Same(?) led different intensity, different distributor

[martin321]

Hi,

I have made several big seven segment led display with 230 plcc4,Cree CLM4B-RKW/AKW.
Bought led from Farnell for the first display, when they where out of stock I ordered from Mouser.com.

The order number is the same, they link to the same datasheet.

http://no.farnell.com/cree/clm4b-rkw-cwaxbaa3/led-red-1-6cd-624nm-plcc-4/dp/2430038?CMP=i-ddd7-00001003

http://no.mouser.com/search/ProductDetail.aspx?r=941-CLM4BRKWCWAXBAA3


The leds from Farnell is much brighter, if I connect it directly to the psu at 2.4V it draws 300mA.
Doing the same to the one from Mouser, with 2.4V it draws 23mA.

So now one of the six digits are a lot brighter than the rest and gets to warm.

Who has sent the wrong leds or is there something I am missing here?

Martin

[Ian.M]

Characteristics    Color      Symbol  Condition   Unit Minimum Typical Maximum
Forward Voltage    Red/Amber  VF      IF = 50 mA  V            2.4     3.0

You cant run them from a constant voltage supply, as batch to batch the Vf can vary considerably.  You need a constant current driver circuit, or if you don't care about efficiency, a higher voltage DC supply such that a 0.6V variation in Vf is negligible + a current limiting resistor for each individual LED or series string.

[rstofer]

Those darn LEDs just don't behave!  All that Ian.B has said is correct, you need a ballast resistor to control the current in each LED as Vf can vary quite a bit.  The thing is, you won't use the same value for every LED even if Vf matched because of 'bins'.  If you read the datasheet, you will see where Vf can vary by 25% but luminous intensity can vary by 100%.  At final test, the devices are 'bin sorted' and coded accordingly.  Some LEDs are just brighter than others.

http://www.cree.com/led-components/media/documents/CLM4BRKWAKW1087.pdf

[martin321]

I might have worded myself poorly, so I have added an image and the way the leds are connected.


The first batch of leds from Farnell I soldered on to the seven segment pcb. They all are the same intensity.
Second batch of leds from Mouser is soldered onto the next five seven segment pcb's, these are all the same intensity.

When I connect them together and apply 12v to all of them, the one from Farnell is a lot brighter and gets to 100 degrees C fast.

Voltage is low on the image attached but one digit is a brighter.
Is this normal led behavior?

Martin

[Benta]

First, 1 ohm at 300 mA is too low to balance the LEDs properly, you need more than just 300 mV drop to compensate for unequal VF.
Second, are you sure all the LEDs are from the same bin?

[Ian.M]

Its 1R at 300mA/6. That's 50mV so not even a ***** hope of balancing them and its probably low enough that the LED array can destroy itself by thermal runaway leading to current hogging.

[Benta]

You're right, I missed that it is the whole array and not per string.
I would opt for around 1 V over the resistors.

[martin321]

I don't understand how balancing the leds can have anything to do with the brightness difference between individual digits?
All 230 leds that makes up one digit have the same brightness.
They are all balanced the same way with 1ohm resistor.
If this is wrong so be it, but how can this make one digit a lot brighter than the 5 other digits. The difference is blinding. About 1.2 amp at 12v from 5 of the digits and about 2.4amps from 12v on the first digit I made with leds from Farnell

It’s not individual leds that make up a digit that’s not the same brightness, its different digits compared with one other digit.

Is it possible that either Farnell or Mouser sent the wrong leds?

Martin

[wraper]

You are doing it wrong, first of all 12V is not enough for 5 leds in series. If LEDs have Forward voltage of min spec of 2.4V, then 12V is barely enough (2.4 x 5 = 12), but it must be 15+V to cope with maximum possible Forward Voltage as per spec of 3.0V (3x5 = 15). Secondly 1R resistors are too small. Thirdly, don't mix LEDs from different batches as their brightness will noticeably vary.

Is it possible that either Farnell or Mouser sent the wrong leds?

Possible? Yes but unlikely. Because you are using them incorrectly in the first place.

[Zero999]

The leds from Farnell is much brighter, if I connect it directly to the psu at 2.4V it draws 300mAh.
Doing the same to the one from Mouser, with 2.4V it draws 23mAh.

No they don't. That's meaningless. mAh is a measure of charge, normally used for stating battery capacity, not current. You mean mA. I know it may seem like I'm being picky but getting the correct units is important.

As everyone has stated above, the current sensing resistors need a greater share of the voltage to stabilise the circuit. Using strings of four with a constant current source would also help.

[Audioguru]

I have seen that very stupid LED Wizard before make the same mistake. It should be banned from the internet.

[rstofer]

Numbers are always a good thing.  Measure the current through each string and measure Vf for all of the diodes as they are currently connected.

But even if they match electrically, the luminous intensity can vary by 100%.  I linked the datasheet above, just read the specs.  Look at the bin code on the devices and see what you have.

Everything mentioned above about the ballast resistor is quite correct.  The supply voltage is too low and the resistor value is too low.  Furthermore, if you want every string to match, in terms of luminosity, you can expect the string current to vary somewhat and that may change the resistor value slightly.

I might go for 24V and plan to drop half of it across the resistor.  That would require a 1W resistor of about 270 Ohms (assuming 50 mA per string).  220 Ohm would be another candidate.

The voltage might be a little high and a case could be made for something in the middle.  Maybe 16V with 4V across the resistor at 50 mA.  Somewhere around 82 Ohms 1/2W would work well and waste less power going out as heat.

[Ian.M]

put two of the old, and two of the new LEDs alternately in series with a 100R resistor to limit the current then crank up the PSU output voltage *SLOWLY* from about 10V.  If they are near enough the same brightness with the same current going through them, constant current drive will fix your problems.  Otherwise you will have to implement per-digit brightness control.   How you do that depends on how you are driving the digits from whatever MCU or logic you are using.  It could be tweaking the duty cycle for each digit if you are multiplexing them, or maybe PWMing them individually, or different (or extra) series resistors for the segments of the bright digit.

[mariush]

Ok... let's look at the datasheet: http://www.farnell.com/datasheets/2047884.pdf
You have there:

Forward voltage at If = 50mA  typical 2.4v , maximum 3v
Luminous Intensity at 50mA  : minimum 1120 mcd , typical 1600 mcd

The magic words are TYPICAL and MAXIMUM...  Basically they're saying no matter what, the leds will turn on fully with 3v. 

The majority of them will turn on fully with a forward voltage as low as 2.4v but that's not a guarantee, so you really shouldn't use that value. 
Also, a couple of important things :

1. LEDs may turn on partially at forward voltages lower than the specified typical forward voltage but would behave sort of like having built in resistors.  So for example, a led may need 2.4v to be fully on and let 50mA go through it, but at only 2.2v it may allow some current, let's say only 20mA... so in turn it will limit all the other leds in series with it to 20mA, making all that strip of leds less bright.

2. The forward voltage of LEDs goes down as they warm up. A particular LED may be fully on at a typical forward voltage of 2.4v but as it heats up to maybe 40-50 degrees Celsius, the typical forward voltage may drop down to 2.3v  - so considering this and what I said at [1] , imagine a LED initially limiting the whole strip to less than 50mA but as it heats up, its forward voltage lowers a bit and may turn on completely and allow the full 50mA to go through the series of LEDs and then the series of LEDs will become bright.

So in your particular example, considering you already have the LEDs on circuit boards and everything in place... it would probably be cheaper to just replace the 12v power supply with a 18-20v power supply like a laptop adapter.
With 18v in and 5 leds in series, assuming you want to limit them to 50mA .. let's just say you'll have an average forward voltage of around 2.5v so 5 leds would have a total of 5x2.5v= 12.5v  .. so now you can calculate your resistor value :

V = I x R  =>  18v - 12.5v = 0.05 x R => R = 5.5/0.05 =  110 ohm .. but to keep things simple, i'd probably just round that down to 100 ohms.  And the resistor would have to dissipate P = IxIxR = 0.05x0.05x100 = 0.25 watts, so you should use at least a 0.5w rated resistor.

So in worst case scenario, when the leds will all have a forward voltage of 3v, then you'd have  V = IxR => 18v- 15v = I x 100 => I = 3/100 = 30mA  and in the best case scenario of 2.4v, you'd have I = 6/100 = 60mA

Anyway.. rather than using resistors, it would make sense to use LED driver chips to control the current better.

[amyk]

No one has mentioned yet that LEDs are diodes, and thus have an exponential I-V curve. A small change in voltage can result in a huge change in current and intensity, and that combined with the natural variation between batches is why you're seeing such differences.

[james_s]

What it comes down to is that LEDs are current driven and you are trying to drive them from a constant voltage. The LEDs don't "draw" 300mA, you are pushing 300mA through them by using a ballast resistor that is far too small to behave like a roughly constant current supply. The only reason the first batch of LEDs is not burning out is that the forward drop is just high enough that your power source cannot push so much current through them but if you were to raise the voltage even slightly at some point all of your LEDs will behave this way.

[Audioguru]

The magic words are TYPICAL and MAXIMUM...  Basically they're saying no matter what, the leds will turn on fully with 3v.

Absolutely not. You never apply a voltage to an LED, instead you feed it a certain current and it sets its own forward voltage. Some LEDs have a low forward voltage and others have a high forward voltage.
If you use a simple series resistor to set the current then if the LED forward voltage is low then the current and brightness are high. If the LED forward voltage is high then the current and brightness with the same resistor are low.

[mariush]

The magic words are TYPICAL and MAXIMUM...  Basically they're saying no matter what, the leds will turn on fully with 3v.

Absolutely not. You never apply a voltage to an LED, instead you feed it a certain current and it sets its own forward voltage. Some LEDs have a low forward voltage and others have a high forward voltage.
If you use a simple series resistor to set the current then if the LED forward voltage is low then the current and brightness are high. If the LED forward voltage is high then the current and brightness with the same resistor are low.

I never meant to say "apply a voltage to a LED" ... I meant to say that if you have a led driver that would limit the current to 50mA, the voltage drop across the led could be up to 3v.  The manufacturer says maximum 3v, that's what I meant to say by that phrase.
Of course the forward voltage of each led will vary at a specific amount of current flowing through them and as they heat up and so on... using resistors is a very poor way to control the current, especially with so many unbinned leds in a string.

For that project, it wouldn't have hurt to do some manual binning of those leds and use even amounts of each binned category for those segments and potentially use the lowest binned leds for the digit separators

[neko efecktz]

how about this...
measure the voltage drop across the resistor in one led chain in the "normal" brightness display
then measure the voltage drop across each LED in that chain.
Do the same test in the bright display.
this will tell you how much difference there is in the two batches of leds.

[james_s]

It doesn't matter how much difference there is between the two batches, he's not driving them correctly. The way they are wired up is going to greatly amplify any difference there is.