Weird LED Behavior... any explanations?

Thread Starter

anparks

Joined Jun 29, 2009
9
I've encountered a bizarre LED behavior and am wondering whether anyone can shed any light on the reasons behind it.

I Recently built a simple miniature white-LED flashlight onto the top of a 9v battery clip, and took it out on a backpacking trip as a super-small flashlight.

After some time I realized that the light was beginning to strobe- first very quickly, and increasingly slower until it hit about 10 pulses per second. I took a closer look and realized that I'd put a much smaller resistor in series with the 2 white LEDs than I'd intended, and obviously a lot of current was flowing (up to 100ma, measured).

The strobing effect is what I'm interested in- how could an LED turn into an oscillator? Is this a failure mode of LEDs?

I'm an EE major and longtime hobbyist and have never seen anything strobe like this without good reason.
 
Last edited:

Wendy

Joined Mar 24, 2008
23,415
I'm thinking the resistor heating and cooling was causing it to change value, possibly in a negitive resistance fashion. Just a theory.
 

hgmjr

Joined Jan 28, 2005
9,027
I suspect that what you were seeing was the death-rattle from the battery.

As the battery dropped below the level adequate to power the LED it would be unloaded and the battery would then recovery from the overload. This would explain the first rapid flashing and then slowing flash with time.

hgmjr
 

Thread Starter

anparks

Joined Jun 29, 2009
9
Thanks for the replies.
Good theory hgmjr, but I just tested it with both a fresh 9v and a 9v power supply, and am still getting the same strobing - still about 10 pulses per second.

The circuit is super simple- just two 3.4v bright white leds (recommended I=20ma) in series with a (much too small) 10 Ohm resistor, connected across a 9v battery or power supply.
 

Wendy

Joined Mar 24, 2008
23,415
It could also be heating causing the diode to temporarily fail, until the final time it dies completely. How much current do you think is going through this sucker (no schematic).
 

Thread Starter

anparks

Joined Jun 29, 2009
9
Could be. It doesn't seem too hot, although the resistor gets warm to the touch. I measured about 100ma at first with a fresh battery, but measuring again gives much less- about 20ma (probably due to the strobing... not sure about the duty cycle, etc.).

Here's a schematic, really straightforward:
http://www.mediafire.com/imageview.php?quickkey=j1nwjxbmgtg&thumb=4

Here's a photo of it (pretty crappy, I know.):
http://www.mediafire.com/imageview.php?quickkey=xem1mj3fymn&thumb=4

And a short vid of the strobing (fairly large b/c it's straight off my camera):
http://www.mediafire.com/?1mnqymmtwzj

It's not a big deal, or a problem I need solved, or anything. It's just weird to see something so simple oscillating on its own.
 

hgmjr

Joined Jan 28, 2005
9,027
I wonder if you have purchased flashing LEDs. There is such a thing. Do you know the part number of the LEDs that you are using?

hgmjr
 

Wendy

Joined Mar 24, 2008
23,415
Let's see, 2.2V differential (9V-6.8V) divided by 10Ω, this would be 220ma. I suspect those LEDs are screaming their death throes. Even at 100ma their dying.
 

Thread Starter

anparks

Joined Jun 29, 2009
9
Hey- I didn't think of that.
But no, I've used others from the same lot in different applications and not had this happen.

The LEDs were straight from Hong Kong (via eBay), and came in a pack of 200 with no part number, datasheet, or specs except for If=20ma, Vf=3.2v.
eBay product page:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=110376183431&ssPageName=ADME:X:AAQ:US:1123

Unless a flashing LED got mixed in with the others at the factory (?) and I just happened to install that. Seems far-fetched, though.

Maybe I'll try to replicate this and see if it happens again. Then patent it as the busted-LED oscillator.
 

Thread Starter

anparks

Joined Jun 29, 2009
9
Thanks for the input- that's a good point.

Well, the 100ma was read off of a true-RMS DMM while the LED was strobing. It's averaging the current throughout the pulse period, and that average is 100ma. So during the time the LED is on it could be well into the 200s. Yes, they are screaming. But not dead yet.


