ultrabrightLED flashlight

[zachtheterrible]

yeah, I know that I started a subject like this before and I kind of gave up on it. The reason was that I didn't have the LED. I just bought myself a 1 Watt ultrabright white LED from allelectronics for $11. https://www.allelectronics.com/cgi-bin/category.cgi?category=340500&item=LED-110&type=store

This thing is insanely bright! If driven with 3 volts, its quite hard to look at. If driven with around 4 volts, it comes very close to being as bright as a maglight! The problem with driving it at 4 volts is that it becomes very hot (fortunately it is mounted on a heatsink), and as a result obviously not very efficient. I would imagine its efficiency to be about that of a regular incandescant bulb, which sort of defeats the purpose. This is why I would like to normally drive it at 3 volts.

Basically what I'm looking to do is use a 3 volt powersupply. I'm going to replace my maglight bulb with this thing, so I'll be using two D cell batteries. I am wondering if there is a circuit that I can build to step the voltage up to 4 volts when I need to. So i'll have two brightness settings on my light.

I would also like to have a voltage regulator so that the LED stays a constant brightness and doesn't get dimmer as the battery dies. Just how efficient is this type of circuit?

 

[audioguru]

Hi Zach,
You are going to blow your LED without a heatsink and without current-limiting. It is much more efficient than an incandescent lightbulb and is much easier to melt.

Its 4V max rating is what some LEDs would measure when operating at its max of 350mA with an infinite heatsink. All other ratings talk about current. It typically has a voltage across it of 3.3V at its max current. 4V is way off the current graph of a typical LED.

The curves show an LED reaching its max temp at 150mA without a heatsink. They don't guarantee its lifetime but say to keep it cool to extend it, so don't use it at its max.

I don't think it will work with a battery voltage as low as 3V. A 3V battery quickly drops to 2.4V, then 2.0V when it is exhausted. The LED needs about 3V to 3.5V with a heatsink, plus another volt or more for its current-limiting resistor or constant current source.

All good LED flashlights have a stepup constant current circuit using a switchmode IC, inductor and Mosfet. So their brightness remains constant as the battery voltage sags until the thingy stops.

Your LED needs an external heatsink. Lumileds have a built-in heatsink.
I remember that Silicon Chip magazine had a couple of projects for a flashlight with a Lumiled and switchmode supply for it. You can see the project for free if you enter in a Google search: Silicon Chip Starpower. :lol:

 

[Oznog]

Oh yeah you need a sink. The die is very sensitive to high temperatures. Moderate overtemps will degrade the die whereas high overtemps will blow it outright.

Also, if you look at the spec sheet you will see the efficiency drops with higher junction temp.

A 1W device needs a load of heatsinking, and don't exceed 1W even with the best sink. I am unfamiliar with this package but note that the Luxeons are funky in that the thermal slug in back has to be electrically isolated from both the positive and negative wires.

These things do need a well regulated current. This requires either some sort of constant current supply (greatly recommended!) or the ballast resistor has to have enough voltage drop to stabilize the battery and LED forward voltage variations. This may mean a substantial heat generation and efficiency loss. A voltage regulated supply is not appropriate for LEDs, it must be current regulated.

Some focusing would be recommended. That's a 100 deg output which is far too wide for a useful flashlight.

 

[chiba]

Instead of the big D cells, How about using 6 x AA NiMH's with solder tags in series and regulating that down to 3.3V with an LM317T or whatever, If you get decent cells ie 2500mah they should last a long while. You could also switch a few resistors or a pot on the regulator to give you your brightness levels. The big question is will the AA's squeeze inside the torch??

 

[audioguru]

Hi Chiba,
The LM317 linear regulator will waste more power than the LED uses.
Its fixed output voltage will overdrive the LED's current as the LED heats-up. If the torch is left in the sunshine in summer, the LED current will be so high it might melt. In the winter, it won't be very bright.
A switch-mode supply with current feedback is stable and is much more efficient. :lol:

 

[Roff]

Hi Chiba,
The LM317 linear regulator will waste more power than the LED uses.
Its fixed output voltage will overdrive the LED's current as the LED heats-up. If the torch is left in the sunshine in summer, the LED current will be so high it might melt. In the winter, it won't be very bright.
A switch-mode supply with current feedback is stable and is much more efficient. :lol:

Audioguru, as I'm sure you know, the LM317 can be used as a current source, but your power dissipation argument still stands. The switch-mode current source is the weapon of choice.
I don't know if Linear technology is good about giving samples (they do offer them), but the LT1932 looks like a good part. There are lots of others available.

