LED Power Questions

[TropTrea]

For some reason I seem to realy be missing something with all the treads here on LED lighting. Leds have been around for ages only today we do have some higher powered ones. I see threads after thread that seems to make the controling of these LED's more and more complex.

Looking at a basic LED yu virtualy have a simple diode which emits light at much more effecient ratio than most other type of light. Basicly the old diodes could simply be paired up with a simple resistor to control the actual voltage they were using and be run off a simple $5.00 DC converter or even a battery. Actualy the only big difference is that instead of the old diodes using 50 ma we are now talking power going as high a 3 Amps.

As I look at it why cannot these new diodes simply be powered with a simple Regulated DC power source that can be adjusted to the proper voltage to maintain the correct current on these new LED's. After going through many LED spec sheets they seem to operate in the 3 Volt range therefore a simple 12V 3 Amp supply would easily power 4 LED's. For banks of even 60 LED's a bigger power supply of 45 Amps @ 12V should work equally well. That is basicly only a 540 VA supply.

Am I missing something extremy basic here in my thoughts? Are these new LED's so sensative that a power spike from say 3.1 volts to 3.5 Volts would burn them out?

 

[Gorgok]

First, you want to run the LEDs off a constant current, not voltage, as voltage at any given amperage will vary with temperature and the die itself. In fact i think if you run them at a specific voltage, and the temperature rises enough you could have a thermal runaway and start pumping more and more current through the LEDs until one fails.

Also, if all you want is the LEDs to be either full on, or full off, there are plenty of cheap ways to do it. But, why do that? Most people would rather not have that shock just because its not very pleasing (even MH lights turn on slow, off is still a shock), and for some jumpy fish it can lead to further jumping. Dimming lets you control how bright the LEDs are, how fast they get there, etc... I would not go with LEDs if i just had them on/off on any display (my ATS lights will be on/off).

Finally, i don't know enough about very crude ways to power LEDs to really know if that would be efficient, but i don't think just putting resistors or whatnot to keep current in check is gonna be more efficient than a driver regulating it. Wasting power to heat is basically backwards of what LEDs are designed to do.

Basically, why make it crude (simple) when a more elegant solution (complex) can do the job better?

 

[der_wille_zur_macht]

For some reason I seem to realy be missing something with all the treads here on LED lighting. Leds have been around for ages only today we do have some higher powered ones. I see threads after thread that seems to make the controling of these LED's more and more complex.

Looking at a basic LED yu virtualy have a simple diode which emits light at much more effecient ratio than most other type of light. Basicly the old diodes could simply be paired up with a simple resistor to control the actual voltage they were using and be run off a simple $5.00 DC converter or even a battery. Actualy the only big difference is that instead of the old diodes using 50 ma we are now talking power going as high a 3 Amps.

As I look at it why cannot these new diodes simply be powered with a simple Regulated DC power source that can be adjusted to the proper voltage to maintain the correct current on these new LED's. After going through many LED spec sheets they seem to operate in the 3 Volt range therefore a simple 12V 3 Amp supply would easily power 4 LED's. For banks of even 60 LED's a bigger power supply of 45 Amps @ 12V should work equally well. That is basicly only a 540 VA supply.

Am I missing something extremy basic here in my thoughts? Are these new LED's so sensative that a power spike from say 3.1 volts to 3.5 Volts would burn them out?

Look at the VI curves for some of these LEDs. The difference between 3.1 and 3.5 v can be the difference between "barely on" and "exploded."

Most of what you've said is technically correct, but it misses a major point. HP LEDs are designed to be run at the very fringe of their capabilities. A 20mA gumdrop LED acting as a power indicator on your DVD player has a vast range of acceptable operation. We don't care if it's emitting 8 lumens or 20 lumens, as long as it's "on." We don't care about efficiency, since it's using a miniscule amount of power anyways. We don't care about hitting an exact wavelength or spectral distribution, and we don't really care if the light goes everywhere or only in a specific direction. With HP LEDs on a fish tank, we care very much about all of these factors.

So, yes, the theory of operation is the same. You can drive them with similar methods. BUT, if you care about meeting certain criteria, over a wide range of operating and environmental conditions, you really need to treat any LED carefully, regardless of what type it is. If you ignore these goals, you're better off just sticking with MH or T5.

First, you want to run the LEDs off a constant current, not voltage, as voltage at any given amperage will vary with temperature and the die itself. In fact i think if you run them at a specific voltage, and the temperature rises enough you could have a thermal runaway and start pumping more and more current through the LEDs until one fails.

This is true. Operating an LED on "constant voltage" means you really have no idea how it's behaving, and you have no way to control it's behavior over any range of operational or environmental conditions.

Simply put:

why cannot these new diodes simply be powered with a simple Regulated DC power source that can be adjusted to the proper voltage to maintain the correct current on these new LED's.

You can, theoretically, use a constant voltage source to power an LED if you carefully match the voltage to the LED's specifications, but it's not gonna meet any specific set of operational goals over any realistic range of operating conditions. In practice, it would be extremely hard to even get this to work at a useful output for the LED, because differences from LED to LED and environmental changes inherent in operating an LED (it's gonna heat up over time) will mean that your "target voltage" isn't a single specific value you can look up or determine ahead of time.

Finally, i don't know enough about very crude ways to power LEDs to really know if that would be efficient, but i don't think just putting resistors or whatnot to keep current in check is gonna be more efficient than a driver regulating it. Wasting power to heat is basically backwards of what LEDs are designed to do.

The "use a resistor as a current regulator" approach can be pretty much as efficient (or more) than the drivers typically used in this hobby, but it can also be woefully inefficient. Getting the efficiency takes experimental effort AND careful upfront design, while getting efficiency with an off the shelf driver just means you wire it up and plug it in. At any rate, a resistor is obviously passive, so you've got no way to actively control or adjust the current, i.e. no dimming.

 

[TropTrea]

To give you a wough idea what Im presently looking at is "moon Light System using 6 or 8 - XPCROY-L1-R250-00801 - LED, HIGH BRIGHTNESS, ROYAL BLUE .

From the dats sheet I get a maximium of 3.9 Volts forward aand a 500 ma. maximium. It also lists typicial operation is at 350 ma @ 3.4 Volts.
3.0V = 75 ma
3.1 V= 140 ma
3.2 V= 200 ma
3.3 V = 270 ma
3.4 V = 330 ma
3.5 V = 410 ma
3.6 V = 500ma

If you have a power supply with a maxium voltage runnit 3 leds in series and aming at the 330 ma for a maximium then a simple power suppy with a maximium output of 10.2 Volts I wee as very possible option. It would be very easy and simple to build a simple voltage regulator that could have a voltage adjustable between 9.5 Volts and 10.2 Volts capable of handling 700 ma or 8 VA.

Where I see the fear comming to some individuals is that the actual resistance of the LED's is not a constant. As an example from the above data we see that at 3 volts the effective resistance is about 40 ohms at 3.3V 12 ohms, and at 3.6V it is about 7 ohms. But actualy this feature could be used as saftey feature to prevent the leds from burning out on a so called thermo run away.





gorCog mentioned sudden turn ons and offs as a factor but you do need a micro controller to handle that issue either. Now if your using pure LED's for your lighting had anyone thought of using a simple sequence timer cuircuit to gradualy turn them on. There are some easy ways where you increase the voltage in steps or actual step the banks of light on in groups.

 

[der_wille_zur_macht]

Again, you're not really proposing anything that's not theoretically possible. but:

1) It's gonna be harder than you think. You cannot just look up the Vf and do a design based on that. LEDs vary considerably from individual to individual. Also, they'll vary over environmental conditions. If you set up your rig and "tune" the voltage when the LEDs have been on for 5 minutes, you may find the power level is totally different once they've been on for 2 hours and are warmed up.
2) It's not really "saving" you any complexity or cost compared to doing this "right." Sure, you could build a voltage-regulated power supply that would nail a particular output voltage. BUT, you can just as easily build a "correct" constant current power supply, that'll actually regulate the current correctly despite variations in the LEDs or environmental conditions.

