, when you stop they stop after a few minutes timeout. Only button necessary should be to force the light to stay on if needed.
  
  day/night sensor : during the day front & rear should flash which is more visible, at night, continuous of course. Auto-switching is needed for things like unlit curvy mountain tunnels, for instance, which are REALLY annoying especially when you hear a car coming behind you and your bike lights are off because it's midday.
  
  
• standlight / low power battery mode
  
  For stops, traffic lights, long climbs, setting up the tent, etc : at the very low speeds where the dynamo doesn't give power, you don't need to see very far, so no need for a huge amount of light, 1-2W are enough.
  
  
• 5V output
  
  GPS and phone need charging.
  
  
• High power / High efficiency
  
  Extract as much power from the dynamo as possible when needed, reduce drag as much as possible when no power is needed.
  
  
• Stop lights
  
  Dynamo provides an accurate speedometer so a feature like a car's stop lights comes for free.
  
  
• All-weather watertight
  
  Last year I came back home on the bike and the watter bottle was frozen solid. The NiMH froze about halfway through, so I had to do the rest of the trip in complete darkness on an unlit country road. This was a great feeling but not really safe. It should also work under rain, dust, mud, etc.
  

I think that sums up my wishes, see if you have more ! Now here's my current work in progress design :

The brain will be a Cypress PowerPSoC, I have an EVM with it on my desk, it's a nifty chip, quite suited for the job.

Dynamo can provide high power at high speed, this is important downhill. Fast downhill (40-50 km/h) can produce more than 10 watts (possibly 15 watts).

A dynamo is an AC generator with harmonics with a big inductor in series. If you add a resistive load it behaves like a current source. However we can do a lot better by using a more funky load impedance like a negative inductor in series with a resistor. Some use a cap as the negative inductor but it only works at a certain frequency. I will put the dynamo in the center of a mosfet H-bridge, measure the current and PWM the voltage across it according to a realtime algorithm which emulates u = Z . i. I created a model of the dynamo which seems to fit quite well, and simulation shows this approach has potential. It also does synchronous rectification. The dynamo inductance behaves kind of like a boost reg, so I get 15-30V out. I'm still polishing this part.

Compared to a simple rectifier, the amount of power extracted becomes usable at much lower speeds and at high speeds it's a powerhouse.

To put all this power to good use : 3 Cree XP-G LEDs. Since power is proportional to wind speed, no cooling problems !

The design of the LED driver is done, it's a simple hysteretic buck switching current source with short and open circuit protection. Good thing PowerPSoC has 4 channel hysteretic switchmode controller on-chip with assorted control DACs and PWM...

Battery : 2 LiIon 18650 cells, charged and used independently. I will also make a small helmet light (or just buy one) so I want to be able to charge the batteries on the bike.

Battery management is done, a bidirectional hysteretic switching converter pushes current into the battery (charge = buck) or from the battery to the main capacitors (discharge = boost). The uC has ADCs and stuff to check the battery voltage and manage the charge.

Anyway, whaddya think ?

 

[Steve K]

• front : powerful white headlight
  
  I've selected Cree XP-G LEDs, very efficient, I have some with optics, nice usable beams. I plan on using 3 LEDs, each at a slightly different orientation, for a better beam. For instance a flood pointing a bit down + a spot pointing almost straight, etc. This is a nice combination and it minimizes the amount of light lost in the sky or blinding drivers on the other side.

• 
  
  what sort of beam width can you get with an XP-G? I've been pretty happy with a Ledil Rocket smooth spot used with Cree XR-E's. Other lenses, like the CRS smooth spot, had a wider beam that put too much light on the road close to me, impacting my night vision and making it overall less effective.
  
  
  

  [*] automatic operation
  
  day/night sensor : during the day front & rear should flash which is more visible, at night, continuous of course. Auto-switching is needed for things like unlit curvy mountain tunnels, for instance, which are REALLY annoying especially when you hear a car coming behind you and your bike lights are off because it's midday.

  
  or just leave the lights on in the daytime? That would take away power from any battery charging that you were doing, of course.
  
  

  [*] 5V output
  
  GPS and phone need charging.

  
  this would need to be current limited, or uC controlled. You wouldn't want the battery charging power, plus headlight power, to exceed what the dynamo can provide.
  
