Issues with LED’s

I am building a tail light assembly each tail light is 3 pieces that contain 11 rows of 5 LED's in series. I Built a circuit to measure the voltage drop on every LED and matched them using that voltage drop (Most of them Measure 1.90 to 1.98 V with a 225.9 Ohm Resistor in series from a 12.60V Power Supply). However once assembled I get some LED's that barely light and some that are very bright in each series of 5 LED's. Short of building the 55 Led panel and replacing every led that does not match light intensity is there a way to fix this. Did I just get a batch of junk LED's or is this a common issue.

I am using super bright Red 2.0V 50mA LED's.

I took 5 samples on a current limited power supply. I Set the supply to 2.0V and my results were as follows.

LED 1) 35mA
LED 2) 22mA
LED 3) 2mA
LED 4) 76mA
LED 5) 51mA

I am pretty sure this is an issue with the LED's but how do I either filter them out and get ones that match. Or where can I purchase ones that are guaranteed to match.

Using the voltage drop method produced more desirable results. However the array that I just built has 3 LED's that barely light, 5 LED's that are super bright, and about a dozen that are too bright. About 30 of the LED's are within and acceptable variance. These all measured 1.93 Volts on the circuit I built to test them.

Sounds like you're running these LEDs in parallel. The reason this doesn't work is they are nonlinear. Small increments in the forward voltage equal large increments in forward current--as such, LEDs a current devices, not voltage devices.

Each LED (or series string of LEDs) must have a resistor in series with it to, essentially, convert the voltage supply into a quasi current supply. OR use a current regulator in series (such as the CL2 [www.supertex.com/pdf/datasheets/CL2.pdf]).
A good arrangement, for your case, would be 10 strings of 5 LEDs in series with a resistor in series with each string (probably around 82 ohms).

Something like the attached:

As you measured there is a spread in how high voltage the LEDs need. With LEDs like LED3 all in a row there will be nearly no light and with LED4 in one row there will be extrem light. Either you sort them and put together hi current with low current so each row gets about the same current or you adjust with a resistor on each row.

I attached the my LED Board schematic to this.

I am measuring the Voltage of each LED within .01 on the same source. The voltage should be relative to the current. So if I put 5 LED's in series that measure 1.93 V in my test circuit they should all draw the same amount of current. Please correct me if I am wrong here. The problem is not between the parallel lines. It is within the 5 LED's in series. Would I be better off using my test set up and measuring the current on each LED. I would think that would not be any different than measuring the voltage drop though. I do not have Dim rows. I have 1 or 2 out of each series of 5 that are dim or bright.

For Example.

I put led's 1 through 5 in series. and say number 2 of the series is dim. I can replace the number 2 led and get a new LED that matches light output. But doing this with 66 rows of LED's and sometimes taking 3 or 4 tries to get it to work is extremely time consuming. The rest of the LED's in the series do not appear to change brightness when I change LED 2. LED 2 just comes up to meet the rest.

I am new to LED theory If there is such a thing. I do not understand. If I take 5 that measure 1.93v in my test fixture and put them in a string (AKA: series). Why would there be one that acts so much differently than the rest? By OHMs law they should all draw the same amount of current therefore put out the same amount of light.

As Steve has mentioned, you will need to arrange your LEDs into series strings. The typical forward voltage drop of a red LED is 1.8 volts, so 5 x 1.8 = 9.0 volts dropped across five LEDs. At 12 volts you will need to drop 3 volts across the series resistor: 3v / 50mA = 60 ohms. Don't forget to take into account the power dissipated in the resistor: 3v x 50mA = 0.15 watts, so a 1/4 watt resistor is perfectly adequate.

If you are running the LEDs of a vehicle battery, be aware that the battery voltage can reach 14.5 volts while charging. If you now calculate the resistor value using a required voltage drop of 5.5 volts (14.5v - 9v) at 50mA = 110 ohms, at 12 volts you will only supply about 27mA to the LEDs. If you use a 60 ohm resistor, at 14.5 volts the LED current will be 91mA which will probably shorten the life of the LEDs significantly.

The simplest way around this is to use the lowest battery voltage (12.0v), use your eleven strings of five LEDs in series at 50mA with a 60 ohm resistor, and supply the panel via a current limit set at 11 x 50mA = 550mA. The easiest current limit is a LM317 with a 2.2 ohm (1.25v vref / 550mA) resistor across the output and adjust pins, connect the battery supply to the Vin pin, and the LEDs supplied from the adjust pin. Don't forget to heatsink the LM317and don't ground the heatsink!