Check out the big DIY driver thread. You can build a driver that'll take unregulated DC and power your proposed number of LEDs for maybe $5. It'll be extremely efficient, constant current, dimmable via pot or microcontroller, and won't have any of the issues you'll encounter with your proposed voltage regulated approach.

 

[TropTrea]

If your running the diodes three in series using and 2 groups in parallel you coul have a current controling power supply that would be adjustable from 400ma to 1 amp using the diodes I mentioned earlier. Or as I was suggesting one with the voltage adjustable from 9.6 Volts to 10.8 Volts.

With either senerio as I see it if your set to run at 10.5 Volts or 820 ma your relying on all three diodes in that series running the same resistance. As was mentioned if one diode ran away and dropped its effective resistance is say half then you would have an imbalance in the load where voltage going to the other diodes would increase causing there curent to increase.

Using the above as an example running normal at 3.5V each and using 400 ma on one branch if the resistance decreased in half on one diode going wild with a voltage controlled power source you would have 4.2 volts across two diodes and 2.1 volts across the one that went wild. If they are that touchy on excessive voltage burning them out you would fry the two good diodes.

Now looking at a current controlled supply you are safe in this senerio since the power supply will never go over 820 ma. However if you have the two bank of Diodes are in parallel to each other the power supply will adjust the voltage to give 820 ma total and one bank may be drawing 500 ma while the other is drawing 320 ma. If a diode in one circuit actualy burnt out to an open you would have the 820 ma on the other set of three.

So basicly unless one goes either all parallel or all diodes in series I see little difference if one goes with a voltage controled or current controled power supply. With them all in serries a current supply would be be more idea. Whil;e if they are all in Parallel the voltage controled supply would be more idea.

Now when you get into some of bigger arrays of LEDS like for the main lighting on an aquarium you can be talking about 30 + LEDS. running them on an all parallel circuit would mean a power supply running 6 to 12 Amps at about 3.0 to 3.5 volts or if they are in series 200 ma to 410 ma with 90 to 105 Volts. Finding a basic power supply for either might be a pain with these specs.

 

[Gorgok]

Most setups do run all series, as that is the easiest way on a current controlling driver.

And those who want to mix parallel strings of series LEDs, which is better i think than a single really high voltage string, you can always add fast blow fuses rated at the max any single string should take. In the event of a string failing you will only blow fuses instead of LEDs.

I still don't see why a voltage controlled PSU would be better in the case of an entirely parallel LED setup. Say you get a 3.3V 0-40A PSU (21-30A is more realistic for 30 LEDs) wouldn't you run the risk of putting a whole heap of current down one hot LED, only getting hotter with the extra current until it fails. With some bad luck the extra heat it puts into the heat sink also starts the others going over the edge.

Overall, I don't know if the $1.90 i spent on my Cat4101 chips is too much vs the 'simpler' method. Sure i still needed a PSU (24V 14A) but the simpler method does too.

 

[der_wille_zur_macht]

So basicly unless one goes either all parallel or all diodes in series I see little difference if one goes with a voltage controled or current controled power supply.

That must be because you're not reading what we're writing. There's a HUGE difference, both in the theoretical sense, and the practical sense. There's NO good reason to spend effort developing an incorrect supply (one that provides constant voltage) when it's just as easy to develop a correct supply (one that provides constant current).

With them all in serries a current supply would be be more idea. Whil;e if they are all in Parallel the voltage controled supply would be more idea.