  
  

  [*] Stop lights
  
  Dynamo provides an accurate speedometer so a feature like a car's stop lights comes for free.

  
  that would be pretty low on my list of priorities, but don't let me spoil your fun.
  
  

  [*] All-weather watertight

  
  absolutely!
  
  
  

  Dynamo can provide high power at high speed, this is important downhill. Fast downhill (40-50 km/h) can produce more than 10 watts (possibly 15 watts).

  
  at high speeds, I've seen over 100v (after rectification) out of my SON. Make sure your design can handle those sorts of voltages.
  with the proper choice of loads, you can get quite a bit of power from a SON. Whether you want to pedal hard enough to provide that much power is a separate question!
  
  

  A dynamo is an AC generator with harmonics with a big inductor in series. If you add a resistive load it behaves like a current source. However we can do a lot better by using a more funky load impedance like a negative inductor in series with a resistor. Some use a cap as the negative inductor but it only works at a certain frequency. I will put the dynamo in the center of a mosfet H-bridge, measure the current and PWM the voltage across it according to a realtime algorithm which emulates u = Z . i. I created a model of the dynamo which seems to fit quite well, and simulation shows this approach has potential. It also does synchronous rectification. The dynamo inductance behaves kind of like a boost reg, so I get 15-30V out. I'm still polishing this part.

  
  I would have to agree with the general spirit of this, but disagree with the literal meaning. Dynamos, despite what is repeated so often, do not behave like a current source. They are a speed dependant voltage source in series with a mostly inductive impedance whose magnitude varies with speed. Nick Ray, on the Bikecurrent list, did a decent job of mapping out the SON characteristics at various speeds with various resistive loads. This made it possible to locate the maximum power point at over the range of speeds tested.
  
  I've thought it would be interesting to direct the rectified dynamo output into a buck converter where a processor selected the converter's duty cycle, and therefore controlled the impedance presented to the dynamo. This would allow the dynamo to be operated at the maximum power point when needed.
  
  

  Battery : 2 LiIon 18650 cells, charged and used independently. I will also make a small helmet light (or just buy one) so I want to be able to charge the batteries on the bike.
  
  Battery management is done, a bidirectional hysteretic switching converter pushes current into the battery (charge = buck) or from the battery to the main capacitors (discharge = boost). The uC has ADCs and stuff to check the battery voltage and manage the charge.

  
  I'm curious as to what you do to manage battery charging over the full temperature range. I bike commute in temperatures ranging from 2F to 98F, and have had some troubles with battery charging in the winter. It was complicated by the fact that most or all of the commute took place in the dark and often at low speeds, so there was little spare power for battery charging. Combined with the fact that battery impedance goes up when it's 10F out, and the standlight quickly drained the battery.
  
  The system could have been improved with a proper max power point tracker, and a lower power standlight, but it was a tough scenario at best.
  
  

  Anyway, whaddya think ?

  
  I think it's got more bells and whistles that I have designed into my system (but I have my bell on the handlebar. No whistles, though ). I don't see any other design issues or missing features.
  
  Load management might be one area that could use reviewing or explaining. i.e. if you have a large load (full lights and charging dead batteries), how do you avoid pulling down the dynamo output? There needs to be a way to cut back the load if the dynamo can't support it.
  
  Somewhat along the same lines, which I mentioned earlier, can the circuitry tolerate the dynamo's open circuit voltage if the load is very light (blinking lights only and not charging any batteries)?
  
  Another issue, which really applies to any bike light, is reliability. This is as much an issue of construction techniques, component derating, packaging, connector choice, etc., as it is circuit design. I know that when I'm coming down a steep, rough hill in the rain at night, I really don't want my lights to fail!
  
  have fun!
  
  Steve K.

 

[BrianMc]

Food for thought (taken close to the MTBR beam shot specs of ISO 100, F4, 6 seconds):

This beam is from 3 XP-G R5s @ 1 A, using three Carclo 10417 ('narrow') 16.4 FWHM (full width half maximum rated with XP-E, so actually about 18-20 degrees FWHN with XP-Gs (from green up 12th fairway, signs every 25 m):



As shown this is too bright for night vision or oncoming motortists even if pivoted right and down. I use it at 0.5 A with a fresnel lens in front of the Carclos to bring it to about 12-15 degrees FWHM and aim it down. Oncoming motorists are less affected but I am still working on it, so I have no beam shots. About 100 feet, aimed any higher and you light up sky.