I am using an LM317 set to 11.5 volts. and plan on adding the current regulation after I nail down the LED Intensity issue. I have limited my current for now using resistors. I probably will end up changing the resistors to LED Drivers eventually. I am using a sequencer (555 and 4017),an XOR Gate, and A Power Mosfet to run all of the LEDs. For the issue I am referring to here those are taken completely out of the equation. I have other issue with the sequencer that I need address but they are not priority right now. Those issues are so minor they may get ignored.

These arrays are built and the current limiting is done. The issue is not in my supply it is in each group of LEDs. I have one panel the is acceptable. It took me 2 days to come up with a set of 55 LEDs that worked together well.

Each array when completed draws approximately 330mA The voltage on the LED Rows (Series Of 5) Measure between 9.7 and 10.2 v. The average being 9.94. These are 2.0v 50mA LEDs but I am limiting the current to 30. Using a 100 Ohm resistor in series with each row.

I came up with 82 ohms because I took into account that the voltage in a car's electrical system can go as high as 14 volts (unless that has changed in the newer cars -- been a decade or so since I visited this).

You can actually use an LM317 as a current regulator. Check out the application notes on the LM317 datasheet.

Yeah My car is far from new. It is a 69 Mustang.I will try to get a picture of one of the lights powered up to show what I am talking about later tonight. As for the current limiting. I will be doing the LM317 Voltage regulator with current limiting but I am working at the bench level for now. Once I get that dialed in I will add the regulation on the controller board. And STRESSSSSSS test it...

if even intensity is required, then I would definitely use current regulation in place of resistors. And, if you want a system that automatically applies consistent current to each series string, then mux the current from a single current source/sink.

These are all 1.92v LED's on my tester. There is a diffuser over the top of them to make it so I could capture with my camera.

It in the nature of LED manufacture that there are small variances in the voltage/current characteristic of individual LEDs. However, your setup does appear to have an unusually great range. This would only be expected if all the LEDs were operating at about their absolute minimum current.

What is the value and tolerance of the resistors you are using ? Have you used the same value resistor in each string ?

LEDs are never normally matched for brightness by your method. Usually the calculation is done by using an approximate value of the LED voltage (to one decimal place), multiplying by the number of LEDs in the string, then the resistor value is taken as the remaining volts over the string current (the normal LED specified operating current).

With the normal voltage applied to the whole assembly, check the current in each string by measuring each resistor's volt drop x R. Let us know what range of values you get.

I can't see an obvious fault in your string/series arrangement.
You may not be testing in the "normal way" but it does seem clear to me that you have a batch of highly variable LEDs.
I would approach the supplier for an explanation or replacement.
Failing that it seems to me you simply have to test & select for brightness.
That's going to be time consuming & its not a commercially acceptable approach.
For you tho', I'd suggest you connect in a single string as many as you can manage, & supply the test string with controlled current at what you believe to be the nominal operating forward current.
A bit of fiddling yes, but then you can just "aside the dim boys & the overly bright fellas also. - problem solved - repeat for more strings until all out of spec items are set aside.
Don't waste time attempting to use the out of spec LED's. If you need more buy more.
Compared to your time binning 20% is not unthinkable.

Don't rely on the forward voltage! It's not reliably consistent across parts/lots/temperature/age. Also, any LEDs that you connected in parallel might have been stressed and thus may have lost some of its/their brightness, And if none of that is "it", then consider the quality of the LEDs.

Also, are you sure they are 50ma LEDs? These look like T1 3/4, which are typically 30ma max. Though, you're muxing them, right? If so, then check the specifications for pulsing them--e.g. max pulse current @_max_ duty cycle, freq.

I have another 200 on order from a different supplier. This is a one shot deal. These are custom for my car. I Probably will not build another set like this anytime in the near future.Scrapping 20% or more is an acceptable loss to me. At this point scrapping all of them is an option I have considered. I have way more time into it than what the loss would cost. I have never tied large quantities of LED's in series parallel before and after this probably wont be doing it again. But you never know.

As for my testing method, This was an attempt to measure with a meter and get a matching set. It did work somewhat. I started out with 1000 LEDs and filtered out about 200 of them that were way out of range. Some measured as low as .8v and some were over 3.5v. I have pretty much come to the conclusion that I this is not a design issue but an issue with the LED's themselves. It would be great if there was a way to measure these to eliminate the bad ones. But I am going to have to just weed the bad ones out. The whole thing is geting benched until I get these new LED's. If I don't have the same issues then I may just redo the panels with LED's from the new supplier. The out of spec LEDs. will go where they belong. In the recycling bin. At least I got my controller designed, built and working.

