HID bicycle lights

[jleaman]

Scott why not try these get about 5-8 of them and make a nice low current light weight one but really bright.


http://www.besthongkong.com/product_info.php?cPath=3&products_id=34
Jase

 

[Variac]

I think mainly because you can get much brighter ones than those you listed. It is a lot easier to work with one LED and one reflector, etc. than 5. Cateye used to make a light with a cluster of 5 which was prettty bulky and heavy, but now are using one much more powerful one. The big ones I mentioned , like the are now up to about 3 watts


BUT your place is a good resource, they also sell the big ones: http://www.besthongkong.com/newsdesk_info.php?newsdesk_id=47
http://www.besthongkong.com/product_info.php?products_id=171


I need to check prices, the places I mentioned also sell lenses and reflectors that fit the star LCD's.

Also, it is possible that the HID lamps and even the fluorescent lamps are actually more efficient than LCD's . People assume that LCD's are the most efficient, and I'm not sure its true...

 

[Eva]

Eva,

If You try to google "royer oscillator" You will find a lot like this attachment.
Regards
Heinz!

I don't know where you got that circuit, but it won't work at all for normal low-cost compact fluorescent lamps in the power range that we are discussing.

Note that a normal 11W compact fluorescent lamp like mine requires proper filament heating and 400V for starting, and 100V 100mA for operation. There is no way in which such a circuit could provide 100mA to the lamp as it has a 562 ohm resistor and a 50K pot in series with it that would burn instantaneously! Also, there is no way in which two TO-92 transistors rated at 2A max could provide 11W lamp power with input voltages as low as 6V.

Actually, as it has a 33pF 3KV capacitor in series with the lamp, it seems like a low power neon tube dimmer (5W or so, like the ones used in computer modding).

 

[powerbecker]

"I don't know where you got that circuit, "

It´s from AN 45 LinearTechnology (1991), You can load it there.

"but it won't work at all for normal low-cost compact fluorescent lamps in the power range that we are discussing."

Of course it´s not a recipe, interesting is the topology.
It deliver very easy high hf-voltages. For instance I build with this
topology a source for hf-surgery : 500kHz, low distorsion sinus, 150 W @ 93% effectivity.

The actual schema is for 5W lamps, worked @ 60kHz and deliver 1400Vpp.

"Note that a normal 11W compact fluorescent lamp like mine requires proper filament heating and 400V for starting, and 100V 100mA for operation. There is no way in which such a circuit could provide 100mA to the lamp as it has a 562 ohm resistor and a 50K pot in series with it that would burn instantaneously! Also, there is no way in which two TO-92 transistors rated at 2A max could provide 11W lamp power with input voltages as low as 6V."

If You use "real" HIGH Voltage ccs You don´t neat the heaters!!

"Actually, as it has a 33pF 3KV capacitor in series with the lamp, it seems like a low power neon tube dimmer (5W or so, like the ones used in computer modding).

Yes, behind many (big) screens there are some of this lamps.

You can use also a little bit other schema to came with high effectivity to high voltages : Output current from a power-comparator drives via transformer a L-C series connection, feed back the input current with a current transformer to the comparator to get nearly zero switching losses.

The biggest oscillator I had to build with this deliver 16kW @ 100kHz to power a 4MW LC series connection (8000Vrms/500Arms), witch produce a strong high frequency magnetic field.
Regards
Heinz!

 

[Eva]

If You use "real" HIGH Voltage ccs You don´t neat the heaters!!

There is no way in which you could make a conventional fluorescent lamp light at lower than maximum intensities without heating the filaments. It will just stop lighting as soon as you try to reduce the duty cycle. Also, since conventional fluorescent lamps have a strong negative impedance, they can't be dimmed by trying to reduce the input current to the transformer through a buck regulator (the voltage will increase as current is reduced, thus keeping power constant), dimming must be achieved by duty cycle control. Furthermore, skipping filament heating for conventional lamps is said to reduce its useful life by an order of magnitude.

This circuit is much better suited for conventional fluorescent lamps:

An externally hosted image should be here but it was not working when we last tested it.

In that circuit, dimming by duty cycle control is achieved by increasing the drive frequency. That's not due to the inductor, but because the capacitor in parallel with the lamp imposes a crossover time on each transition, during which the lamp conducts little or no current due to a lack of voltage (filament heating is required to keep the gas ionized during that period, otherwise the lamp won't conduct again in the next cycle). As the frequency is increased, the relationship between the time the lamp is conducting and the time the lamp is not conducting (crossover period) decreases, thus decreasing light output.

