Nic Jones (![[info]](http://www.insanejournal.com/img/userinfo.gif){kind=small} nicwrites) wrote,@ 2008-07-23 22:21:00 |
 |
|||
|
|
|
|
|
|
|
Flash! Aaa-aaah!
So, riding a bicycle presents some challenges. The one that scares me the most is cars. To lower the risk I want to be easily seen. I wear a high visibility jacket and have a range of flashing lights. My headlight has been a bit anaemic, though, and as an electrical engineer I thought it was my duty to do something about this.
A look at wikipedia confirmed what I pretty much knew: LEDs have around the highest luminous efficacy for white light. Phillips led the way with a new breed of high intensity LEDs and they've been available at retail for a year or two now. There are a couple of other companies that make them too, including Cree and Nichia.
These high intensity LEDs can't really just be plugged in to a battery. They're devices that have to be current regulated and mounted on a heatsink.
I spent some time looking at torches, garden lighting, and catalogues for sources for the LEDs. Eventually I chose one of these ZD-0352 lamp modules. I thought they were $40 (misread the label on the shelf) but even at $60 I'm happy with it. The pattern of the light out of it suits a bike headlamp really well and I didn't have to worry about heatsink design or optics. The thing that really sold me was that the one on display was painfully bright to look at even from the far end of the store. 3 Cree XR-E 1W cool white LEDs add up to 270 lumens of light.
Inside the base of the module is a full wave bridge and a small SEPIC converter. The controller ship is marked "IP57" but google doesn't turn up any information about it. The module is meant to be a drop-in replacement for a 12V halogen but the converter pretty much prevents it from dimming. It doesn't start shining until about 9.5V and is at maximum intensity at about 12.5V. They're specified as "12V". I've driven them up to 15.7V without releasing the magic blue smoke but I wouldn't push it much further - the filter cap on the rectifier inside them is rated at 16V so that's probably the maximum.
Power draw of the light module
While wandering through Bunnings I saw this on the shelf for $3.47:
It came with two 20W halogens. It's really solid, too. I went back a week later and they were on the shelf for $34.70 so I think I did quite well, as it's a great housing for the light.
The bike light circuit. Click to enlarge
The circuit is the result of the random parts I had sitting around. It's pretty hacky but it's working well enough. The switching controller is an LM2576T-5.0 buck converter with 3A internal switch I'm using upside-down to make an inverting buck-boot converter. There's a 9kΩ resistor in the feedback line to change it from a 5V converter to a ~15V one.
If you do the calculations, it's really meant to have a 750µH inductor (the switcher is quite slow; only ~50kHz). I had a 680µH inductor in my parts bin but the controller wouldn't start. The next one down I found was a 400µH inductor and it works with that.
I'm using a 555 and a FET for the flash. I'd have liked to have wired the 555 into the ON/OFF pin of the LM2576 but because of the topology I'm using it in there was no obvious way to get it working.
Neither the controller or the FET need heatsinking, which makes things easier.
A rotary switch is used for power and bypassing the FET. There are probably more elegant ways of doing that but I was running of of room.
The other things that could certainly be improved upon is the fact the circuit will happily run the batteries into the ground. It's not good to do that to a NiMH battery pack and some form of under-voltage shutoff would be a really good idea for anyone who has stumbled on this design from the wild west of the internet.
While digging about for a knob for the switch I found an old laptop battery. I took the best 7 cells and the bimetallic overtemp/overcurrent cuttoff and made a battery pack. I had lathed the housing out for AAs but these cells are just slightly bigger than normal AAs so I need to do a little more work to make them fit.
I don't know their capacity but I'm guessing it's about 2000mAh. There are a couple of strategies for charging NiMH batteries but the easiest is to charge them at C/10 (200mA in this case) and you'll have a full charge in 15 hours, and you can't do too much damage if you leave them on the charger. So I built a little charging circuit and set it for 200mA.
The circuit is pretty straightforward: The voltage drop across a 3.3Ω wire-wound resistor is isolated and amplified using a dual op-amp, and it drives a BJT. It seems to work reasonably well, but needs a regulated voltage source. I'll probably add a linear regulator and stick it on some prototype board.