But the point is: Is this a failure mode of LEDs? To strobe like that? If anyone has seen anything like this before, I'm interested.
 

someonesdad

Joined Jul 7, 2009
1,583
This is an interesting phenomenon -- I too would like to know more. It would be interesting to see if you could duplicate the behavior with a power supply or see if the battery is essential. Hook things up to a scope and look at the current through the LEDs. While you've got the scope hooked up, look at the light output of the LEDs -- you can use my poor man's integrating spectroradiometer with no licensing fee -- just solder a photodiode to a BNC connector (works great with DMMs and scopes for seat of the pants light measurements).
 

Thread Starter

anparks

Joined Jun 29, 2009
9
Hi. I appreciate your interest! This may be a waste of time, but I'm still curious and would love to find out what's really going on.

I graphed the voltage across the 10 Ohm resistor with my (astounding) 1ch portable scope, and attached some "screenshots." I also made my own version of your spectroradiometer with a little solar panel from an old broken calculator, and hooked it up to my 'scope. Check out the pic.

The peak voltage across the resistor was 0.85, giving 85 ma peak current through the LEDs when they are on. They stay on about 10ms and then are off for between 70 and 90 ms. It's kind of irregular sometimes (see "erratic" photo), but 90% of the time it stays very regular.

Yes, a power supply set to between 6 and 12v yields the same strobing as the 9v battery. lower voltage causes quicker strobing, and higher voltage slows the strobing. Below about 6 it stops oscillating and just glows faintly (scope confirms). I'm afraid to go above 12 b/c I don't want to destroy this thing. May never happen again.

Aaron
 

Attachments

THE_RB

Joined Feb 11, 2008
5,438
...
and higher voltage slows the strobing. Below about 6 it stops oscillating and just glows faintly (scope confirms).
...
Definitely sounds like a heat failure. When the voltage is higher the chip is left at a higher temperature after it fails and the current stops. Hence it takes longer to cool and the freq is slower.

I wonder if the design of the LED chip has some type of inbuilt overheat protection, that cuts the power... I would try and replicate it with other LEDs from the same batch.

A "hiccupping" type of repeating heat failure is very common in component failures but I've never seen it in a LED!
 

someonesdad

Joined Jul 7, 2009
1,583
Definitely sounds like a heat failure. When the voltage is higher the chip is left at a higher temperature after it fails and the current stops. Hence it takes longer to cool and the freq is slower.

I wonder if the design of the LED chip has some type of inbuilt overheat protection, that cuts the power... I would try and replicate it with other LEDs from the same batch.
This is exactly the same thing that occurred to me as I looked at the scope photos. If so, it would be interesting to know how the vendor did it. Some opening of the case and microscopic examination would be fun (alas, I don't work at a facility with such tools anymore; otherwise, I'd be glad to help).
 

Thread Starter

anparks

Joined Jun 29, 2009
9
Yes, I agree that higher input current resulting in lower frequency is evidence for some sort of failure prevention mechanism. Kind of like the old incandescent flashers that used bimetallic strips to open the filament as it heated, then close again as it cools to produce a strobing effect. Could very well be what's happening here, on a much smaller scale.

I'll try to duplicate the effect with others in the batch and see what happens.

Thanks,

Aaron
 

wood78987

Joined Jun 22, 2009
1
This is not a failsafe mechanism, its the actual PN junction of the LED breaking down. With 100mA flowing through an LED rated for only 20mA this is inevitable.
You probably noticed the first time you ran the LEDs they lasted for much longer before starting to flash than they do now. The PN junction is becoming more and more damaged until it will eventually fail. Change your LEDs and your series resistor to at least 100 ohms.
 

Audioguru

Joined Dec 20, 2007
11,248
The chewing gum that holds the LED die on the leads of the cheap LED causes poor contact when the LED gets warm.

Have you ever seen the inside of a counterfeit power transistor? Lots of chewing gum.
 
Last edited by a moderator:
Top