This is the high beam unit, also with triple XP-G R5s. It has some features you mentioned. It uses three aspheric lenses at different focal lengths to project the die image of light that doesn't hit the barrel holding the lens and some of the light that hits the barrel, softening the square projected beam. The bottom one is about 11-12 degrees FW (severe cutoff all four sides) the middle one is aimed a bit higher, and is about 10 dergees FW, and the one aimed furthest down the road is about 8 degree but that required the dome to be removed from the XP-G, which reduces total output, but since the de-domed output is narrower, a greater amount is die image and it is of course concentrated on a smaller area. My crude light meter and 'collection sphere' say the two lights (low beam without the fresnel) are almost the same output. I expect that is in part an artifact of the collection device but they are not a lot different or that would have been detected.



This light produces a bright, narrow, high throw beam, which into a lowering sun is easily seen at 1/4 mile even on half power. This is aimed down a bit, aimed with the cutoff at the horizon, it will go out to 150 m, but asa wekare more spread out beam. It is too narrow to be seen closer in from the oncoming lane than 200 feet or so. Unless aimed at the driver. It has stopped left hooks on four occasions and the lights have gained me proper stops from entering traffic and that was not the case before.

This shows the separate and integrated beams:


Uploaded with ImageShack.us

The tipped lower beam is in part from a tipped lens and part from not getting the screw holes mounting the 10 mm MCPCB with the XPG perfectly aligned with the other two sets. It helps on the road. The MR11 housing I used is very tight or I would have slid this beam to the right lining up the left side cut offs.

This light is too narrow for your purposes, though as a battery driven helmet mount, I think it would be very effective supplement. The principle is proven. Mounting the lenses further inside the focal point increases the size of the beams. So for instance a 10/14/18 degree FW triple should be possible. These are prototypes and I found the alignment suffers from shocks to the helmet light like hitting door frames (surprise, surprise! :sigh

So I would suggest you consider three separate heads that can be separately aimed or two, one having the two higher beams set and not movable, and the other able to fit either another aspheric or the Regina reflector (one of the narrowest for the XP-G) as an alternative close spot as it is about 6 degrees FWHM so should be maybe 10-20% at 12-15 degrees and would give decent spill and send light to oncoming cars or cars entering from intersections closer to you.

Maybe I have been lucky with my old Cygolite (now about 50 Whr 12 volt) NiMH. Rode it down to -5 *F, regularly for two winters under 0 *C /32 * F still got over 2 hours run time. Maybe its sheer mass and discharge kept if going and it was nearly done when I got back, but I would not know. Regardless, I would like to be rid of batteries, for all the usual reasons.

I suspect there would be a market for a battery version of the light if it improves on the Phillips headlight for road use.

 

[Peufeu]

what sort of beam width can you get with an XP-G?

With the same lens the XP-G gives a wider beam than Seoul, and with some lenses, better, with others, worse. I've bought an assortment of lenses. Up until low I have tested them for like 10 minutes so I can't give you precise details !...

Also where you mount the light has a lot of importance. The helmet gets you the most efficiency because you only send light where you're looking so less light is wasted in useless places. But a light on the helmet eliminates the ground shadows so you don't see potholes. A much lower position like on top of the front wheel (the standard position actually) creates big shadows on road bumps and holes, making them much more noticeable, so it is really nice to have. On the handlebars, that's a compromise between the two.

Also depending on the height you might choose different beam angles. With a low position you can use a tight beam to illuminate a narrow, but long portion of road ahead. On an upper position (handlebars), not so, if you point it down you'll get a bright spot on the ground and if you point it more or less horizontal you get good throw but you don't see your front wheel.

So I'll use a mix of lenses. On the helmet I want to keep it light so I'll just use a compromise between spot and flood. I'd like to find a flood with a bright center spot.

wider beam that put too much light on the road close to me, impacting my night vision and making it overall less effective.

That's why I'd like to mix lenses.

or just leave the lights on in the daytime? That would take away power from any battery charging that you were doing, of course.

Flashing gives you excellent visibility with 1/10th of the power...

this would need to be current limited, or uC controlled. You wouldn't want the battery charging power, plus headlight power, to exceed what the dynamo can provide.

