a new, cheap and useful kind of LED lamp

[vtpeaknik]

I've just gotten hold of a new kind of LED lamp that seems very nice for general room lighting with no grid power.  It's much brighter than any LED lamp I've tried before, and spreads light over a very wide area.  The white color seems pleasant.  And it's very affordable too (less than $10).  It's called the "UFO" light because of its "flying saucer" shape.  I think you can get a one-ring version of the same concept here but I bought a 2-ring version here.  (Broken english there, but was quickly shipped across the ocean.)

I measured the current (for one ring of 24 LEDs) at about 400 mA, at close to 5V supply, i.e. about 2 watts.  The lit LEDs feel warm to the touch after a short while.  They claim it'll run 30 hours on one set of 4 AAs, that does not seem possible, 5 hours is more likely.  And the way this thing is built, removing, charging, and replacing the AA cells is a akward.  I would rather run it directly on 12V.  I think that can be arranged:

Looking at the picture of the innards, you can clearly see that each ring of 24 LEDs is built as 4 quarter-rings of 6 LEDs each, all connected in parallel.  (I see no current-limiting resistors - could those be built into the LEDs?)  It would be easy to rewire the ring as two pairs of quarters (12 LEDs each) in series.  Running it on  about 12.8V from a full set of lead-acid batteries would be similar to running it on very fresh alkaline AAs at 1.6V each.  Perhaps that's a bit too much, could add a resistor to reduce the voltage to about 10V or 11V, something like 4-8 ohms (or a couple of 1N4001 diodes in series).  I think I would wire the outer ring like that, and keep the inner ring connected to the AAs, that way it's dual-powered.  (There isn't a whole lot practical difference in the brightness between one and two rings, it's "only" a factor of 2 and our eyes work "logarithmically".)  Will need to disconnect one of the red wires that links the two rings.

[Simen]

Your best option would be to insert a LM317 regulator connected as an constant current regulator. They require 3v higher supply voltage than the output voltage, so in your case; 4 AA = 6V, would mean that you could run the led-ring from 9V up to the LM317 max input at 40V.

Setting the regulator to deliver your measured 400mA would ensure constant brightness no matter the voltage from 9V and upwards, and your diodes would be safe from burnout.

Though, i would have connected the led-ring to a regulated 6.0V supply, and verified the Amp usage first, before settling on the current...

[vtpeaknik]

Running it at 6V off a 12V+ supply (burning half the watts in the regulator) would be defeating the purpose of efficiency.  By re-wiring the LEDs into two series groups the current would only be about 200 mA.  Adding some sort of voltage reduction scheme that reduces the 12-13V down to 10-11V would also reduce efficiency, but only by about 15-20% - less than the loss inherent in charging NiMH AAs from the 12V system.

And perhaps voltage reduction is not necessary.  2.4W spread over 24 LEDs is 0.1W each, they should be able to handle that, no?  I'm still puzzled as to whether there are any invisible resistors or some other current-limiting scheme built into this lamp.  They say you can run it on alkalines or NiMH cells, that's a range of voltage already.  Sorry for my incomplete current measuring, I was in a hurry with limited gizmos on hand.  I was aiming for 5V not 6V since I would never run it on alkalines! (Well perhaps the rechargable alkalines - anybody still using those? Look up the "Juice" brand, they currently offer a nice little smart charger that can charge those and also NiMH cells and handles each cell individually.)

[ghurd]

I doubt there is a resistor built into the LEDs. Maybe.  

I have seen some indicator-brightness LEDs with internal resistors for 14V.

Resistors in the switch is more common.  No resistors at all is even more common (Poof!).

It'll run 30 hours on 4 alkalines... but it sure won't be very bright at the end.

G-

[vtpeaknik]

So do you think they're supplying 6V directly to the LEDs with nothing to limit the current?  Are there LEDs made for such a high voltage?  I thought 3V working voltage was more typical.  Perhaps each LED is actually a pair in series?  They lit up (dimly) with only 4V supply so that does not seem likely?  I did measure the voltage across the switch/control unit while lit, and there was a tiny bit of it.