 

[Someone Electro]

I seen an tiny circuit that acts as an constant curent source. and can run an white LED from an single button batery.It can squize the LAST volts out of baterys.
But thats for an litle LED but i think whith a bit of tweaking and a biger transistor it cod run that 1W big ass LED.

 

[Oznog]

I seen an tiny circuit that acts as an constant curent source. and can run an white LED from an single button batery.It can squize the LAST volts out of baterys.
But thats for an litle LED but i think whith a bit of tweaking and a biger transistor it cod run that 1W big ass LED.

You're talking about a boost converter here. The buck converter is the one to lower the voltage. Both require inductors for all implementations. The inductor needs to be selected based on Isat (cannot be exceeded at any time), rDC (to keep heat generation down), and inductance to get an acceptable ripple at the desired frequency.
SuperTex makes dandy constant current LED drivers like the H9910. The 9910 needs a lot of voltage to run though.

Boost converters are kind of a pain because they provide a pulsed output which is not desirable for LEDs. A cap will not fix this when driving an LED, it has to be yet another inductor.

 

[chiba]

Here ya go, step-up dc-dc converter with Better efficiency (~80%) than a couple of lm317 in current/voltage mode and you can use the D cells.

https://www.alldatasheet.co.kr/datasheet-pdf/view/ZETEX/ZXSC100.html

a sample torch and pcb from siliconchip magazine:
https://www.siliconchip.com.au/cms/A_30805/article.html

 

[audioguru]

Hi Chiba,
Links to Silicon Chip projects don't work because they want you to pay to see them. Google has the links but they are complete without the nagging to pay. "You can see the project for free if you enter in a Google search: Silicon Chip Starpower."

 

[chiba]

audioguru, your search points to another article, here i have zipped the one showing the torch and pcb using the ZXSC100 chip.

 

[zachtheterrible]

Hi everybody, thanx for the feedback. I AM SO PISSED OFF! last night i was fiddling around with my LED. I had it hooked up to a wallwart supply. I was also tinkering with a motion sensor that I got when I ordered the LED. The motion sensor used 12V, my LED 3V. So when I got done with the motion sensor, I switched the supply back down to 3v and connected it. I heard a small pop, and a bright flash of light and my LED was fried!!

I should've known better. I had watched the voltage coming out of the supply before and noted that it would take about 3 minutes before the voltage was back down to 3v because there must be a rather large capacitor in there with a bleeder resistor value that is way too high! So now I have to wait till tomorrow to get another one (i ordered it right away despite its $11 dollar price tag and $6 S&H. I ordered a couple other things to justify the S&H :lol: )

-------------------------------------------------------------------------------------

So, enough of my wining. The LED already has its own heatsink, which is a round aluminum plate about 2 or three centimeters diameter. The LED actually works beautifully with a battery voltage of 3v. May even be lower, those batteries i'd been using have been accidently shorted I don't know how many times. Audio, I looked at silicon chip magazine, and the only switchmode supply for luxeon LEDs was made to work with 12 volts. Perhaps you could provide a direct link?

Ron H, I like that part, the LT1932, and with 80% efficiency is very good.

Chiba, I'd really like to see that project, but I have to pay for it, can you post a picture?

Oznog, yeah I took the reflector off my maglite and fitted it with my luxeon and WOW! I could easily light stuff up on the other side of my garage to read writing! I have a four car garage. I really wanted to test it out at night but never got the chance, seeing as I blew the thing up at around 5:30PM :cry:


So my main requirement is that I can run this thing from 3v. 4.5V would be acceptable as well.

Obviously a direct drive of the LED is the most efficient thing there is, but the problem with that is that the brightness will start to drop. I was thinking of driving my LED directly from the 3v batteries while they are fresh, and then using some sort of regulating when the voltage drops too low, this would increase battery life I'm sure. When driven from the 3v, I don't even have to use any kind of current limiting. The LED doesn't even get warm.

When driven from my two AA batteries, I measured the current throught the thing and got a little bit under 20mA. The meter must have a high resistance though because it caused a noticable dimming when connected in the circuit. So I'm guessing that there is around 30mA or so when directly driven from 3v.

 

[chiba]

>Chiba, I'd really like to see that project, but I have to pay for it, can you post a picture?
zipped above, if it does not work i'd be happy to post it in pic form.

>So I'm guessing that there is around 30mA or so when directly driven from 3v.

1 watt should be 1000mw / 3v = 333ma, your meter on the wrong scale or you got the .1watt led?