Yes, in this case the dynamo voltage would collapse. Fortunately, with a bit of firmware this isn't hard to control.

at high speeds, I've seen over 100v (after rectification) out of my SON. Make sure your design can handle those sorts of voltages

It won't, the input voltage is clamped to about 30 volts max, but without loss. It just has to emulate the right load impedance. If this works it will only take the power that is needed.

Whether you want to pedal hard enough to provide that much power is a separate question!

I think using 5% of the muscle power for lights is OK. After all, I already chose to waste a lot more power by not riding a race bike !...

Dynamos, despite what is repeated so often, do not behave like a current source.

It's a simplification : if the output voltage is small relative to the open circuit voltage (because it's clamped by, say, a 3.5V LED + diodes) , then the inductor makes it behave as a current source. If not, then it's not a current source. What makes it "current source-ish" is that the frequency and open voltage are proportional to speed, and the internal impedance (L w) is also proportional to speed.

I've thought it would be interesting to direct the rectified dynamo output into a buck converter where a processor selected the converter's duty cycle, and therefore controlled the impedance presented to the dynamo. This would allow the dynamo to be operated at the maximum power point when needed.

Yes, but that system could only emulate highish resistive load impedances. That would not allow you to boost the output power at low speeds, which is the reason some use a big capacitor in series with the dynamo. Basically, when you put in series the dynamo inductance and a negative inductance you get direct access to the voltage source. It is a technique that is used to compensate AC power factor. A capacitor is an approximation of a negative inductance but only at one frequency, which is why the capacitor + dynamo has a peak in the power response. Emulating a load impedance with a switching converter is more flexible.

I'm curious as to what you do to manage battery charging over the full temperature range.

I'm curious too. The chip has a temp sensor, so we'll see. Nothing there yet. For your commute case recharging the battery at home seems unavoidable...

Load management might be one area that could use reviewing or explaining. i.e. if you have a large load (full lights and charging dead batteries), how do you avoid pulling down the dynamo output? There needs to be a way to cut back the load if the dynamo can't support it.

Somewhat along the same lines, which I mentioned earlier, can the circuitry tolerate the dynamo's open circuit voltage if the load is very light (blinking lights only and not charging any batteries)?

It's actually pretty simple. Picture a H-bridge motor driver with a big power supply cap. Replace the motor with the dynamo. Measure current. Have the switches switch at the right time so the dynamo sees the impedance you want. The voltage created on the supply capacitor gives you a good idea of the available power. If it rises above a set threshold (about 28V), reduce power taken from the dynamo (and drag). If it lowers, increase power taken. If it lowers too much, reduce the load power.

Another issue, which really applies to any bike light, is reliability. This is as much an issue of construction techniques, component derating, packaging, connector choice, etc., as it is circuit design. I know that when I'm coming down a steep, rough hill in the rain at night, I really don't want my lights to fail!

Definitely. I've spent a lot of time on protection. The current design can have a short between any dynamo wire, ground, and LED V+, or an intermittent open connection on any of those, or a loose battery, or a short on the +5V phone charger, and nothing bad should happen. Now, for the rest of the reliability equation, I'll need to be very careful on the mechanical design and assembly. Also, that's what the helmet light is for !...

It has stopped left hooks on four occasions and the forn lights have gained me proper stops from entering traffic and that was not the case before.

Yes, I noticed that too, my helmet light puts out about 6-800 lumens and that apparently is needed to get some respect on the road. It is also very effective on pedestrians.

So I would suggest you consider three separate heads

That's what I wanted to do and I'm happy to see that it works for you. Good !

 

[BrianMc]

BTW I am testing where there are few cyclists except occassional recreational ones in spandex (including me), families tooling along sub 5 mph in neighborhoods, or the occassional BMXer. So if I am getting respect in a place where bikes are not expected on main streets (I do use alternate routes if available), the lighting is working. Something like 1000 lumens in a tight beam flashing on the helmet seems to work when aimed right at the motorist.

You are correct about the light angles with different placement. A close spot is best lower, a thrower on the helmet, and the middle beams on the bars. If you need to drop load at low speeds, you drop the farthest middle light first, then the other, never the spot just drop its power. Assuming the helmet light is battery powered and not your high. Better redundancy that way should there be a failure in the dynamo system.

Looking forward to your efforts. I was thinking of redoing my lights as two single MR11s, now maybe I'll separate the low beam into three distinct units to make aiming easy and flexible...

 

[Peufeu]

Something like 1000 lumens in a tight beam flashing on the helmet seems to work when aimed right at the motorist.