I'm leaning towards running it (rewired) on 12V with both a diode and a 5-ohm resistor in series.  Could also add something to limit the current when running on AAs, that might extend the LEDs' life?

[ghurd]

"So do you think they're supplying 6V directly to the LEDs with nothing to limit the current?"

Yup!  That's what I think.

The high drain makes the battery sort of limit itself, like the single LED things that use a few LR44 button cells.

"Are there LEDs made for such a high voltage?"  Not 48 for $7 including shipping half way around the planet.

"Perhaps each LED is actually a pair in series?"  Not by the looks of the guts.

Reducing the current will extend the life.

I would maybe rewire it for 12V by separating the 8 quarters.  Series 2 quarters and give that section its own resistor.  Repeat 3 times.  Meaning 4 resistors.  More reliable.

LEDs don't follow Ohm's Law.  Best to measure the current (in all 4 sections individually) in the final layout before calling it done.

Expect to need quite an assortment of resistors.

Another option for more efficient 12V is running 3 quarters in series, 3 more in series, then the last 2 in series.  The 12V battery drain will be reduced by 25%.

G-

[vtpeaknik]

Great ideas, G.  I'll try out 3-quarters-in-a-series off of 12V and will report eventually.  If that's still bright enough, I think the way to go is to have two such strings running on 12V, leaving 2 quarters still in parallel still connected to the AAs via the switch.  That'll mean 36 LEDs on 12V using somewhat less than 200 mA or about 2 W, and about a 200MA load (and much less light) on the AAs when they're used.  Lots of options!

[rossw]

In a past life when LEDs were new and interesting, and people were doing innovative things in electronics, and where I was working at the time had a decent budget for R&D, we used bucketloads of a very nifty LED that was a constant-current device.

That is, it didn't have a "resistor" inside, it had a constant-current source inside.

These LEDs would operate safely and indefinately (without a series resistor) on anything from 3V to about 14V if my memory serves correctly.

They were about twice the price of a standard LED (at the time), but the space saving and convenience in driving them was well worth the difference.

I've probably still got a few in my junkbox...

I just had a quick google for them but find no (modern) references for such a device, so perhaps the "here is a LED, here is a driver, you marry the two" design philosophy has taken over.

[richhagen]

My son is in Cub Scouts, and they had to build an electronic circuit for their Webelos engineering pin.  I helped out with this a bit, and we built very simple pwm drivers from 555 timer IC's and used 4 D cell batteries in series through the circuit to power 3 Cree LED's in parallel.  Such a circuit could be used to drive the LED circuits you light has from a six or 12V source with reasonable efficiency, with the added bonus of being able to adjust the light output.  Here is a link to a circuit similar to the one that I modified for use in our little 'tent lights'.

http://josepino.com/circuits/?pwm

I had a few differences in mine, which was referenced from a different source and tweaked by experimentation, namely that I used a 500K pot (because they were available), and had resistance in series with one of its legs to reduce the duty cycle as I did not want the individual LED's to be of too high of intensity (as expected, the first thing they did was try to shine them in each others eyes).  Also, I had a capacitor across the power input leads as the circuit developed a lower frequency flicker without it.  I also had a third 4148 connected across the output in case of any minor inductance, as would typically be done with a motor controller.  

This is a very blurry photo of the circuit.  There are only a few components, and it is just on a proto-type board.  One could likely make it much more neatly than I.  There are six leads from the board, power+, power-, three leads for the pot, and the negative lead to the led's.  

Happy Scouts with their lights.  I did not measure its efficiency, but I expect it is fairly efficient with a low power 555, and relatively high resistances on the r/c portion of the circuit.  Adusting the resistance to lower the duty cycle further would likely be needed but should be doable to run your LED's from a 12V source, as I did run one of these test circuits at 12V as is with it turned down, but had I have turned up the pot I am pretty sure I would have blown the LED's.  I also have three led's hooked in series to a slightly different 555 pwm circuit running at 12V now, which I used for plant starting in the spring.  

