Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

LED Lights

Status
Not open for further replies.

crazyhamm

New Member
I was looking to arange 120 3w LED's in a series, 100 3w RED LED's **broken link removed** and 20 BLUE 3w LED's **broken link removed**, however I am new to all this and I dont know how to power it.

So I have two questions, How do I power this thing? & Can I install a dimmer on something like this?
 
Last edited:
That's pretty insane, that's almost 750watts of LED's. First you'll have to find out what drivers you're going to use, those type's of LED's should optimally each over their own dedicated LED driver, and with that many it's a good idea. Run from a 5 volt power supply you're going to need a 5 volt supply capable of about 200amps minimum. How do you plan on heatsinking the LED's that's quiet a project to be working on.
 
My guess (and it's only an educated guess, as I haven't even looked at your LED auctions or a datasheet) is that you would be best to find a series-parallel arrangement that would run from a medium-voltage, high-current power supply (let's say somewhere in the neighborhood of 12 to 48 volts, regulated), with the appropriate current-limiting resistors for each series string. This will take a bit of back-of-the-envelope calculation.
 
Last edited:
I wouldn't trust these type of cheap chinese high power LED's in series without some serious study no back of the envelope calculations here =). There are no datasheets for these, what you see is what you get on the e-bay listing which specifies generic characteristics. The manufacturing variation in the LEDs will require characterization of the true VF drop at the tolerated drive current, and will still need to have heat sinks, series strings if you do so will have to be on the same heat sink to try to balance the diodes as well as possible.

If you don't have the ability to characterize the LEDs to match strings many of the LEDs will burn out or be dim, unless you derate their current by 25% or more, even then you'll still need drivers, and even if you derate you'll still get brightness variations without matched LEDs, you don't power LEDs like this with a simple resistor ballast and expect any kind of tolerable results, you'll need constant current drivers.
 
Last edited:
Ok so I've been using this site **broken link removed** to do the math for me.

Just please let me know if this is gonna blow up on me or not, just type these numbers into the site listed above.

12 Source voltage
2.4 diode forward voltage
700 diode forward current (mA)
100 number of LEDs in your array

PLUS

12 Source voltage
3.8 diode forward voltage
700 diode forward current (mA)
20 number of LEDs in your array

RUNNING off of a 12V 30A regulated power supply

So is this a failure or success?
 
Ok so I've been using this site **broken link removed** to do the math for me.

Just please let me know if this is gonna blow up on me or not, just type these numbers into the site listed above.

12 Source voltage
2.4 diode forward voltage
700 diode forward current (mA)
100 number of LEDs in your array

PLUS

12 Source voltage
3.8 diode forward voltage
700 diode forward current (mA)
20 number of LEDs in your array

RUNNING off of a 12V 30A regulated power supply

So is this a failure or success?


That calculator is seriously flawed. It often works, but in some cases it does not. In the first set, with the 100 off 2.4 V LEDs it is wrong.

The calculator has two problems. The main problem is that it does not have any minimum voltage difference between the LED voltages and the supply voltage. The smaller the difference between the supply voltage and the LED voltages, the worse the current control will be. You cannot take chances with 700mA LEDs.

Secondly, the calculator will not give a value less than 1 Ω. In fact, if it calculates less than about 5 Ω, it is probably time to find another solution, but there is no warning about that.

The problem you will get is that small variations in voltage will give large variations in LED current. With a total of around 20 A flowing, you will get small variations in voltage due to voltage drop in the wires.

You would be far, far better to use an active current control circuit for each string of LEDs.
Something like the circuit in **broken link removed** but with a smaller value resistor, a larger transistor, and a heatsink, would be a start.
 
I've been working on a 100 LED array for a friend of mine's fish tank which uses a 50/50 mix of cool white and royal blue Luxeon Rebels. I looked into using multiple buck converters with a 24V power supply but the costs were adding up so I investigated using 120VAC mains power as a cost saving measure. Here is what I came up with.

Constant Current Controller Specs (per channel):
Based around the Supertex HV9961: **broken link removed**
120VAC input (fused)
Designed output of 90VDC @ 750mA
Fused output stage to protect the LEDs. Controller IC also supports a hiccup mode.
High current mode = 0-750mA
Low current mode = 0-150mA

There are 2 boards each with 2 channels which are easily driving 25 LEDs in series a piece. All 4 channels and high/low selection can be controlled separately to allow for any kind of mixing you would like. The interface is optoisolated so there is no risk of exposing yourself to dangerous voltages while the controller is running.

Here is a picture of the constant current controllers. I put a Comcast remote in the picture so you can get an idea of the size. Since these were prototypes, I made them on perfboard but I could likely make them a fair amount smaller if I layed out a proper circuit board. Those 2 boards controll all 100 LEDs.
**broken link removed**

Here is a picture of the 100 LED array lit up at full power. I had to turn the exposure way down on my camera to get a proper picture. In real life they are extremely bright.

**broken link removed**

It would be trivial to adjust the components in the controller to take a 220VAC input if needed and the output stage can be adjusted to almost any current you want.
 
I should also note that the total BOM for the 2 boards is under $50.

I have a future upgrade planned that will add a 400V PFC front end to the controller that will allow for either much higher string voltages or physically smaller components for the current 90V string voltage.
 