Yeah, but since it's a bit violent I use it mostly at intersections, when the guy should yield, that makes sure he sees me. Also I've noticed bright clothing gets you more respect during the day.

A close spot is best lower, a thrower on the helmet, and the middle beams on the bars. If you need to drop load at low speeds, you drop the farthest middle light first, then the other, never the spot just drop its power. Assuming the helmet light is battery powered and not your high. Better redundancy that way should there be a failure in the dynamo system.

Originally I wanted to put everything on top of the front wheel, but after some tests it appears that the shadow of the wheel itself on the road is a bit annoying. So probably, as you say it is best to put the spot on the top of the wheel and the flood on the bars. Unfortunately my break/gear cables are white so I avoid putting lights on the bars, they make the cables annoyingly bright ar night.

The helmet light is important as a backup (should the main light fail) and also to ensure that when you're behind a car, the guys on the other side of this car (who only see your head) do see you. I'd like a retro-reflective hlemet with a red blinkie at the rear, but couldn't find any readymade.

I'm still designing the electronics, and perhaps dropping the PowerPSOC in favour of a more conventional uC like MSP430. We'll see.

 

[markus_i]

Yo guys.

....
front : powerful white headlight

Check the Philips LED bike light (see corresponding thread) for a light with a good LED/reflector combination. The usual LED reflectors/optics are next to useless for on-road cycling - both thanks to blinding the oncoming traffic and for wasting half of the light where you can't use it.

rear : really visible red

Area is your friend here - illumnated area. Brightness is secondary (except for fog lights). Again, there's a nice commercial product from busch&müller, the Toplight Line, which uses a specific line-generating optic system to create a large illuminated area (and I've seen this type of optics also at the tail end of some of the more expensive cars).

automatic operation

When you ride the bike the lights light, when you stop they stop after a few minutes timeout. Only button necessary should be to force the light to stay on if needed.

day/night sensor : during the day front & rear should flash which is more visible, at night, continuous of course. Auto-switching is needed for things like unlit curvy mountain tunnels, for instance, which are REALLY annoying especially when you hear a car coming behind you and your bike lights are off because it's midday.

Why? Just keep 'em running. LEDs won't wear out fast enough to notice. Ok, if you want to charge batteries, you'll want to cut down during daylight.
Oh, and forget about the flashing. It may be 'visible' in some sense, but won't help you on the road because it's also next to impossible to judge the distance/speed of a flashing object.

standlight / low power battery mode

For stops, traffic lights, long climbs, setting up the tent, etc : at the very low speeds where the dynamo doesn't give power, you don't need to see very far, so no need for a huge amount of light, 1-2W are enough.

Don't overdo it - 0.5W with a decent reflector should be enough. Although for setting up a tent, I'd prefer a headlight.

5V output

GPS and phone need charging.

Nice to have - but not really necessary. But easy to implement (electrically - I'm not talking about waterptoof connectors...), so why not.

High power / High efficiency

Extract as much power from the dynamo as possible when needed, reduce drag as much as possible when no power is needed.

Not quite. Extracting as much power as possible might be too much - I'd put in some (speed dependent) upper limit.

Stop lights

Dynamo provides an accurate speedometer so a feature like a car's stop lights comes for free.

Not quite. Stop lights should come on when the brakes are operated, not when the speed changes. You may operate the brakes without changing speed (e.g. going downhill), you may change speed without braking (going uphill) - and you might not want to add a few tenths of seconds delay to a brake light (between gripping the brake lever and actually slowing down) when you _really_ need it. Using the dynamo frequency would be the easier way, but just because you've found a solution doesn't mean that you've solved the problem.

All-weather watertight

Forget it. That means basically vacuum-sealed, or you _will_ have condensation inside. Providing some holes for the water to drain out is at least as effective.

Bye
Markus

 

[Peufeu]

The optics in this philips look pretty good, but the price is just lol.

Anyway I just checked the datasheets and the XP-G and Seouls have about exactly the same light distribution versus angle. What makes the difference in beam quality is that the XP-G is much closer to the PCB than the Seoul : using the same lens and holder, it is not placed in the right spot. So, it should be possible to correct this by removing a bit of plastic from the lens holder to get the lens closer to the LED. I've tried this and it seems to work...

