Powering LED array from mains

So, one thing I've been mulling over is halloween, and persuant to that, as usual I am messing with LED's.

Here's the question... assuming I am using "typical" LED's (I am using UV, but that isn't important) in an array, is there any good reason (other than the potentially dangerous voltage) to NOT rectify and filter AC wall current directly into DC and use it?

3.2-3.4v @ 25ma is the rating for the LED's, if I were to string 36 of them in series, that makes 118v give or take. Unless I misunderstand, all I would need then is a current limiting resistor (56 ohms by my calculator) per string of 36 LED's. This also works out conveniently to just about three watts exactly per string, including the dissipation from the resistor.

I'm thinking a bridge rectifier and nice sized electrolytic cap ought to do the job. I'd probably toss in an NPN power transistor or an N-Channel MOSFET so that I could dim and control via logic switching the ground side. Since each string only draws 25ma, the required transistor shouldn't need to be very beefy either, other than being able to handle the voltage itself. I'd be fine with anything rated over 150v I would think. Thoughts?

focalist:
Thoughts?

The theory is seems sound but make sure you a fuse in it...just in case.

focalist:
So, one thing I've been mulling over is halloween, and persuant to that, as usual I am messing with LED's.

Here's the question... assuming I am using "typical" LED's (I am using UV, but that isn't important) in an array, is there any good reason (other than the potentially dangerous voltage) to NOT rectify and filter AC wall current directly into DC and use it?

3.2-3.4v @ 25ma is the rating for the LED's, if I were to string 36 of them in series, that makes 118v give or take. Unless I misunderstand, all I would need then is a current limiting resistor (56 ohms by my calculator) per string of 36 LED's. This also works out conveniently to just about three watts exactly per string, including the dissipation from the resistor.

I'm thinking a bridge rectifier and nice sized electrolytic cap ought to do the job. I'd probably toss in an NPN power transistor or an N-Channel MOSFET so that I could dim and control via logic switching the ground side. Since each string only draws 25ma, the required transistor shouldn't need to be very beefy either, other than being able to handle the voltage itself. I'd be fine with anything rated over 150v I would think. Thoughts?

Your basic theory is correct as best as I can tell. However the biggest concern is safety if you insist on direct connection the the AC hot and neutral lines. Especially when you require (or want) to pwm modulate the leds via an arduino analogWrite output pin, because that requires that you have a 'ground' connection wired from a arduino ground pin and the neutral side of the AC power bus. That means your AC connection requires to be properly 'polarized' or you run the risk of applying 120AC hot to your arduino ground pin, which if you were to plug in your arduino USB cable would probably ruin your day and a lot of hardware. I think as a minimum you would want to use small AC transformer to isolate the AC hot and nuetural from your LED voltage. This could either be a small wattage (wattage as needed for the number of strings you will be using) 1:1 transformer (120vac to 120vac) or two small 'filament transformers wired back to back, 120vac input to 12.6vac to 12.6vac to 120vac output.

Lefty

I'd think the 56? resistor would be moot, it'd only drop a volt and a half or less.

But a bridge and filter will charge up close to 170V (120 ?2) under no-load conditions.

Actually, I might use the 170V peak voltage for the calculations anyway, so string up 50 LEDs instead of 36. Or then use a dropping resistor, or better, a capacitor, which will dissipate very little heat. For 36 LEDs at 170V, we'd want to drop 50V, so 2000? assuming 25mA. A 1.2µF capacitor will have a reactance of 2210? at 60Hz, and will dissipate almost no power, where a 2200? resistor will dissipate over a watt.

Another thing to consider, to run directly on AC, wire two strings in reverse parallel to use both halves of the cycle.

For the sake of safety, I would probably put a PC817 optoisolator feeding the power transistor, which then optically isolates the control from the mains.. no electrical connection at all. I have a bunch of them I bottom-fed off of ebay a while back, they are a good item to have around for just this type of thing.