Also, this circuit has a very nice feature: The power applied to the lamp decreases linearly with increasing phase lag of the inductor current with respect to the drive voltage. This means that precise dimming is easily achievable for different lamps, different temperatures and different LC values just by adjusting drive frequency to get a fixed phase lag, thus estabilishing a phase-locked loop. The current phase lag is quite easy to detect by sensing the voltage drop across one of the switching MOSFETs during its conducting period and finding the zero-current crossing point (where Vds is exactly 0).

Anyway, if you just drive the MOSFETs straight from a SG3525 at a constant frequency close to the resonance of the LC network, the lamp will light nicely.

Even the following circuit produces excellent results despite its simplicity, altough filament current ends up being somewhat higher than required and it's not easy to reduce it.

An externally hosted image should be here but it was not working when we last tested it.

And the best of all, net light output from a fluorescent lamp is similar or better than from an HID lamp at similar power levels (altough light is much more concentrated in the latter case).

 

[jleaman]

id like to post that besthongkong.com was a rip off i got screwed over totally i waited 4 months and have not recieved a single product from them.

Im now using > http://www.lsdiodes.com



Scott's idea here is to use the same battery pack as what comes with his bike lighting system as it is but get better / more lighting and a longer lasting light run time. I think super bright white led's with a high cid will be the trick.

 

[powerbecker]

"There is no way in which you could make a conventional fluorescent lamp light at lower than maximum intensities without heating the filaments."

Well, Eva, I have no practically experience about that.
On the other way, with lamps described eg in AN45 it works.

In my limited understanding the physical process in all this gas-discharging lamps is the same. Of course the characteristics depends from pressure and kind of fillings, dimensions and temperature. And always helps heated electrodes to work or start with less voltage and also all represent after ignition a negative resistor.

"It will just stop lighting as soon as you try to reduce the duty cycle. Also, since conventional fluorescent lamps have a strong negative impedance, they can't be dimmed by trying to reduce the input current to the transformer through a buck regulator"

This seems clear for me, because this is not a real current source which will enforce the current, one can this also imagine as a cycle by cycle ignition.

I think it will be a interresting experiment with Your circuit to feed them from a (DC) currentsource, implied all parts are useful for a higher outputvoltage.
May be I get the time to try it in my company with a constant current poweramp which can deliver 800Vss DC to 100kHz.
But I had to wind a 1:2or3 step-up transformer and my free time is limited.
If You do some experiments care to Your probes, at higher frequencies they break down at much less voltages as, say they withstand @ 10kHz and less. This way once I killed a 1000:1 HV-probe!
Regards
Heinz

 

[SCD]

Thanks for chiming in Jason;

I have been reading this discussion with interest. I am not familiar with the florescent lighting as a portable helmet mounted system.
I have riden with folks that have the LED system. That fellow has had lots of trouble with his battery pack and I think the lights are not quite brite enough for me.
Other fellows have the HID systems that are quite incredible. One made by Niteride the other by Light and Motion. Both systems perform very well. The cost of these is way up there in the $700.00Cdn range. Well beyond my reach.
The HID sysytems are what inspired me to start this discussion. I am hoping that other riders will be interesed in developing / designing a DIY system that might bring the cost down a bit.

Now that daylight savings is here and daylength is increasing so fast right now the preasure is off a bit for this year. That allows lots of time for discussion and design.


Keep the ideas and thoughts coming. This is a ways off of audio but it is still DIY

Thanks for the help

 

[powerbecker]

Hello Eva and all,

I had to do my work, but this things seems to put me off:

By Zetex I found a lot of interesting information and links about ccfl and leds: http://www.zetex.com/
For ccfl it is very usefull to download AN65: 124 pages interesting stuff from
http://www.linear.com/designtools/app_notes/index.jsp

A part at the end I cut out to show it here. There they say, that the hcfl-lamps (hot cathode) work equal with their ccfl-circuits...without using the heater!
Regards
Heinz

 

[Variac]

My stupid friend has convinced me to try some brevets this year. The first official one in 300km. Even with summer light we will have some darkness!!!

He has the Princeton Tec 3 watt LED. It isn't quite bright enough, but I think that means a 5 watt might be.

OR:
A wide angle 3w on the bars paired with a 3w spot on my helmet. That would consume 6 watts which is low enough for decent battery life.

I'll report back!


Jason, Useful to know that that LED place you linked us to isn't reliable. Make sure to tell us what happens!

Mark

 

[jleaman]

Jason, Useful to know that that LED place you linked us to isn't reliable. Make sure to tell us what happens!