A slightly more complicated version had another capacitor so that the operating frequency and the duty cycle could both be fine tuned, but that was getting a bit complex for 10 year olds (and me as well), although it did operate better.  Anyhow, I was just thinking that something like this might work well, and efficiently for your lights, although based upon what you appear to have paid, it would be a significant portion of the cost.  Rich  

[vtpeaknik]

As G said, the LEDs light up at about 3V and it takes quite a current to get them above 3.5V.  I've now tried it as three quarter-rings (6 LEDs each) in series, with a 33-ohm resistor also in series.  At 12V supply, it uses about 70 mA (about 12 mA per LED), rising to about 100 mA at 13.8V supply (similar to running on 4 AAs).  This seems about right for my needs (no solar charge at night)?  But if you are going to run it off of a 12V battery that may be charged (by wind) while using the lamp, you may want to up the resistance to 47 or 51 ohms or so.  The other group of 3 quarter-rings should have its own resistor.  Thanks for the tip, G!

Thanks for the other comments too.  Yes a PWM circuit would be nifty, but seems like way too much work and expense relative to this $7 lamp.

[ghurd]

Careful with the resistor.  I'd use 1W.

If you can do it with a switch (no time lag between the comparison)... may want to check "how bright" they are to human eyes in the dark with higher value resistors.  May find there is almost 0 difference between the 33 and 51, but the current goes down (better LED and battery life).

Could be impressive for a night light with a total of maybe 5~10ma to the entire light.

The ones that light will be very efficient.  They won't all light with cheap LEDs (or even good LEDs).

Hi-Lo switch can be SPST, with a ~1K resistor jumping the terminals.

I agree about the PWM in a $7 light.  But I have used the cheap housing and changed the LEDs to better LEDs.  Not sure it made sense either.  

Be nice to see 5mm white LEDs with built in current regulators.  That would be handy.

G-

[vtpeaknik]

Why 1W resistor, G?  At 70 mA it's about 160 mW, and 100 mA it's still only 330mW being dissipated?  I suppose you're worried about the worse-case scenario of it being connected to a battery bank that's being equalized (15V)?

[ghurd]

And a 1/2W at .33W can get pretty warm.  Some plastic gets warped without much heat.

G-

[vtpeaknik]

Now wired the other set of 3 quarter-rings with a 51-ohm resistor, while the first one still has 33 ohms.  My naked eyes cannot see a difference in the brightness of the LEDs of the two sets.  (That's looking at the LEDs directly, in daylight, no switch, just all lit up at once.)  They are both a lot less bright than the remaining 2 quarter-rings running on 4 AAs (and presumably over-stressing those LEDs).  The voltages on the resistors, with a 12.4V supply, are 2.54 and 2.67, i.e. the currents are 77 and 52 mA, or 13 and 9 mA per LED, with 33 and 51 ohms respectively.

Efficiency is nice, but I'm also trying to set up a lamp that is fairly bright, not a night light.  With both parts at 52 mA, the total would be 104 mA or 1.29 W supplied, of which 1.01 W reaches the LEDs, spread over 36 LEDs, or 28 mW each.  (1.5W total reaching LEDs if using 33 ohms.)

[Norm]

Putting 2- 6 volt lamps in series,( the ones I

had were similar....20 in a ring....$8.95) then hooking

up to a 12 volt battery works.....for awhile !

    .....then things start goofing up !

A no brainer for people like me is to just make

a special 6 volt battery pac with a large enough capacity to last all night....

I give up on Leds for a really good light for now.

I rather like the 4 watt fluorescent lamps for

$4.75 clearance sale at Walmarts for now.

Who knows....someday they may have A large

surface area LED lamp???