 

[audioguru]

Hi Chiba,
Thanks for zipping the Silicon Chip torch project that I haven't seen before. It is exactly what Zach needs. :lol:

I am confused by which LED Zach is using. His 1st Cree XLamp is very wide angle and needs an external heatsink but now he is talking about testing a narrower angle Lumileds Luxeon that has its own built-in heatsink. :?: :?:

 

[zachtheterrible]

Chiba, wow, that is EXACTLY what i'm looking for. Great efficiency, small, what more could I ask for? I am going to use that circuit, unless anybody can provide something that will top that (although I doubt it :lol: ).

What does everyone think of my idea of direct driving the LED until it becomes dim, then hooking it up to the DC-DC converter?

Chiba, I don't think its drawing 333mA. Like audio said, the maximum temperature is reached @ around 150mA, and I don't think it was even getting warm when I was using it. Hmm, maybe my batteries voltage is a lot lower than I thought. That would mean that I wasn't getting as much light output as I would with fresh batteries, and even that was extremely bright! Can't wait to get another one in the mail tomorrow.

 

[mstechca]

:lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol:

Zach, Have you ever thought of using a RESISTOR and connecting it in series with the LED? It is by far, the most simplest fixed brightness control you could make, and it costs you under a quarter! Or you can go fancy, and turn the resistor into a potentiometer for a few quarters more.

 

[zachtheterrible]

:lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol:

Zach, Have you ever thought of using a RESISTOR and connecting it in series with the LED? It is by far, the most simplest fixed brightness control you could make, and it costs you under a quarter! Or you can go fancy, and turn the resistor into a potentiometer for a few quarters more.

mstecha, have you taken the time to read the thread? Because i explicitely said that i want something to regulate the current, so as for the LED to remain at the same level of brightness as the battery voltage drops :lol:

I also want something nice and efficient, i don't consider a resistor to be all too efficient :lol:

Had you read further, you would have come across my saying that I don't need a resistor in series with the LED, seeing as there is not enough current to damage it :lol:

 

[Oznog]

Had you read further, you would have come across my saying that I don't need a resistor in series with the LED, seeing as there is not enough current to damage it :lol:

But you DO need one, or another current-regulating solution. That's what we're saying and that's one of the main reasons you blew the one you had. The only exception is some very small batteries have such a high internal resistance already they don't use an external resistor, but that's only utilized in coin cell powered LED flashlights as far as I know and it's not a great idea.

The 1W device on a 3 cm disk is only minimally heatsinked in open air at normal temperatures. I know, I use 1W Luxeon devices too. In a flashlight, this will be insufficient since there is no exposed air cooling it. Thermal connection to the case should be made.

The heatsink actually has a substantial thermal lag. You may need like 5-10 min of runtime to see how hot it's actually going to get. And remember this isn't like a resistor where "as long as it's not on fire it's fine". The junction MUST have low temps to work efficiently and preserve its service life and that means you need a cold heatsink.

The Zetek single cell DC-DC converter is not a current-mode device and will not do what you need. Also, the circuit listed will not work with an LED. It uses caps to filter the output ripple inherent to buck converters and caps don't work for direct driving LEDs unless you bring the ballast resistor back in. You need to find an LED driver chip or a "current mode PWM" to build a circuit around.

The Cree part you have is cool, but for $15 Lamina can give you over 4x the power with the BL-2000 series. And they're not hard to come by, Mouser will sell you 1 for that price. Lamina has an incredibly low thermal impedance from junction to backing and thus heatsinks quite well (a quality heatsink is required even for a brief power-up test).

 

[audioguru]

I don't need a resistor in series with the LED, seeing as there is not enough current to damage it.

Hi Zach,
You receive your replacement LED but don't know its forward voltage that is different for each one. 3.3V for most, 4V max for some others and maybe only 3V for your new one. Then it will draw 350mA from your 3V battery and quickly melt without an external heatsink.

What if the LED's forward voltage is only 3V at 350mA but your battery is brand new and is 3.2V? Pop and poof again! :lol:

 

[mstechca]

have you taken the time to read the thread? Because i explicitely said that i want something to regulate the current, so as for the LED to remain at the same level of brightness as the battery voltage drops :lol:

I have found a low battery indicator here:
**broken link removed**

If you can't see constant brightness from that, then maybe hook up your source to a 7805 voltage regulator, because one of those will give a 5V output.

I also want something nice and efficient, i don't consider a resistor to be all too efficient :lol:
Had you read further, you would have come across my saying that I don't need a resistor in series with the LED, seeing as there is not enough current to damage it :lol:

The only way I see it being ineffecient is the fact that it turns power into heat. Let's face it. Would you rather waste a small amount of power to keep an LED running for a long time, or would you rather see your LED die instantly like it already did?

:!: