are some LEDs expected to burn in???

[darron]

i was making an LED strip to try out on some "plexi" (just acrylic... not plexiglass)...


i got a heap of these LEDs to use for it

http://australia.rs-online.com/web/search/searchBrowseAction.html?method=searchProducts&searchTerm=697-3608&x=0&y=0




i wired on about 6 and hooked it up to a 3.3V (for LEDs) switchmode supply. they came on super bright as i wanted! then about 5 seconds later sort of popped and smelt. before 'minor burning' they also changed to a shade of green for a tiny moment.

i remeasured the supply disconnected and it was 3.4V... the LEDs rated maximum is 3.8, so i was going to run them at a maximum of 3.3.... which is the typical.

i thought it might be the supply so i also hooked up a 10k load and wired on another 5 or so.... they all did the same thing.


BUT... though they are 1/2  -   3/4 the brightness of the first few seconds, they all seem fairly consistent in brightness, which isn't something i would expect from a chain of half-burn LEDs.... the smell worries me.... even though working i don't like knowing they are half cooked...





it would be a shame not to get the maximum brightness out of them. should i use a 3v supply or something instead?


anybody have any thoughts???

[darron]

update:


i left it running for an hour on test. half of them stopped.. one of the mis charred...


where did i go wrong?

[Incubusguy]

When you say '10k load', do you mean a 10k series current-limiting resistor? If not, consider the following:

While they have a maximum forward voltage, remember that after the turn-on voltage of the diode is surpassed, the resistance drops down to ohms; hence the need for a series current-limiting resistor.

You may have been under the maximum forward voltage but you were probably way over the 20 mA maximum forward current.

[Gurner]

Have you double checked your PSU?

In my experience...blue LEDS will normally tolerate 3.8V to 4V forward voltage with no resistor before you hit general smellage (but even so, going over 3.3V is a seriously bad idea). So if you're using a 3.3V supply then you ought to be fine. The bit that raises my suspisicions is this....

"i wired on about 6 and hooked it up to a 3.3V (for LEDs) switchmode supply"

that doesn't read right - do you have some sample address of the internet type, to the PSU you're using, I suspect it's way more than the 3.3V you think it is.

[darron]

When you say '10k load', do you mean a 10k series current-limiting resistor? If not, consider the following:

While they have a maximum forward voltage, remember that after the turn-on voltage of the diode is surpassed, the resistance drops down to ohms; hence the need for a series current-limiting resistor.

You may have been under the maximum forward voltage but you were probably way over the 20 mA maximum forward current.

i tried a 10k in parallel afterwards to see it if it made a difference on the supply, no difference...


so you're saying i need a series limiting resistor? i tried a 220r before and the LEDs almost disappeared, so i'd need something very small and i'd have to re-measure it when i know how many LEDs i need.

the supply i was testing with worked on upwards of 100 LEDs in parallel before for a sign...


i've always been alright supplying an LED with the "typical" rated voltage and drawing the typical sort of current - from the datasheet.


the LEDs were cheap... give them 3V maybe?

Have you double checked your PSU?

In my experience...blue LEDS will normally tolerate 3.8V to 4V forward voltage with no resistor before you hit general smellage (but even so, going over 3.3V is a seriously bad idea). So if you're using a 3.3V supply then you ought to be fine. The bit that raises my suspisicions is this....

"i wired on about 6 and hooked it up to a 3.3V (for LEDs) switchmode supply"

that doesn't read right - do you have some sample address of the internet type, to the PSU you're using, I suspect it's way more than the 3.3V you think it is.

here's the supply i was testing with...

http://australia.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=436-042

what do you mean more than i think though? is there a different way that i should be measuring it?

[Gurner]

That PSU looks fine...have you tested its output with a DVM?

It may well be that it needs a more significant load to pull the regulation in (ie to be assured of solid 3.3V output)...so worth checking.

[darron]

That PSU looks fine...have you tested its output with a DVM?

It may well be that it needs a more significant load to pull the regulation in (ie to be assured of solid 3.3V output)...so worth checking.

mmm.. i would think so too.... glad other people are thinking i'm not crazy.


like i said. i tested it to be reading 3.4v with no load, and i tried a 10k load on it along with the LEDs and they still burnt...

[Gurner]

like i said. i tested it to be reading 3.4v with no load, and i tried a 10k load on it along with the LEDs and they still burnt...

Oops...missed that bit - I still suspect the PSU (possibly overcorrecting from the initial voltage sag when you switch onthe LEDs) ...see what happens when you slap in a low value resistor in series with the led ....and just while you are trying to get to the bottom of this, only connect one led up.

[darron]

i think that's the go and this will be as far as we get... shame though, to need to regulate the regulator down a bit though because i'll have to change the resistor based on the number of LEDs connected probably. i'll make a careful selection with slow testing.


thanks for responding guys...

[linny1982]

dude check the led calculator. i put in all the details and it says your voltage is way to low. best i got was 6 in series with a 68r on 24v. and it only draws 40mA

[darron]

dude check the led calculator. i put in all the details and it says your voltage is way to low. best i got was 6 in series with a 68r on 24v. and it only draws 40mA

the idea was to connect them in parallel so that i don't have to change the voltage for ever led added. i'll make up something new anyway as the switchmode was only testing them.

i don't like to string LEDs in series much because you rely on the forward voltage drop and the final limiting that you need to do is too precise.

[R.G.]

OK, what's wrong here is that you have a fundamental misconception about LEDs.

LEDs are current mode devices. You drive them with a current-limited power supply, and the voltage is whatever the the LED needs. The light output from the LED is approximately proportional to the current through them.

LEDs do have an approximately constant forward voltage. This varies slightly from device to device. And like an ordinary diode, they do not limit their own current. Applying a voltage to a diode - any diode - results in the diode letting relatively unrestricted currents flow. Something else must limit the current, or the diode (including LED) will kill itself, as you've found by experiment.

Paralleling LEDs with no voltage-difference-compensating resistors is another problem. Because the LED forward voltage is a little different from device to device, the low voltages ones eat all the current, die, and then the next-higher voltage one eats all the current, and it dies, then the next, and so on until you have a burned-out, stinky mess which is probably releasing non-RoHS substances into the air.

Never apply a voltage source to an LED. Either no light (because the voltage is too low to make current flow) or LED-death is the most likely result. Apply a current source to an LED to force it to conduct the current you want it to have. Arrange to keep the current at the desired level to get the amount of light you want.

[Gurner]

I've a bench PSU...regulated adjustable output blah blah - with every 'bunch' of leds I receive from the far east...I do a couple of tests to see how far they can be pushed.

Blue leds  ....3.2V -> 3.3V straight out of the bench PSU...no smoke, no heat, no smellage, no "damn there goes 2p I'll never get back".....no drama.  So while I agree with the sentiment, real world tests don't result in instant death (or zero light). I drive such blue leds direct from a PIC output pin with no series resistor (albeit there's be a tad of pin internal resistance within the pic itself)...I feed the PIC with a regulated violtage supply matched to the forward voltage of the LEDs I'm using. I do this not cos I'm a tightwad or want to stray from the norm, I do this becuase often I'm *very* pushed for PCB real estate....and if it works (I check the current draw of the total led circuit...to ensure the current per pin is less than 20mA), then that's good enough for me!

[darron]

i suppose that i've been lucky with most of my selections before. for example getting LEDs where fed off this supply (3v3) only just start to conduct and "eat" a few mV.

R.G., thanks for your input to make them see more of a current source than voltage. do you think you'd be better off starting with a voltage closer to say 5V and limiting down with resistors?

stacking a heap in parallel does make me think about the differences that i have seen, and also your description from where i have intentionally blown up bad LEDs with higher current

[darron]

shame the modern cheapie meters that i use don't have a ammeter for DC..... should invest one day....

[Gurner]

shame the modern cheapie meters that i use don't have a ammeter for DC..... should invest one day....

Well @3.3v out of your PSU, if you use a series resistor (like a 10R, 47R ....something low), then measure the voltage drop across the resistor, you can establish the current through the led.

[amptramp]

If your blue LED's turned green just before self-destructing, this is a sign of extremely high junction temperatures.  We used to use a value of 3 nanometres longer wavelength for every 10°C temperature rise back when I was doing dot-matrix displays for the F-16 in the 1980 to 1983 time frame.  These were 575 nanometre green LED's and if the aircraft had been sitting out in the sun for any length of time, they would look yellow.  R. G. is right - you have to drive LED's with something approaching a constant current or they are toast.

[R.G.]

Blue leds  ....3.2V -> 3.3V straight out of the bench PSU...no smoke, no heat, no smellage, no "damn there goes 2p I'll never get back".....no drama.  So while I agree with the sentiment, real world tests don't result in instant death (or zero light).

It's true that instant death is not necessarily what happens every time. Witness that Darron loses them slowly. I'd put that as "real world tests don't *always* result in instant death". It is one possibility, though, and the overheating is not good for any of them - they just recover on cooling if they aren't pushed too hard. LEDs also vary in how much abuse they can recover from. I sometimes hate myself for quoting popular songs as wisdom, but a Bob Dylan tune comes to mind: "How much abuse will you be able to take? There's no way to tell by the first kiss."

I drive such blue leds direct from a PIC output pin with no series resistor (albeit there's be a tad of pin internal resistance within the pic itself)

PIC pins are designed for driving LEDs directly from the pin to save putting resistors in, at least on recent PICs. They are designed to limit to nominally safe LED currents. Check the datasheet.

R.G., thanks for your input to make them see more of a current source than voltage. do you think you'd be better off starting with a voltage closer to say 5V and limiting down with resistors?

It depends on your design. The most common thing to do is to use a higher voltage and resistors, all right. However, this works better when the voltage is at least 100% higher than the LED voltage, and best when it's several times the LED voltage. This is because of the differences in the LED voltage and the power supply voltage. If you want 20ma per LED, and have a 5.000000V power supply, then the resistor value to keep 20ma is R = (5.0-Vled)/0.02.

Notice that 5.0-Vled varies a lot with the LED voltage. The bigger that difference is, the less the resistor varies.

Put another way , Iled = (5.0-Vled)/R . If R is fixed, then the current varies directly with the difference in 5V and Vled. Even a few tenths of a volt changes the current, not to mention then the variation because of the 5V supply not being exactly 5.0000V and the resistor having a 10% tolerance. So making the LEDs be the same brightness (which is where you'll get to eventually, once you see different brightnesses in the LEDs) starts getting tricky, even not including the variation from LED to LED in light-per-miilliampere. You can do it with 5V and a resistor. It works best with 1.5V LEDs and less well as the Vled approaches the 5V supply. That's why series connections are often used. At least then you can guarantee that every LED has exactly the same current in it. After that, all you need to do is match LEDs for light output at a current. And you only need one current limiter, which may then be made an active circuit and waste a lot less power.

By the way, if you're doing this for backlighting, there is a variety of acrylic sheet ("Plexiglass" is just a brand name for acrylic sheet, at least in the USA) which has internal light diffusing beads so that edge lighting the sheet makes it glow all over the flat faces. I always thought I'd like to use a piece of that as the "indicator LED" on a pedal, so that the whole face of the pedal lights up when you stomp the switch.

Given what I know of the internet, there will be pedals like that on the market within the week, I guess.

[Gurner]

Just for fun, I've just tested a blue led to destruction....voltage straight out of my bench PSU into the led with no series resistance.

2.8V....it starts to light up....brightness increases in line with voltasge up to about 3.2V....brightness then increases a little more upto 3.8V....not much increase in brightness between 3.8V & 4.8V, it starts to fade at about 5V...and dies once I cranked the voltage across it to 5.5V

I made a video nasty of it...my son's at my main PC with his schoolmate (driving a train ...& trains trump 'LEDs being nuked' type videos apparently)....and that's the PC which has an SD card reader in....if I get a mo later I'll upload it.

[bluesman1218]

There is a calculator, on the web somewhere, that will tell you what value CLR to use based on your LED's specs, power source and brightness desired. I found it on another thread here, but am not on my computer with the bookmark. Try this link http://led.linear1.org/1led.wiz