Note that using a long-throw light on the bike itself is a bit pointless : since a bike leans around corners, at speed you only turn the handlebars slightly, so the light always points to the outside of the curve, not where you're actually going. That's why a helmet light is really useful (and besides, you can aim it away from people's eyes if needed).

Area is your friend here

For the rear, I was thinking about a transparent plastic box, sanded to make a nice diffuser.

 

[Peufeu]

I am also wondering about something -- wether I should split the circuit in two or not.

Option 1 : all in the same box, probably located on the front of the diagonal tube of the bike, close to the dynamo and LED.
- cons : less room for batteries (2 cells only), not modular
- pros : small, light, wire between front and rear is low-current thus small

Option 2 : put the dynamo management and white led driver in a very small enclosure located somewhere at the front (top of the wheel with the front light, handlebars...) and put the batteries at the rear, in a larger box, in front of the rack or under the saddle.
- cons : two enclosures to design (but the front one doesn't need to be opened, so it can easily be made waterproof), wire between front and rear needs to be a bit bigger (nothing huge though)
- pros : more modular, more batteries and battery types possible.

 

[Steve K]

a good question...

For me, the issues are always (1) where to mount the box(es) of electronics/batteries and (2) how to minimize the wiring.

My current preference is to minimize the electronics and batteries to allow them to be included in the headlight and/or taillight assembly. This means that you don't need any new mounting points on the bike, and wiring is minimized.

If that's not possible, I've used the box/package as part of the bracket or fixture used to mount the headlight. It might require a sturdy fitting on the bike, though.

The biggest factor for you might be the batteries. Seems like they would be the largest and heaviest items, requiring the most attention for packaging. Due to the thermal issues of charging, I would assume they need to be mounted with the charging electronics, and maybe tied in with the dynamo controller? Go with whatever minimizes wiring, since wire problems are probably the biggest source of failures in a typical dynamo/LED system.

how big are the batteries, and how big will this whole light system be?

regards,
Steve K.

 

[Peufeu]

> how big are the batteries, and how big will this whole light system be?

It should be able to get you home always, so I set the battery design target to a 4 hour climb at very low speed, not using the dynamo. So, one of those :

http://www.batteryspace.com/lgli-io...7v2200mah814whorequivalentkorea18650cell.aspx

gives 8 Wh, ie 2W for 4 hours, or 1W for 8 hours, which is plenty. And it weights 47 grams. I think this cell size is the best energy / price / weight compromise.

Of course the beauty of the dynamo is that if you go fast, you get lots of power, and if you go very slow, you only need a watt or two from the batteries to get enough light. And if all the batteries are dead, you turn the bike around and go back home downhill with full lights. And even at 4 km/h you'd get a faint light, not really good, but better than nothing if that's all you got.

When the cell in the helmet lamp is drained, I will put it in the dynamo light for recharging. So, it would be better to be able to fit two cells, one charged, one charging. More is not really necessary. Two cells side by side also correspond more or less to the width of a bike's top tube. Charging some NiMH could be a bonus, but I can always use the 5V for that.

> how big are the batteries, and how big will this whole light system be?

So, we have two cells, with holder this means 2x 20mm x 20mm x 70mm.
The other large component is a big capacitor (25 x 35 mm) which is needed by the controller. Adding this we get a board about 10 cm x 4 cm which fits on the top tube and is large enough to hold all the electronics (all SMD on the other side). Or 6x7 cm. Or 2.5 * 18 cm but layout would be hell.

Wiring ... there's the two dynamo wires of course, the LEDs (in series) and a small wire going to the rear light.

I have to test and decide where I put the LEDs. Handlebars, on top of the front wheel, or both ?... that will depend on the beams. Maybe a combination of the two gives the best light. Putting the light low certainly makes the potholes easier to see.

I would like to avoid more clutter on the handlebars, too.

I can't put the batteries on the fastening point on top of the front wheel because that won't hold 100 grams safely (there is only 1 screw), so the top tube seems a good place (and allows short wiring).

So, with a 10 cm x 4 cm PCB, plus enclosure, where do I put it ?...

- on top of one of the bikes tube (top tube or the one below) : short wiring to the dynamo/LEDs, easy to fasten, easy to take the cover off to get at the batteries.

