Making a small high power LED array - best way to drive them?

[mcfly666]

I'm going to make a simple high power led setup for my aquarium, I just need to figure out the best way to drive them.

The array is going to be made up of two or three 10w COB leds, specs are below...

Voltage: DC 32-34V
Current: 350mA
Power: 10W

I could get a driver for each of them, but I can only find simple 12v dc ones. I was thinking about using an old laptop charger and running it though one of these:

http://www.ebay.co.uk/itm/Digital-co...item3f29e7f18e

Would that work?

[tom66]

Re: Making a small high power LED array - best way to drive them?

Step down module, so can only output less than its input. You want a step up.

Ideally, you'd use a converter which had a current limit instead of a voltage limit, which would work better, as the LEDs warm up the voltage will drop. Without control, this can lead to increasing current, lower voltage, increasing current, etc. until the LEDs or power supply are destroyed -- not good! Yes it's possible to find a sweet spot where this doesn't happen but requires you to measure them carefully and also maybe use a series inline resistor which wastes power.

Also remember for 10W LEDs you're going to have to keep them cool, consider a decent heatsink and maybe a fan to blow air over such a heatsink (though be careful with the fan - the contaminants, if any, in the water condensing would quickly stop it from working.) The LED die temperature typically may not exceed 90 to 120°C. The LEDs are nearly the most efficient light emitting devices (only sodium-discharge lamps beat them, and they're orange only) but they still waste around 85% to 90% of input power as heat.

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

^^ Ah sorry Tom, your right. I was looking through Ebay and might have got the wrong link.

Though in the description (Chinese language barrier perhaps?) it does say its a "DC-DC 400W 6-40V to 8v-80v 10A Boost Converter Step-up Module Power Supply" (confused)

Anyway the one I was thinking about was http://www.ebay.co.uk/itm/Digital-le...item417762165a

I might get a cheaper analog one, and just use a DMM to set it up. Though the one below seems to have a pretty hefty heatsink

I'm going to wire the LED's in series (bit rusty) are the volts for the LEDs accumulative but not the current EG run the LEDS @96v - 350 millamps?

Thanks Tom.

[budm]

Re: Making a small high power LED array - best way to drive them?

"Voltage: DC 32-34V
Current: 350mA
Power: 10W"
Is this LED module with built-in LED drive circuit? Any link to this LED?

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

OFc, thank you. Though in my previous post I think I made a mistake about wiring them in series - the voltage will be out of spec for the stepup boost converter. I will have to wire them in parallel at just over 1amp (I think)

The link is below, I think they have a chip on them (IC led floodlight) I thikn COB is "chip on board" ?

http://www.ebay.co.uk/itm/2514064918...84.m1439.l2649

Cheers

[tom66]

Re: Making a small high power LED array - best way to drive them?

In general you do not want to wire LEDs in parallel unless they are on the same substrate because matching them will require ballast resistors. You will have to find a way to put them in series with a higher voltage converter outputting a lower current.

Each LED device is probably 9 or 10 standard white LEDs wired together in one die, so you're probably looking at around 90 to 100V to drive them in series. This isn't impossible, but it's approaching the point where you'd probably have to design your own driver (something off the shelf will probably be too expensive.) You could also run each LED from different power source.

[mariushm]

Re: Making a small high power LED array - best way to drive them?

Really, the easiest way to do this would be to use a 36-48v power supply and then use a simple led driver to get your 350mA.
For example, this thing is idiot proof : http://uk.farnell.com/on-semiconduct...2-3/dp/2382343 (mouser and digikey have them in stock, farnell will have them soon)

Just put it on a large pcb copper plane or heatsink it (you'll need large heatsinks for those 10w leds anyway) and it will limit the current through a led at up to 350mA (adjustable) and with up to 50v input voltage.

Use one for each led you're going to have... then put the led + driver sets in parallel connected to your power supply.

Other simple to use led drivers...but these need additional external parts (inductor, diode etc easy to find)

http://uk.farnell.com/zmdi/zled7000-...5/dp/1898426RL (max 40v input, 750ma)
http://uk.farnell.com/diodes-inc/ap8...sop/dp/1843875 ( 48v input, 500mA)

etc etc

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

Thanks for the info tom, I've now found there's two types of these dies. One has the specs above, another has a 12v feed with increased current - guess the 12v one will be easier to work with off the shelf parts.

Damn it, Ill have a little play with them when they get here and see how I get on, is there anyway a mortal can see the I/V characteristics of a die? I suppose if the I/V is close you could run them in parallel without a ballast resistor?

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

Thanks for the information @mariushm

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

Just read through your post @mariushm

What an awesome idea, the LED's are going to be connected to a large piece of aluminum which acts as a lumair/hanger with some regulated computer fans so cooling wont be a problem (I think)

All I need are one of these and to mount the regulator/driver on the heatsink with the led's (so simple!)

Do the semiconductor run @ the full 350ma without changing the Radj?