I'm trying to simplify this as much as possible, assembly, bulk, and cost. I figure on wiring this up and then basically potting the whole backside (everything but the power and control lines) in epoxy, so there's going to be no open conductors anywhere.

The problem with using AC directly is that you are dealing with AC, and that becomes a problem if you want to do things like strobing. High voltage DC is still DC and far more controllable. If I were simply going for a full-time illumination, doing the reverse-parallel thing would be a good choice though...

retrolefty:
Your basic theory is correct as best as I can tell. However the biggest concern is safety if you insist on direct connection the the AC hot and neutral lines. Especially when you require (or want) to pwm modulate the leds via an arduino analogWrite output pin, because that requires that you have a 'ground' connection wired from a arduino ground pin and the neutral side of the AC power bus.

No... his 'ground' pin would go to the rectifier, not the mains.

fungus:

retrolefty:
Your basic theory is correct as best as I can tell. However the biggest concern is safety if you insist on direct connection the the AC hot and neutral lines. Especially when you require (or want) to pwm modulate the leds via an arduino analogWrite output pin, because that requires that you have a 'ground' connection wired from a arduino ground pin and the neutral side of the AC power bus.

No... his 'ground' pin would go to the rectifier, not the mains.

And might I ask where the rectifier gets it's 'ground' reference from, get a clue. :wink: Direct connection to both hot and neutral is being proposed and required. It will work. It's just not the safest practice and that was my point.

Lefty

right, but if optically isolated, that becomes a non-concern as there is no shared ground.. correct? The optoisolator's LED doesn't share a ground with the transistor (wouldn't be isolating then, would it?).. the only thing to pass between the led and phototransistor sides is light...

focalist:
right, but if optically isolated, that becomes a non-concern as there is no shared ground.. correct? The optoisolator's LED doesn't share a ground with the transistor (wouldn't be isolating then, would it?).. the only thing to pass between the led and phototransistor sides is light...

Correct, the use of an optical isolator means that you will not have a direct electrical connections from the AC mains and your arduino ground pin. You will however still be controlling a pretty high voltage loop. It should work, as I said, it would however not be as I would personally construct it. Be sure your switching transistors/opto device is/are rated well above the peak DC voltage that the rectifier will produce. Perhaps a complete schematic drawing of your proposal will help in looking for possible hazards or risks?

Lefty

focalist:
right, but if optically isolated, that becomes a non-concern as there is no shared ground.. correct? The optoisolator's LED doesn't share a ground with the transistor (wouldn't be isolating then, would it?).. the only thing to pass between the led and phototransistor sides is light...

I got the impression the OP was trying to save components...

to NOT rectify and filter AC wall current directly into DC and use it?

Why rectify it at all? One LED strip in one direction and another LED strip in another direction and you are done.

Hehehe.. all I was thinking is that skipping a wall wart could be done :slight_smile:

I mean, at a lower voltage (say 12v), I would need to run three in series then a resistor, and repeat many times. For a hundred LED's, that means I would need thirty three resistors. Running the higher voltage means running more LED's in series, simplifying the construction. I am well known for my ebay bottom feeding for bulk LEDs and such, I have about three hundred UV (400-405nm) sitting in a bin next to me. This begs to be made into a very precise UV strobe.

Though it may LOOK like LED's are on continuously when connected to half-wave AC (or "bounding" full-wave rectified, unfiltered DC), they would in fact be flickering, and at a rate that would be very noticable with motion.. and noticably dimmed in terms of what you would see. Conversion to DC so that the LED's are driven continuously is pretty much a given. The big ole electrolytic is there to smooth out the voltage dropout that is happening at 120hz (2x line frequency) and provide what is hopefully "clean" enough DC to the LED's to allow them to operate without flicker.