- on top of the handlebars : can contain some of the LEDs, easy to fasten, easily accessible buttons or display, adds more clutter

- below the handlebar T : sams as above, less clutter, much harder to swap batteries

- split (dynamo/LED driver at the front, same place as LEDs, batteries on the rear rack) : removing the weight of the batteries makes things a lot simpler, but complicates the wiring.

 

[Steve K]

So, with a 10 cm x 4 cm PCB, plus enclosure, where do I put it ?...

do you have a front rack? For a device that large, mounting to the rack would allow a solid mounting location that wasn't in the way of the handlebars, you, etc.

I'm thinking of mounting it near the top of the rack, inboard (so it doesn't interfere with panniers). The only downside I can think of right now is that you might get more spray from the front tire.

by the way.... did we discuss connectors? They seem to be a major source of trouble in all electronics, and bikes are no exception. I've got my own non-standard solution, and I think the issue has been discussed in many places over the years.

and how were you going to swap batteries in and out? A battery holder, I assume, plus some easy access panel/door? That raises issues in terms of sealing the enclosure, as well as the reliability of the battery holder contacts. Are there issues with battery charging too? Does "proper" Li-ion charging require a temp sensor attached to the battery itself? Personally, I keep life simple by sticking to Nicads and NiMH.

regards,
Steve K.

 

[Peufeu]

> do you have a front rack?

No, I don't have the mounting points on the bike. I've had handlebar baskets, but they all ended up breaking, since this method of one-point fastening cannot work reliably...

> For a device that large, mounting to the rack would allow a solid
> mounting location that wasn't in the way of the handlebars, you, etc.

Yeah but the front wheel would project a big shadow on the road.

> by the way.... did we discuss connectors?

Nope, and yeah, they're a problem. I was thinking about using the same connectors as ATX PC power supplies (on mobos), of course inside the enclosure.

> and how were you going to swap batteries in and out?

The case needs to be opened, no matter what... I have a few solutions in mind, I will have to test those in the shower.

> as well as the reliability of the battery holder contacts.

The holders are specified for vibration, but the datasheet doesn't mention the Gs (lol). I've designed the circuit to withstand intermittent open contacts.

> Does "proper" Li-ion charging require a temp sensor attached to the battery itself?

Charging is only allowed by specs in a certain temperature range (0-50°C). I guess getting a little hotter shouldn't be so bad. The on-chip uC temp sensor should do, I'll just put the uC under the batteries. Apart from that, they're really simple to charge, just don't get over the max voltage, or boom.

I've attached weights to various parts of my helmet : 1 x 18550 cell is the least noticeable. 3 AAs are a lot more annoying.

 

[BrianMc]

Fitting a front rack for electronics package with no fork rack drop out lugs:



1984 Schwinn World Sport fork with no lugs. Porteur style aluminum rack with legs carrying weight straight to front axle and using P-clamps to keep them in place for easy wheel removal and replacement. Fender lugs carry SKS quick detach fender mounts.

It will carry more weight than you want that high on the front of the bike.

If you are only going to carry electronics and not put 50-60 pounds up front, then the P-clamps can provide enough support and a second pair can be placed at the usual tourer rack boss height.

Hope this helps.

Brian.

 

[ZFlash]

Anyway, whaddya think ?

Thats the lightning system I've really dreamed of!!!!

I'm very much interested in your idea with the synchronous recitification AND the implemention of a variable capacitor.
I know, it's not the right explanation, my idea was to switch several capacitors via a µc...:candle::fail::candle:

My question is: I own a bottle-dynamo "Lightspin" whose frequency is much higher compared to a hub-dynamo.
Will your "powerhouse" :devil: be able to switch that much faster?

As for road-compatiple beam I'm very curious where to get those asymetrical lenses used in the supernova e3?
They seem to me the perfect solution, even if they are using a soul emitter.

Three of them plus three XP-G with the Regina-reflector for "high beam" is what I'm dreaming of.
Maybe switchable with one led only in situations wheres enough light to bee seen and load the two 18650er and/or to keep the drag less.
Thinking also about a self built Headlamp, XP-G, Regina, ak47 driver and a small heatsink which is 433MHz remote-controlled via the Handlebar-switch to toggle between high beam (ordered several, size as much as a stamp).

Please keep us informed with the progress of your project,
I'm highly interested to heat up my soldering iron!

 

[ohropax]

So, with a 10 cm x 4 cm PCB, plus enclosure, where do I put it ?...