Much thanks!

[tom66]

Re: Making a small high power LED array - best way to drive them?

For those ICs, you'd have to design and manufacture a custom PCB to use it.

[mariushm]

Re: Making a small high power LED array - best way to drive them?

The first one in DPAK package doesn't need a pcb. You can bend the pins towards the air, put some thermal paste on the back tab and figure a way to keep the IC pressed against the aluminum/copper heatsink. The tab goes to ground, so it won't have to be insulated from the other ICs (but may be a problem if the led heatsink is the anode, you may cause shorts in this case) and you can simply solder wires to the other two pins (or just one if you don't want to adjust the current).

That first led driver I gave you behaves like a pass transistor, it allows up to 350mA of current to go through it.
By default, without anything connected to it, it allows up to 150mA, depending on your input voltage.

You can adjust it higher by placing a resistor between Radj pin (3) and Cathode (4).. see datasheet ... you have the graph on page 4, bottom right corner.. basically for 350mA, resistor should be around 4 ohm (but keep in mind the value is at Vak=7.5v meaning value and graph is only accurate when input voltage is about 7.5v higher than led voltage... for example if you power the led with 36v, even if you set this resistor the IC may not pass 350mA through it, but a lower value)

Alternatively, you can also put two in parallel and as long as they'll heat up evenly they'll split the current (so two such ICs will allow 300mA to pass without using any external resistor, or up to 700mA with external resistor)

The IC handles up to 50v, there's a voltage drop on it of about 2 volts, so the maximum led voltage would be around 48v, in theory. If your leds are rated for 32-34v, then your input voltage should be at least 36v.

In practice, with the best cooling you could give to this chip, the maximum power it can dissipate is about 4 watts, so you have to be a bit careful. For example, with 48v input and 34v on leds, there's going to be (48-34) x 0.35a = 14 x 0.35 = 4.9 watts.

So even though the IC can do 350mA and it can do 48v-50v input, that doesn't mean it can do it in any condition, or both 350mA and 48v input at the same time. With this particular example, you would use two of these in parallel with the adjust resistor set to 175mA, and then each IC would dissipate (48v-34v)x0.175a = 2.45 watts.

They're so cheap at about half a pound it each that it would make sense to just use 2 in parallel for each led instead of messing about with soldering other drivers on pcb, buying inductors, schottky diodes etc.

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

^^ @ Tom AH bugger, I was hoping the diagram below would work

[/URL][/IMG]

I guess the easiest to work with is to get the 12v LEDS and drive them off of a boost controller in series?

[tom66]

Re: Making a small high power LED array - best way to drive them?

You're really pushing it if you expect to get 5W dissipation in an MSOP, or a SOT-89 without failure. It's just far too much to be safe. You might be able to do it with a very good PCB (with large copper pad) and maybe if you heatsunk the chip externally too but it's a bodge solution to a problem that should not exist. 2.5W might be more doable, but it's still going to get very hot. Both of those drivers are buck converters anyway, they are not linear regulators and WILL require a PCB.

You should probably look at an LM317T if you want to do it properly (but ok with wasting power), you can build that without a PCB by mounting it on a large heatsink. It is only 40V rated to ground, but you can float it, which means it can drop up to 40V across it safely.

You can bolt it to same heatsink as LEDs.

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

Hmm I'm willing to learn and maybe its because its really late here 12:30 but some of that goes well above my head

The LED lights only cost 4.99 for five, am I right in thinking that if I bought the 12v versions = Voltage 9-12v current = 900ma

I could just simply run them off of a dc/dc boost converter in series @ 12v - 2.7 amps, well within spec with an off the shelf psu and not having to worry about different I/V of chips?

Thanks

[mariushm]

Re: Making a small high power LED array - best way to drive them?

No.

So let's recap.

A led has a forward voltage and a current. The forward voltage is the minimum voltage at which the led starts to work. The current is the maximum current it's designed to run at, before the color starts to go out of specs, or you risk decreasing the life of the led.

So when the page says forward voltage 9-12v, that simply means that the led may start working at a voltage between 9-12 volts. You can power it with a bit higher voltage, as long as you make sure to keep the current below the specified value.
So if you're lucky, you could get a batch of leds that will all light up quite well at 9-10v, in which case you can use your 12v adapter without boosting the voltage.

Now, when you connect several leds in series, the forward voltage of the leds gets added up and the current remains the same. So, if you have 5 leds with forward voltage of 9-12v and current of 100mA, if you put these in series you'd need a power supply of at least 5 x (9-12v) = 45-60v and a current of 100mA.

If you put leds in parallel, the minimum voltage is the forward voltage of a single led but the current gets multiplied. So you'd need a power supply that can do at least 9-12v and 5 x 100mA.

But, this method is not recommended because if a led breaks down, the power supply still gives those leds 500 mA, but now this 500mA value gets split between 4 leds and each led now has 120mA, instead of the maximum recommended of 100mA.