Mark

I'll keep us all posted on the new place. i was told from another member on here that the new company that i just put up for share's is really good. ii just did my order so far so good 3$ for shipping to canada WICKED deal the other company charged me 12$ us

 

[Eva]

powerbecker:

It's what I said previously. The lamps discussed in that paper have nothing to do with conventional fluorescent lamps when it comes to drive them. If you look at the diagrams, their circuit is used to drive tiny tubes drawing 5mA with working voltages as high as 1KV (neon like). Conventional fluorescent lamps draw much more current (100mA the smallest ones, even 6W) and use much lower working voltages (around 100V).

Also, I suppose that you have seen that the two automotive medium-size lighting systems shown were CCFL but required external heaters for proper starting in cold ambient conditions, causing 30% losses due to capacitive shunting. Hot catode fluorescent lamps doesn't sufer from any of these issues because they have built-in heating (note that the voltage is 10 times lower and the current is 10 times higher so shunting capacitive losses are negligible).

Anyway, somebody trying to build a DIY fluorescent system is going to have a trouble finding CCFL lamps and getting them at a reasonable price, so we should forget about these.

 

[Variac]

A project making a 3 watt light:

http://www.littlefishbicycles.com/bikelight/index.html

I'm using the same place for the LED's but getting 5 watt.

 

[suzyj]

A project making a 3 watt light:

http://www.littlefishbicycles.com/bikelight/index.html

I'm using the same place for the LED's but getting 5 watt.

I can verify (having used that light for a couple of years including two 400km brevets and a 600km brevet) that the light linked to is a good-un.

If I were building another today, I'd use the new Cree XR-E LED, which manages some 70lm/watt.

Cheers,

Suzy (slightly surprised at finding a cycling thread on DIYAudio - where's the framebuilding thread?)

 

[Variac]

YOW! I was this close to ordering the 5 watt luxeon!!


But this is amazing:

http://www.cree.com/press/press_detail.asp?i=1160427137863

State o' the art!!!!

Now I gotta think again.....





There are more than few bikers lurking in high places around here...

 

[bremen nacht]

YOW! I was this close to ordering the 5 watt luxeon!!


But this is amazing:

http://www.cree.com/press/press_detail.asp?i=1160427137863

State o' the art!!!!

Now I gotta think again.....





There are more than few bikers lurking in high places around here...

Well, low places...hello!


Are the Cree LEDs available yet?

Problem with chip LEDs is heatsinking them. I haven't found anyone that does alumin(i)um backed pcb in proto quantities, would be grateful for any leads.

I use a Luxeon 3W, not even at full power, for my unlit country-lane commute and it's fine. Most drivers dip their lights so I know they've seen me, and even if they don't the Luxeon is so bright that I'm not blinded after they pass.

These guys do really good prices on LEDs and lenses:

www.dotlight.de

 

[AndrewT]

Hi,

I haven't found anyone that does alumin(i)um backed pcb

why would you need aluminium for a heat sink, when copper is so much better. Just as copper is better at carrying current, so you only use aluminium when some other reason (characteristic/property) overides the better material's properties.

 

[bremen nacht]

Hi, why would you need aluminium for a heat sink, when copper is so much better. Just as copper is better at carrying current, so you only use aluminium when some other reason (characteristic/property) overides the better material's properties.

I was referring to aluminium backed pcb, which is a sandwich of copper, very thin FR4 then 1 or 2mm aluminium. I don't know whether anyone makes anything similar with copper.

http://www.clarydon.com/services/metal_backed_pcb.html

Luxeon use it for their mounted LEDs.

 

[djQUAN]

I have use a high power LED (a cheaper taiwan copy of a luxeon) and it is blindingly bright.

I'm using a Li-ION batt pack scavenged from a broken laptop and AN-003 from ESP (raised the current to 320mA) and it works quite well considering the price.


right now, I'm bench testing the Li-ion charger that I made since searching the net leads to charge controllers that always use a dedicated IC so I made one that uses an op amp, some transistors, passives and a 16V laptop charger.

 

[suzyj]

Problem with chip LEDs is heatsinking them. I haven't found anyone that does alumin(i)um backed pcb in proto quantities, would be grateful for any leads.

For Diy stuff, I'd just machine a copper block, bung some solder paste on it, and sweat the LED onto that. As long as the block has cutouts so that it doesn't contact the anode and cathode, it'll work very well. Thermal conductivity would be better than with an aluminium backed PCB too.

Of course, for quantity stuff, the PCB is the way to go, as it'll be lots cheaper.

Cheers,

Suzy