This concept becomes even more appealing as I am thinking about the scads of power LED's I have around.. I have gotten 1 watters as cheap as ten cents each delivered, I probably have a solid hundred whites in various temps. You'll find them on ebay in lots of ten for between one and two dollars for white, around twice that for Red,Green and Blue, and UV or Infrared are still around fifty cents a watt. Great little emitters for cheap money, they need to be heatsunk though. I've had a lot of good luck just plopping on some metal with some white lithium grease as heatsink compound. I've gotten ten watters for around a dollar each, and those want in the range of 9-12v at 1A. I typically have been running these from around 12v with a current limiter built from LM317s and 1.2ohm resistors. I may even have ten, which would bring us into mains range.... as I have discovered, it is now more expensive to build the circuits to POWER led's than it is to buy the LED's themselves. Since I use this stuff in high wattages to do special effects photography, alternate ways of driving these high power arrays are always welcome. It may sound crazy to run 100 watts of white LED's, but in many cases I am simulating a xenon photo strobe-- and that's a heck of a lot of watt-seconds I am trying to emulate.. my current photo strobe project is now approaching five hundred watts of LED's.

I never said this was a GOOD idea...

driven continuously is pretty much a given.

Not really. Many leds lights are actually working from ac.

[quote author=Jack Christensen link=topic=128022.msg962913#msg962913 date=1350660980]Actually, I might use the 170V peak voltage for the calculations anyway, so string up 50 LEDs instead of 36. [/quote]Careful there. I believe the spec allows line voltage to be anywhere from 108 to125VAC. Indeed, the lights at my house dim whenever a heavy load is switched in (electric oven, dryer, etc). Make sure you have a little voltage headroom to make sure all of your LEDs are fully lit even under poor line conditions.

dhenry:

to NOT rectify and filter AC wall current directly into DC and use it?

Why rectify it at all? One LED strip in one direction and another LED strip in another direction and you are done.

Indeed, I have seen such as complete surface mount modules for direct mains use (can't find the link today alas). I believe the LEDs are back to back in pairs to limit reverse voltage - certainly I would do this, LED reverse breakdown is low. Of course you need twice as many LEDs for the same brightness, so the rectification method seems superior.

3.2-3.4v @ 25ma is the rating for the LED's, if I were to string 36 of them in series, that makes 118v give or take. Unless I misunderstand, all I would need then is a current limiting resistor (56 ohms by my calculator) per string of 36 LED's. This also works out conveniently to just about three watts exactly per string, including the dissipation from the resistor.

I'd go with Jack's calculations... :stuck_out_tongue: With 56 Ohms, you'll get about 1Amp peak through the LEDs and probably fry 'em all! (I assume your mains voltage is 120V (not 220V)?

One more consideration - The voltage across the LED's is approximately constant. That means the less voltage you drop across the resistor, the worse your "brightness regulation". With 50V across the resistor, a 10% change in the 170V peak voltage is 17V. That entire change appears across the resistor, which means a 34% change in current (and brightness). So with an unregulated power supply, don't try to stack-up as many LEDs as possible in order to minimize the voltage across (and power wasted in) the resistor.

So with an unregulated power supply, don't try to stack-up as many LEDs as possible in order to minimize the voltage across (and power wasted in) the resistor.

I would suggest that that advice be taken together with consideration for efficiency. You want the led string to have a combined forward drop as close to the main's voltage to maximize proficiency. However, that creates bigger current variability when the mains fluctuates.

You will need to compromise between the two.

There are two problems with this approach. First is safety which is covered, second is longevity. Connecting LEDs to the mains is a very bad idea without any type of regulation. LEDs, especially anything using a blue die (Blue, white, pink, purple, sometimes green) or a UV die like you are using are very sensitive to surges and even to static electricity. When you connect to the mains without regulation (using a transformer or not) every surge is going to be passed through to your LEDs. You are proposing a Mains circuit to a rectifier (hopefully through a fuse and preferably an Isolation transformer if the wiring is exposed) to a resistor and then to the LEDs. This leads to problems.