Do 10+ watt boost converters exist?

[davep]

Hi there helpful folks. This is a rather very long post with the main goal in locating a decent (10+ watt) constant current boost circuit that will output ~ 10.5 volts across 3 lux III's @ 1A. So no need to go through the pain and suffering of reading this if not interested (you should be though...)

I will be building a flashlight using 3 lux III's driven @ 1A by a buck (step down) converter- however I am also interested in finding a high power current regulated step up (or boost) converter as well so I can experiment with a smaller body (less/smaller) batteries flashlight. In the end I will have 2 prototypes- 1 bigger, heavier, and much more efficient; the other the opposite, much smaller, but with the same regulated output for less time.

I have found a good buck regulator that has enough (12W) to power 3 Lux III's at 1A. The consequence of this implementation is that the minimum voltage required from my batteries is 12v since ~10.5 V across leds + 1.5 volt headroom for buck regulator to operate. This means I am needing to use 12 Nimh aa or aaa (I refuse to use anything but Nimh by the way).

If there are any high power (10+ watts) constant current step up converters that anyone knows of, please let us know.

I really hope there is. If not, maybe someone has a proven efficient circuit design that I could start from. There is nothing in the world that can power 3 lux III's from under 12 volts that I have found so far. I have been asking google for 3 hours (quite a sad way to spend a friday night I just realized) so maybe someone has some suggestions.

Anyways, as far as I know, ic or microprocessor controlled PWM designs are the most efficient, reliable, small, and widely adaptable- particularly the all in one designs, namely the led dynamics powerpuck. This is a good product as far as I can see. However, it has only 3W output unfortunately for the 350ma. The 700 ma is 6W but only utilized with 1 on each 5W- which is bad $in/light out compared to 3 lux III's and one driver. The best solution I have found (on paper anyways- I will get it soon in the mail) is the led dynamics 3021 buckpuck- has anyone used this by the way, or I am a pioneer?).

However everyone likes smaller things, and a good step up converter would allow a good cute constant output powerhouse to be made with less/smaller batteries since it requires less voltage. I also plan on integrating a 0-100% dimming pot (not resistive- so it is efficient) as well as a battery status indicator (I'll figure out something).

By the way, if anyone has a good efficient homemade circuit solution, please let me know. I can figure out what I can build from discrete parts if someone has some good design parameters to maximize efficiency and maintain rock solid current. I have decent knowedge of electronics but it is always good to have somewhere to start.

Anyways, hopefully there is an integrated solution (too bad there is not a more powerful powerpuck...

I hope I havent' wasted too much of anyones time reading this. Hopefully however you all will understand that constant current is an absolute necessity and that there are a few good current drivers and a lot of bad current drivers. The only disadvantage to a constant current flashlight is suddenly very diminished output when the batteries finally die without you knowing. However, there are some methods that can estimate the remaining runtime of a constant current output flashlight (even nimh with shallow voltage drop). The easiest way to do this should some type of analog solution (such as opalec newbeam red led- although I don't know how well or exactly how theirs works. If I take enough drugs I may even try to display the remaining runtime with a discrete lcd readout.

Has anybody tried this stuff or even thought of this stuff, or am I just a freak fiending for the perfect flashlight?

By the way if it works out, I will owe it all to Klaus and the helpful folks in this forum (and my craziness).
I will keep people posted. It will take a few weeks though to get this stuff

Dave

 

[CM]

It might be more efficient to use one driver per LED. Most high efficiency (synchronous boost) converters have a relatively low input voltage limit and your output voltage requires a pretty good step up from that which means sacrificing efficiency. At 10+ watts, you need every bit of efficiency that you can get, preferably in the 90+ percent range. If you only manage to get 70 to 80% efficiency, you're burning 2 to 3 watts of power just in the converter alone (remember that there's more heat generated by the LED on top of that). That's a lot of heat. Wayne (dat2zip) is coming out with a next generation badboy circuit that gets great efficiency using a synchronous boost converter. It might be worth the wait.

 

[bikeNomad]

Generally speaking, you can add a current sense resistor and amplifier to a switching voltage regulator and get a constant current regulator.

However, you may have loop stability problems driving LEDs because of their nonlinear impedance. Luckily, you can heavily damp the control loop if you're driving LEDs because speed of response is not important (though you want to avoid overshoot as well). Plus of course you'll need a bigger inductor for higher current.

If you're looking for higher power than single-chip solutions, you probably want to look at "PWM controllers" which can drive external transistors.

Another choice is "Battery charger" ICs which already are set up for constant current output.

I recommend that you get Linear Technology's SwitcherCAD III (www.linear-tech.com) and simulate some circuits.

Some of their parts that you might find interesting are:

LT1513 battery charger (can provide 1A output from 7 Ni cells)
LT1371 3A switching regulator (would need sense amp)
LT1070 5A switching regulator (also would need sense amp)

A handy equation from one of their data sheets to decide whether a particular transistor, IC, or inductor would be suitable is:

IPEAK = (IOUT * (VOUT / VIN)) + ((VIN * (VOUT - VIN)) /(2 * f * L * VOUT))
Where:
IPEAK = peak output current (for sizing inductor and transistor)
VIN = Minimum Input Voltage
VOUT = Maximum output voltage
f = Switching Frequency in Hz
L = Inductor value in H

So for a 5-cell Ni to 10.5V output at 1A, you'd get:

IPEAK = (1 * (10.5/4.5)) + ((4.5 * (10.5-4.5)) / ( 2 * f * L * 10.5))
for an f value of 500KHz and L of 10uH:
= (1 * (10.5/4.5)) + ((4.5 * (10.5-4.5)) / ( 2 * 500e3 * 10e-6 * 10.5))
= 2.33 A

 

[davep]

This is very good info. Thanks a lot. I will look extensively into all this stuff after exams. An advantage of building my own circuit with 1 per led is that it will likely cost a less than the all in one solutions- ex: the buckpuck I ordered cost $30. However, sourcing many little components just for testing may be a hassle. Again, I will look into this stuff soon. Any other comments are very welcome.

Also, I should update my "only nimh batteries" statement. I would consider pila rechargeables too if they hold up to very high discharge. Also, I would hope they have graphite electrodes. Lithium ion rechargeables are the best solution for a small flashlight as long as they aren't drained below 2.5-3V. If they have no protection circuitry to prevent minimum voltage discharge, then this circuit needs to be introduced since li-ion can't drop below 2.5 or 3V (depending if type 1 or 2 graphite) per cell without damage.

I would also like to find larger or smaller than the typical li-ion diameter to make my own battery pack. If people have info on potential batteries, please let us know.
Thanks,
David

Any comments are welcome