A best of both worlds method is to limit the current going through each led, then put the leds in parallel.

As an example, if you go with those leds that say forward voltage 9-12v and 100mA, you could just assume all leds will run just fine at 11v so then with a 12v power supply you can do the math and determine a suitable current limit resistor for each led.
V = I x R => 12v input - 11v forward voltage = 0.1A (100mA current in led) x Resistor

Resistor = 1/0.1 = 10 ohm.
The power dissipated in the resistor would be IxIxR = 0.1x0.1x10 = 0.1 watts, so a 0.25w rated 10ohm resistor would be enough.

So now you have +12v ----- 10 ohm/0.25w ---- 11v/100mA led --- GND and you can parallel several of these and run them from 12v adapter.

If those 12v leds are higher current, just use the formula to pick suitable resistor.

This resistor method doesn't work well with those 32-34v leds however, hence why I recommended actual led drivers.

[tom66]

Re: Making a small high power LED array - best way to drive them?

If they're in series, no, you'd need 60V or so at 900mA.

You can put them in parallel, but you need ballast resistors if your supply is not current regulated... These will reduce output power slightly.

Example calculation for ballast:
LED V = 10V (goal)
LED I = 750mA (guess)
Supply V = 12V

So, you need to drop (12V-10V)=2V at 750mA, resistor=2.7 ohms. (R=V/I)

This resistor will dissipate (0.75A * 0.75A * 2V) = 1.12W

So choose a 2W or 3W wirewound 2.7 ohm resistor.

LED power will be 7.5W/LED.

Whatever you do, don't connect them directly without the resistor.

edit: marius beat me to it :P.

[tom66]

Re: Making a small high power LED array - best way to drive them?

BTW,

The reason for the ballast resistor is less about the LEDs failing, and more about the fact that any diode has a temp coeff which basically tells you how much its forward voltage varies with temperature. Typically it's on the order of -3 to -20mV/°C for LEDs. Basically, this means for each degree C the LED increases in temperature by, its forward voltage will decrease by 3 to 20mV. This is called a negative temperature coefficient and most diodes, including LEDs, have such a behaviour.

Let's assume best case and say you start off with LEDs matched within 0.01V and they all conduct nearly the same current. Let's say they're within a degree of each other at the start.

Now, let's assume the LEDs are starting to warm up. One begins to get a bit hotter than all the rest. This means that its forward voltage will decrease slightly. For arguments sake, let's say it gets about 10°C hotter - not unexpected. So what happens to its forward voltage? Well, in the best case, it drops by about 30mV. Worst case around 200mV. In the best case, the LED would probably conduct about ~1% more current than the others - barely visible to the eye. But, it would also get a little hotter. This now leads to its forward voltage dropping further.

It doesn't take much of a stretch of the imagination to see that if the diodes aren't perfectly thermally matched and made exactly the same (which is all but impossible, especially for cheap LEDs from eBay) that one LED will get a lot hotter than the rest, taking a lot more current, and will subsequently fail. The most common failure mode for LEDs is open circuit. So, now your other LEDs are ripe for the picking - ready to fail in the same manner...

It gets worse. LEDs also drift with age. The forward voltage can increase or decrease, too. Matching LEDs in parallel is a nightmare.

The ballast resistor essentially fixes the LED's voltage and current at a defined point on the curve. If the LED does try to decrease its forward voltage, the current will increase, like before. However, unlike before, there's not a virtually unlimited power supply behind it, so the current won't increase substantially. The LED will hit a defined operating point once warmed up, and it will stay there.

So, a resistor is essential for a power supply not current limited.

[mcfly666]

Re: Making a small high power LED array - best way to drive them?

Sorry guys, I got the V/C mixed up when accumulating. The idea of using the dc/dc boost controller was so that I can regulate current or voltage (does both i think) this is the digital one, you can get a cheaper one with pots on which does the same thing.

http://www.ebay.co.uk/itm/DC-DC-400W...item19e1ad35b8

I don't know if the above counts as regulated current to run in parrallel tom?

So with a series on 12v leds, I could go say 11v = 63v and 900ma all running off a boost controller and if one led breaks then the whole array will go down (quite safe)?

Anyway - thanks for all your help guys. I'm handy with a spanner/soldering iron but the technical/brainy science behind these things I'll forever be playing catchup on + atm I've had a few so i'm a little 'clouded'

Good night guys

[tom66]

Re: Making a small high power LED array - best way to drive them?

Just a note, no way on earth that thing can do 400W so I would take all of its specs with a very heavy pinch of salt.

You could try 63V at 900mA but that is still about 50W which I think might be a challenge for that unit!! It might look beefy to some but just look at the thin wire on the inductor. The 12V input current of over 5A must flow through that. Typically multiply that by 2x because I doubt it will be in continuous conduction mode. In addition the output capacitors are ridiculously undersized.

Glass fuse is a nice touch, so if it fails it will fire glass everywhere. Free fireworks** (of blood.)