The constant current supply is always the most ideal solution. Even if you put them all in parallel, the voltage required to maintain a particular operating current will change during conditions present in normal operation over the lifetime of the LED. If you base your target voltage on the current an LED pulls when you've first turned it on, it'll be WRONG once the LED warms up. If you base it on what the LED pulls on your bench when it's warmed up, it'll be WRONG when you put the fixture above a warm fish tank. If you base it on a cold day, it'll be wrong when it's warm out. If you base it on when the LED is brand new, it'll be wrong two weeks later when the LED has broken in. And so on. Constant voltage is just a bad idea for HP LEDs, end of story.

I appreciate the effort and thought you are trying to put in to solving this problem, but it's one that's already been solved, really well, and your approach has some significant flaws compared to the typical solutions.

Series vs. parallel does introduce some interesting challenges, but those challenges are independent of the way you choose to drive the array. And, the problems you've raised are also all fairly well solved, if you search around some of the threads.

 

[TropTrea]

Okay lets leep at the current controled supply which I agree is no more expensive to build but generaly harder to find rebuilt at a reasonable price.

If we have 6 of the LED's I earlier mentioned they are all in series with the curent supply set for 400MA. Typicialy the circuit will be running at roughly 21 volts. Each Diode will have an effective resistance of 8.75 Ohm's. or a total of 52.5 Ohms for the circuit. Now say one of the diodes does falter and its resistance drops to 4 ohms you now have a total resistance of 47.75 ohms. The circuit would then drop in voltage to maintain the 400MA to a total voltage of 19.4 Volts. But since the resistance of the LED's is dependent on the voltage and current flowing through them there would be an increase in the voltage across good diodes causing there effective resistance to go down as well. So in reality I agree everything would eventualy equalize out even if one of the LEDs were to get a complete short. Or if that faulty LED even went to an open circuit it would simply result in the bank going dead with no harm to the other LEDs.

The only issue I see here is finding that power supply with an adjustable current of between 150ma and 400 ma which will put out 18 to 21 Volts. With a larger bank of LED's this can even be a bigger issue.

Now looking a a Parallel circuirt again we are talking about a power supply that would adjustable from 900ma to 2.4 Amps running in the range of 3.1 to 3.5 Volts. Finding a reasonably priced supply at 5 Vots DC with this kind of Amperage is much easier. Only after the supply then either a current or voltage controler would have to be added.

Thanks for swaying my logic here. I do see the point of current control considering that on a do it yourself project the difference in cost between the two controlers is virtualy the same. However if buying pre built devices I do think the voltage control would be a less expensive way to go.

Yes you would have more risk involved but now with prices of High Output LED's running under $5.00 for a bank of even 6 your talking $30 for the LED's and as opposed to the cost of a 21 Volt power suppy of costing twice the as much as the LED's themselves.

Just a quick browsing gave me a MeanWell 5 Volt 4 Amp Supply that would easily handly 10 LED's in parallel for $17.80 compared to a 24 Volt .500 MA supply that only handle 6 LED's at the same price. But when you up to higher voltages the prices on the power supplies seem to skyrocket. So it looks to me like the economical way is simply keep the system running 24 volts or less.

 

[Gorgok]

My 24V 14A PSU that supplies my drivers cost $40. It can easily handle the load i give it of 60 LEDs and could do little more. This is one of the advantages of the DIY driver boards over pre-built LED drivers i think. Instead of each unit having the extra costs of a small AC to DC circuit in them you can just use a big one. For a smaller setup where you only need say one ELN-60-48 (7-14 LEDs) the $30 it costs as a standalone unit is better though.

 

[00Warpig00]

You probably want to avoid higher voltage LED driver's from a safety standpoint if you are doing a DIY. Many have been zapped by 120V AC, unless you have a health condition or are wet it's normally not lethal. In the event of higher voltage DC currents a 700ma hit at 50+ volts of DC can take you out. High voltage DC is NO JOKE!

Nick

 

[TropTrea]

Nick On that saftey end I do know where your comming from. I have personaly felt up to 480 Volts AC as well as 50 Volts DC. The 50 Volts DC did pup me down and realy shook me up more than even the 480 AC. But I had also seen people ending up in the hospital over 120 AC so it is a matter of not the power but conditions.

