looking for a driver

[07accordEX]

what other drivers are there besides the shark and that KD one?
i need a cheap alternative to drive ~4 crees in the bike lights im making

 

[freedom2000]

HI,

Depending on what you are looking for, this link could help you.

You can pack these drivers in // to get the current you need.

JP

 

[07accordEX]

hmm so lets say i wire 2 of the 1.4a drivers with a 5 mode driver

then theoritically its putting out 3.8A at 950mA per led when 4 leds are wired in series

but they only work with 1 cell
is there a way to make it able to accept 4 cells?

 

[yellow]

they dont work with led in series.
The idea behind the linked driver pack was to get the current to have a P7 be powered suitably. Thats 4 led parallel.

With led in series You still need a boost driver and all the good ones are not cheap.
Another source besides from Sandwich Shoppe / SHARK are the drivers from taskled:
http://www.taskled.com/compare.html


PS: 4 cell setup (Li-Ion?) and 4 led in series wont work.
Try to come to an uneven number!
(2-3 cells/4 Led; 4 cells/5 led; 4 cells/3 led)

 

[freedom2000]

Hi,

Increasing the current delivered to a pack of leds in // is easy with AMC7135 boards --> just wire them in // and you add the currents.

Exemple : 2*1400mA AMC drivers in // will provide 2.8A
2.8A will be enough to light 4 Cree Q5 in //. Each led will get 700 mA which is really good for this led and won't get the die too hot (so preserving lifetime)

If you wire the leds in serie then you must add voltage.
Exemple : 2 Cree Q5 with a nominal VF of 3.7 V will need 2*3.7 V = 7.4V to be properly driven at 700 mA

And here is the problem... most components only handle 6 to 6.5V maximum. This is the case for AMC7135 and for µcontrollers...

As a consequence you are limited to his higher limit 6.5V when using these components... so the easiest solution (if not using expensive drivers) is to increase current... so to wire the leds in //

On my post above, I suggested that you could use 8 AMC7135 chips (or 2 1.4A boards of 4 chips each). This would be a cheap way to drive your 4 Cree Q5.
BUT these drivers have a drop out of 0.12V. So you need at least 3.7 + 0.12V = 3.82V which is the value of a fully charged Li Ion... So a single Li ion will be efficient ONLY 5 to 10 min... after the AMC7135 chip will be useless and more... will spoil 0.12V for nothing... and your leds will be direct driven with an unsufficient voltage (less than 3.7V) and so will get dim...

This why I have choosen this "strange" battery configuration 1 Liion + 1 NimH
a combination which gives always more than 3.7 + 0.12V... thus the AMC7135 will be active and the leds always driven at full power:candle:.
And a combination which always gives less than 6.5 V that is safe for µcontroller and AMC chips (a fully charged liion is 4.2V a fully charged NimH is 1.5V thus 5.7V...)

But read carefully my posts to understand all the dangers (for your cells only) when using this configuration with "unbalanced cells" (nimh discharged and Liion charged for exemple... or not using a protected Liion ...)

Well, driving leds at more than 2A is not that easy !

But it is so bright :wave:

JP

 

[freedom2000]

Hi again,

I didn't notice that you were building your own bike light... so you can do what you want

There is so another solution to drive your leds --> the LM317 linear driver
It is basically the same kind of driver than the AMC7135 but :
- with a higher drop out ... 1.2V ... ten times higher than the AMC7135 (not good)
- with a much higher range of voltage --> this is good

Here is an exemple of how to get 1A current regulated output :

A single 1.2W resistor and the LM317 that's it

With a single LM317 you can rise current up to 1.5A (with heat sinker of course) and voltage up to 25V

So everything is possible :
two leds in serie --> Vin must be higher than 2*3.7 + 1.5V = 8.9V (3 18650 cells in serie will be perfect.
To get 4 just add another LM317 +another resistor !

four leds in serie --> Vin > 4*3.7 + 1.5 = 16.3V --> 5 cells 18650

well my advice would be 2 leds in series and 2 other ones in another serie.
If you want you can add a 10 Ohm potentiomer in serie with the 1.2V resistor (use a 2 or 5W potentiometer) and you can dim the leds continuously from as low you want to full power.

JP

 

[07accordEX]

yeaaah that was what I had in mind also as i used the same thing in my lasers, but how does this compare to the a good buckboost driver?
would the output die down as the voltage of the batteries start to decrease?

 

[freedom2000]

these linear drivers work well (very well) provided that the voltage of your cells remains above the led VF + the drop out of the driver. If not these drivers will switch to direct mode and will not boost your voltage... and the led will get dim.

If you respect this single advice then your led will remain driven at full power until the cell voltage decreases to under 3V (for a Liion) or under 1V (for a NImH). So until your bat needs to be recharged.

All the combo that I have tried to explain above provide a voltage/current which is "too high" for your leds. The driver being there to regulate current to the nominal value of your led. All the voltage above VF + drop out will be dissipated into heat in the driver... So you will spoil a bit of power into heat in the driver but you will get a full power led to the end of lifetime of your cells.

To sumarize a bit. I propose to have extra voltage into your cells that will be dissipated into heat but will drive your led to exactly the current/voltage it needs.
The main advantage of these drivers is that the led is never over driven --> it has exactly the needed current provided that the cell has enough voltage in it.

The main disavantage is the relatively poor efficiency of these devices... See here to understand of bit more As you will see, to have a high effciency you need to be (and remain) as close as possible to the ideal conditions that is a voltage just above LedVF + driver dropout.
This is exactly what I have done with my combo 1 NimH + 1 Liion to drive a single Cree Q5 or a single SSC P7

JP

 

[VanIsleDSM]

I can build you a nice buck driver if you'd like. I have the parts all ready, I've made one already and tested it with 4 CREE in series.. that's what it's designed and optimized for, as I plan to use it for the new CREE MC-E 4 die LED. It's very overbuilt, very efficient. Made to accept 6 Li-Ion cells.

I've got a new Bench PSU on the way so I can get some actual efficiency numbers throughout the input range.

 

[Mr Happy]

This other thread seems relevant to this one here.

Now if you use a linear regulator with a chain of LEDs in series then the inefficiency and the drop out voltage is shared between all the LEDs, so the value per LED is reduced proportionately. For example, if you used a 7135 to drive 3 LEDs the 120 mV drop out voltage when it falls out of regulation gives only 0.04 V loss per LED.

I think the best way to view the 7135 is as a peak safety device rather than a regulator. You select the batteries and the LEDs so that when the batteries are fully charged the LEDs would only slightly be overdriven, and you allow the 7135 to protect against that. Once the battery voltage falls a bit then all the LEDs go into direct drive and you get close to 100% efficiency.

When using NiMH with their flat discharge curve, the loss in regulated brightness during direct drive is not so important.

 

[freedom2000]

This other thread seems relevant to this one here.

Now if you use a linear regulator with a chain of LEDs in series then the inefficiency and the drop out voltage is shared between all the LEDs, so the value per LED is reduced proportionately. For example, if you used a 7135 to drive 3 LEDs the 120 mV drop out voltage when it falls out of regulation gives only 0.04 V loss per LED.

.

Really excellent thread --> very very exciting idea :thumbsup:

JP

 

[shakeylegs]

This thread may be of interest:
http://www.candlepowerforums.com/vb/showthread.php?t=201392

 

[freedom2000]

Yes , it is the same link as above !

But a very good link :thumbsup:

JP