I have been using a 3xCree setup with 5V regulated output (GPS) for quite some time now. But: Is there any particular reason for the PCB being that huge? Think about the Taskled LED drivers, all less than 1 inch diameter. Think about the 3xCree PCB with integrated driver/coil/etc from lux-rc.com, 20mm if i recall correctly. The PCB that I used for driving the LEDs and the GPS was no more than 15x35mm in size and fitted inside the fork shaft easily. Having the cables/connectors coming out of the aheadset top cover gave a really clean look. Furthermore there is still room left for 1-2 18650 Li-Io cells. If you could consider making the PCB as small as possible, maybe separating the big capacitor, it would be much more versatile and even allow stuffing it all inside the headtube. Cabling from front to rear is an orthogonal issue IMHO.

BTW: I created a hub generator LT-Spice model based on the findings of http://www.enhydralutris.de/Fahrrad/Modellierung_eines_Nabendynamos__Mathias_Magdowski_.pdf and have run a number of simulations covering 3-5 series connected LEDs with various capacities and various speeds. From the results i conclude that you cant extract more than 7.5 Watts at regular driving speeds and 10W at max at about 60km/h (40mph?). I did not simulate higher speeds because i don't think anyone drives that fast more than a few secs. I did all these simulations in order to find out whether i could design a dynamic maximum-power-point driven buck driver for a P7 LED. Well this doesn't mean i don't support your project, quite the contrary in fact, but don't be overly optimistic about the available power. Actually, i would love to stand corrected on that. You can see my results here in case you're interested.

Cheers,
Marcus

 

[BrianMc]

It looks like housekeeeping of the database ate my last posting.

Velo-Orange carries some nice bike racks that carry their weight to the axle. Some of these were made my the 'Rack Lady" who is half of Banjo Bikes. Links here.

Heavy Duty racks at Cetma in different widths. All less expensive thatn another bike assuming your bike's frame can handle touring loads.

Hope this helps.

 

[Peufeu]

> Is there any particular reason for the PCB being that huge?

Yeah, the PCB has 2x 18650 Li-Ion cell holders, and 1 big capacitor (since dynamo can go down to 10-20 Hz and the flickering is really annoying) which basically determine the size. Therefore, I didn't try to miniaturize stuff too much.

I've thought about stuffing everything into the head tube, but taking a dynamo-charged Li-Ion cell to put it in the helmet lamp would be ... cumbersome to say the least. Since I have one box to hold the stuff, I've settled on putting everything in it (except the LEDs of course).

> don't be overly optimistic about the available power.
> Actually, i would love to stand corrected on that

We'll see, I guess I'll have to build it and try it to really know !
In yout simulations, did you try capacitors in series with the dynamo ?

> housekeeeping of the database ate my last posting.

Mine too (I quite liked your rack ideas). Since I have an oxy torch I might try to make a welded steel version... I've never done welding with it though, only copper plumbing...

> All less expensive thatn another bike assuming your bike's frame can handle touring loads.

Well the beauty of these racks is that the load goes direct to the axle, not the fork !...

 

[BrianMc]

> > All less expensive than another bike assuming your bike's frame can handle touring loads.

Well the beauty of these racks is that the load goes direct to the axle, not the fork !...

I know. However, not all the load goes to the axle with my rack, though I cantilevered the design so that most of it does (>80% if I remember right). You should see the load up from move going over rough pavement or railroad tracks! The springiness of fatter 35 mm tires is needed with more than 30+ pounds or up front. Handling gets weird, too above 35 pounds.

The need for a stronger touring frame depends on what else you will carry on the back and your weight. Only you can say if such a front rack solves your cargo plans. I wouldn't want to do this with a light carbon fork, for example.

I am heavy and carry a lot of groceries with my bike. The frame is missing the rack lugs but is the basically a touring frame.

 

[ohropax]

In yout simulations, did you try capacitors in series with the dynamo ?

Of course, my simulations cover a 3-dimensional matrix of (speed,capacity,load) and the results exhibt the 'resonance' effects of the capacitor as expected.

The funny thing i discovered is: driving at regular speeds with a 3 or 4 LED setup fed by a low Vdrop Mosfet rectifier and capacitor of choice, you already have 90% of the theoritical power output doing its magic inside the LED. Thus in order to gain any increase in light output, any fancy electronic circuit has to have an efficiency of siginificantly over 90%. Just to put things into perspective...

regards,
Marcus