Something is seriously wrong if you're getting such widely varying forward voltages! These voltages should be far more consistent. The best way to do that kind of cherry picking is to run them with a constant current source (set to the nominal operating current) and then measure the forward voltage. That's the only way to insure that the conditions are exactly the same for each LED you measure (assuming the ambient temperature doesn't suddenly radically change ;)

But, I suspect that if you get a hold of some quality LEDs they will match right out of the box. Where are you getting your LEDs?

I got them off of eBay from a US seller. My concern is that they all come from only a few sources. So I am going to sample some from different manufacturers and go from there. They work enough to get the car legal to drive but they are not acceptable by any means. They were cheap. about .06 each because I bought 1000. Once bitten twice shy. My concern is that I am going to get the same thing from another source and waste more money. So I am going to get small quantities and and go with the ones that match the best out of the box. It is possible that i got some that were not correctly labeled on the current rating but I can grab 5 and run them with a 150 ohm resistor and get basically the same results.

Many very cheap components now originate from China. I bought a bag of 100 white LEDS, and a number of them turned out to have the relative leg lengths wrong, anode and cathode swapped over.

With the automatic machinery that pumps these out by the billion, how could that happen ?
I wonder if they are deliberately trying to put the rest of the world out of business.

I think the upshot of both the original posters experiences & some knowledgeable comment here is:
EBay/Chinese no-name brands (typically low cost) are "suspect".
Probably armed with specific manufacturer information & quality access to their tech support it's possible this might be resolved.
For the amateur at home typically working on one project at a time it's going to be necessary to test every LED.
Generally we want adequate brightness, within tolerances dictated by perception, at the rated forward current.
The working lifetime of bright & super-bright LEDs is often claimed >10,000 hours, but I have also seen >5,000 hours.
This looks good, but almost everyone experiences premature failures, which may or may not matter much in the application, nevertheless need to be aware of this.
One supply alternative is to work with a US based supplier who has the necessary contacts & experience to offer only more consistent product. Ultimately this should overall result in a lower cost outcome, with lower levels of frustration.

I've been buying LEDs from SuperBrightLEDs.com for close to a decade and have always received consistent, quality product. You'll pay a bit more, but probably worth it in the long run.

I Received the 200 I ordered today. I Build up a panel and it was spot on. And the best part is it only cost $4.00 plus less than $2.00 for shipping. The LED's are made by a company called Ligitek.

I am very happy with the company I ordered them from. They have good product for a great price. And the shipping is super cheap. It has never taken more than 7 days to get my order. For my 142x74mm Prototyping Boards I paid $1.15 each.

The only pitfall in my book is that they are not a US Based company but they do stand behind there product. And they do have US Based distributors. My order came from Colorado. The Company is based in Bangkok Thailand.

I did not see a lifetime ratting on them but If they last 5,000 hours that would be great.

Here are the specs on the LED's

Emitting color: Red
Diameter: 5mm
Lens color: Water Clear
Forward voltage(V): 2.0-2.2
Current(mA): 20
View angle: 25
Luminous intensity(MCD): 8,000 - 12,000
Manufacturer Ligitek
MPN LED 5mm Red Water Clear
SKU A-706

Oh I forgot to mention. The other LED's are going into the scrap bin boards and all. Cheaper and Easier to rebuild than try to salvage any of the others.

Well pleasing to see it was, as suspected, an LED quality issue.
I would be very interested to hear when your project is completed if ANY of this new batch have proven "out of spec".
Also since plainly you have discovered a good value for money supplier, can you please advise us all of their contact/order details & name.

Thanks & good luck

"(Tayda Electronics)":http://www.taydaelectronics.com/

If you watch their facebook page they have discounts a couple times a month also.

I will post up a picture of it when I am done. Should be a few days.

Well, what a surprise, ..................
Tayda Electronics has been my supplier of first choice for two years now!!
Their product range is not extensive, but the service is great, website easy to use, shipping costs minimal, prices are low & quality is good.
To be honest I can't really say overall that their product quality is better than any other supplier, but I have been always pleased with what they supply, & the above thread however provides a strong quality endorsement.
Most importantly they seem happy to ship absurdly low quantities although I go to some trouble not to take unfair advantage of this .
I have no connection with Tayda (other than as a satisfied customer) but I heartily recommend them to anyone.
Website: http://www.taydaelectronics.com/