Going back to the overheating. This is another big question for me. The effeciency of a LED is based on the fact that they have a high light to heat ratio. Yet I see some LED set up with gigantic heat sinks and multiple fans. While other seem fairly bare bones.

My long range project is to light up a 40 gallon frag tank with LED's as an experiment. My thought was running channel Aluminium 1 1/2" wide 36" long with 16 LED's on it. Then have roughly 3 or 4 of these channels for the array. (48 to 64 LED's total) For these I'd using the higher powered LED running them coloser to 800MA each. But it comes out all in series 144 to 224 VOLTS DC at 800MA. OR cutting down to 24 Volt DC supply with banks of 6 LED's needing about 10 Amps. Running all Parrelel would be tough finding a 3.5 Volt Suppy at 50+ Amps.

Then there is the other option of running one big power supply and seperate Curent Regulators for each group that in series with each other. Would also give an option of adding a pot to adjust the Blue, Royal Blue and White LED's seperatly. Or even using a Timer Circuit to have a delay between the turn on times of the various banks.

 

[radicaltimes]

It's not the voltage that can kill you. On an AC circuit 20 ma can be enought to cause death. That being said the 24 volt ac on my furnace hurt more than the 277 volts I got hit by. 277 is one leg of a 480 circuit. I would assume that 20 ma on a DC circuit would be just as bad.

 

[TropTrea]

yes it is the current that kills. But there is some natural resistance in the body which makes lower voltages safer to some extent.

 

[kcress]

It's voltage that drives the current. Without voltage there is NO current. Stop thinking "it's the current" because it's the voltage that enables any current to flow.

The higher the voltage the higher the possibility of driving a high enough current to be dangerous.

The path you guys are logically starting down can be confusing and leads to the wrong conclusion. The current follows Ohms law. I = V / R

Notice the higher the voltage with a fixed resistance the higher the current. You cannot talk about humans being shocked/electrocuted without looking FIRST to the available voltage.


Talking about current being dangerous is like saying "falling is dangerous". Falling isn't dangerous - until the height of the fall is taken into account.

 

[der_wille_zur_macht]

Going back to the overheating. This is another big question for me. The effeciency of a LED is based on the fact that they have a high light to heat ratio. Yet I see some LED set up with gigantic heat sinks and multiple fans. While other seem fairly bare bones.

That's because the first several people who built DIY LED rigs went way overkill with thermal management, so people who didn't know better just followed suit.

My long range project is to light up a 40 gallon frag tank with LED's as an experiment. My thought was running channel Aluminium 1 1/2" wide 36" long with 16 LED's on it. Then have roughly 3 or 4 of these channels for the array. (48 to 64 LED's total)

That's WAY too many LEDs for a 40g tank, assuming you use typical best-of-breed LEDs.

For these I'd using the higher powered LED running them coloser to 800MA each. But it comes out all in series 144 to 224 VOLTS DC at 800MA. OR cutting down to 24 Volt DC supply with banks of 6 LED's needing about 10 Amps. Running all Parrelel would be tough finding a 3.5 Volt Suppy at 50+ Amps.

Then there is the other option of running one big power supply and seperate Curent Regulators for each group that in series with each other. Would also give an option of adding a pot to adjust the Blue, Royal Blue and White LED's seperatly. Or even using a Timer Circuit to have a delay between the turn on times of the various banks.

As you've probably seen in threads on the forums, it's very typical for people to have multiple drivers, or multiple current regulators on a single DC supply. You don't have to (and probably don't want to) think of a single supply for the whole array. Multiple drivers gives you more flexibility, more redundancy (single point of failure won't kill the whole array) and several other advantages.

I would suggest you go find and read the DIY driver thread, as it details some solutions to the problems you're trying to solve. If you can solder an LED array you can easily build your own driver from scratch, which will present a HUGE cost savings over buying a commercial driver - considering cost seems to be one of your important criteria, you might find something there you like.