Also, CADSoft's Eagle is a usability nightmare.
So, riding a bicycle presents some challenges. The one that scares me the most is cars. To lower the risk I want to be easily seen. I wear a high visibility jacket and have a range of flashing lights. My headlight has been a bit anaemic, though, and as an electrical engineer I thought it was my duty to do something about this.
A look at wikipedia confirmed what I pretty much knew: LEDs have around the highest luminous efficacy for white light. Phillips led the way with a new breed of high intensity LEDs and they've been available at retail for a year or two now. There are a couple of other companies that make them too, including Cree and Nichia.
These high intensity LEDs can't really just be plugged in to a battery. They're devices that have to be current regulated and mounted on a heatsink.
I spent some time looking at torches, garden lighting, and catalogues for sources for the LEDs. Eventually I chose one of these ZD-0352 lamp modules. I thought they were $40 (misread the label on the shelf) but even at $60 I'm happy with it. The pattern of the light out of it suits a bike headlamp really well and I didn't have to worry about heatsink design or optics. The thing that really sold me was that the one on display was painfully bright to look at even from the far end of the store. 3 Cree XR-E 1W cool white LEDs add up to 270 lumens of light.
Inside the base of the module is a full wave bridge and a small SEPIC converter. The controller ship is marked "IP57" but google doesn't turn up any information about it. The module is meant to be a drop-in replacement for a 12V halogen but the converter pretty much prevents it from dimming. It doesn't start shining until about 9.5V and is at maximum intensity at about 12.5V. They're specified as "12V". I've driven them up to 15.7V without releasing the magic blue smoke but I wouldn't push it much further - the filter cap on the rectifier inside them is rated at 16V so that's probably the maximum.
Power draw of the light module
While wandering through Bunnings I saw this on the shelf for $3.47:
It came with two 20W halogens. It's really solid, too. I went back a week later and they were on the shelf for $34.70 so I think I did quite well, as it's a great housing for the light.
The bike light circuit. Click to enlarge
The circuit is the result of the random parts I had sitting around. It's pretty hacky but it's working well enough. The switching controller is an LM2576T-5.0 buck converter with 3A internal switch I'm using upside-down to make an inverting buck-boot converter. There's a 9kΩ resistor in the feedback line to change it from a 5V converter to a ~15V one.
If you do the calculations, it's really meant to have a 750µH inductor (the switcher is quite slow; only ~50kHz). I had a 680µH inductor in my parts bin but the controller wouldn't start. The next one down I found was a 400µH inductor and it works with that.
I'm using a 555 and a FET for the flash. I'd have liked to have wired the 555 into the ON/OFF pin of the LM2576 but because of the topology I'm using it in there was no obvious way to get it working.
Neither the controller or the FET need heatsinking, which makes things easier.
A rotary switch is used for power and bypassing the FET. There are probably more elegant ways of doing that but I was running of of room.
The other things that could certainly be improved upon is the fact the circuit will happily run the batteries into the ground. It's not good to do that to a NiMH battery pack and some form of under-voltage shutoff would be a really good idea for anyone who has stumbled on this design from the wild west of the internet.
While digging about for a knob for the switch I found an old laptop battery. I took the best 7 cells and the bimetallic overtemp/overcurrent cuttoff and made a battery pack. I had lathed the housing out for AAs but these cells are just slightly bigger than normal AAs so I need to do a little more work to make them fit.
I don't know their capacity but I'm guessing it's about 2000mAh. There are a couple of strategies for charging NiMH batteries but the easiest is to charge them at C/10 (200mA in this case) and you'll have a full charge in 15 hours, and you can't do too much damage if you leave them on the charger. So I built a little charging circuit and set it for 200mA.
The circuit is pretty straightforward: The voltage drop across a 3.3Ω wire-wound resistor is isolated and amplified using a dual op-amp, and it drives a BJT. It seems to work reasonably well, but needs a regulated voltage source. I'll probably add a linear regulator and stick it on some prototype board.
Also, CADSoft's Eagle is a usability nightmare.