Have you measured the true current going through each LED in any particular string to see how well they're current sharing? There's no way it can be even but I'm curious how many LED's are over/under current and by what percentage from nominal current. I'm very curious because of the longevity aspects when the diodes start to age.
 
The array of 100 are segregated into 4 isolated strings of 25 series connected LEDs. Each string has its own current controller so all LEDs in a single string have identical current. When each string is set to 100% brightness, the current differences between strings is about +-3% which is mostly due to tolerances of the sense resistors and the inductors.
 
phalanx, sorry my mistake, I used the completly wrong word. I didn't mean current sharing, I meant power sharing. Measure the forward voltage drop of each of the diodes in a string with a meter that can read millivolts (or better) I garuntee no two will be 100% identical, meaning the apparent brightness will be lower.

You could measure this with the camera but you'd have to take a photo head on of each individual LED with enough translucent material in front of it to not saturate the camera element, you'll be able to determine the apparent brightness from the RGB value, they will not all be the same.
 
As I get some time, I'll check out some of them for you. I know none of them will be exactly the same but they should all be reasonably close since Lumileds has a very tight binning system for their LEDs which is based around color and minimum flux at specified currents.
 
Last edited:
Keep in mind, reasonable close in diode forward voltage can mean a drastic different in diode light emission, it is NOT a linear scale, and the diodes aged directly related to their heat vs light output. Just because you can not SEE the difference at a quantifiable level with your eye does not mean there isn't one severe enough to figure into the maximum life expectancy of the series string.

See what Lumiled says about 25 long series strings... Find me a data sheet that even mention series strings that long..
 
Last edited:
Keep in mind, reasonable close in diode forward voltage can mean a drastic different in diode light emission, it is NOT a linear scale, and the diodes aged directly related to their heat vs light output. Just because you can not SEE the difference at a quantifiable level with your eye does not mean there isn't one severe enough to figure into the maximum life expectancy of the series string.

See what Lumiled says about 25 long series strings... Find me a data sheet that even mention series strings that long..

I really don't see what is wrong with long strings. The application has close control of the current, which is what LEDs work with. Whether the supply voltage is 100 V for a long string or 12 V for a short string, many constant current circuits can supply far more voltage than the maximum forward voltage of an LED, but it does not matter because the correct current is supplied.

The only issue with long strings is that there are higher voltages around and you have to be more careful with the insulation.
 
Phalanx, good job!,

I like series much better than parallel. We know the current is the same.
Starting out with 110/220 AC or 150/300 DC it seems batter to have a 75 volts LED string than 25 current sources at 3 volts.
This design is clean, (small parts count).
 
Keep in mind, reasonable close in diode forward voltage can mean a drastic different in diode light emission, it is NOT a linear scale, and the diodes aged directly related to their heat vs light output. Just because you can not SEE the difference at a quantifiable level with your eye does not mean there isn't one severe enough to figure into the maximum life expectancy of the series string.

See what Lumiled says about 25 long series strings... Find me a data sheet that even mention series strings that long..

Forward voltage doesn't mean much in terms of light output when the LEDs are binned according to their minimum flux at a specified current. So long as the LEDs are from the same bin (which they are in my case), they will all be reasonably close in terms of flux when run at the same current.

Power dissipated is a different issue. The forward voltage of the emitters can vary even within the same bin which could result in more heat and reduced longevity. I've tried to mitigate these effects in a few ways. First is all calculations were based around the worst case numbers available in the datasheets. Second is the emitters/stars are mounted on massive extruded aluminum heat sinks using screws and heat sink compound. The heatsinks are much larger than what is necessary to keep the LEDs at a proper operating temperature. They have the added benefit of being rigid enough to span a large fish tank while only being supported on the ends. Third is the controllers are set to a maximum of 750mA which is lower than what they are able to tolerate. This helps keep the temperature down. Fourth is the array will likely only be used at 50% intensity due to how amazingly bright it is. This drops the current down to less than 400mA which is where you start seeing noticeable improvements in efficiency and heat output is greatly reduced.

I've let the array run at full power for a few days on end and the heat sinks get quite warm but you can still touch them with your hand without any discomfort (which means they aren't even close to being hot). At the likely operating point of about 50% power, the heat sinks are only slightly above room temperature. Cooling fans on the heat sink are not necessary do to their size but my end user will be adding them anyway which will only improve the thermal characteristics of the system.

In this type of temperature environment, Lumileds testing (which is available on their website) shows lumen maintenance at better than 85% at the 50,000 hour mark. My biggest concern is how everything will hold up being mounted above a salt water environment. My end user understands the corrosion issue and will be accounting for it in the design of the hood going over the tank.
 
Last edited:
Phalanx, good job!,

I like series much better than parallel. We know the current is the same.
Starting out with 110/220 AC or 150/300 DC it seems batter to have a 75 volts LED string than 25 current sources at 3 volts.
This design is clean, (small parts count).

Thanks Ron!

Not wanting to have (or pay for) a dozen lower voltage current sources is what led me to go this route. Right now this is design is optimized for 120VAC rectified to 170VDC and my component selection will not tolerate a 240VAC source. This isn't an issue for my end user but I will be improving it in the future. The controller IC i'm using can tolerate up to 450VDC. I'm currently testing a 400VDC PFC front end with a universal input (90-250VAC) which could give the individual controllers much more flexibility in terms of source power and overall string voltage.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top