Intermittent LEDs

[Diver300]

I recently bought a few of these:- **broken link removed**

The power-on indicator consists of four white 5 mm LEDs around the on-off switch. They are in two strings, each with a resistor and a diode in series and run from 230 V 50 Hz mains.

I find lighting that flashes at 100 Hz bad enough, but these run at 50 Hz. Also, I can't see why the LEDs are in two strings, as the LED voltage is around 12 V, so far below the mains voltage. There is only about 0.5 mA through the LEDs.

Anyhow I've changed the circuit to put all four LEDs in series, and removed one of the diodes and resistors, and added an electrolytic to get rid of the flicker, and they seemed to work.

After a month or two, some of the LEDs have gone out, and some flicker or are intermittent. The flicker isn't the 50 Hz modulation, it's more like a loose connection or the flicker that an unstable neon light has, so the light turns on and off sort of randomly, once or twice a second.

I've swapped out some of the LEDs, and tested the ones that I've taken out. I've found that it takes quite a lot more than 0.5 mA to light some of the LEDs, and others seem to "loose" around 0.5 mA intermittently, so that at 0.5 mA, they turn on and off. At larger currents the brightness varies quite a lot. Once the current is up to 10 mA or so, there's no visible variation.

I'd not seen before LEDs that took that much current to light, nor LEDs where the brightness varied with a constant current.

Are these particularly poor LEDs? How common is it to have variations in brightness, both between LEDs and with time? Is it this behaviour what has made designers vary brightness with PWM instead of reducing current?

Finally, has anyone any idea why the LEDs weren't originally in one string of 4 LEDs?

 

[Les Jones]

Can you post the schematic of how it was originally connected and how you have it connected now ?

Les.

 

[MikeMl]

I have had high-power LEDs get permanently damaged if overheated by not having good thermal contact with their aluminum heatsink. Their failure mode is greatly reduced light output, and then they flicker like an old dying florescent tube...

 

[Diver300]

Here is the circuit. The average current in the LEDs is similar in each circuit. It is only around 0.5 mA, so I don't think that the LEDs got too hot, but they were probably turned on for a month or more before they started misbehaving.

I measured a couple of the LEDs that I had taken out. I found that at around 0.5 mA, the forward voltage of one of the LEDs would randomly change by around 1 V as the LED turned off. It is as though there is a resistor in parallel with the LED that is turning on or off.

 

[audioguru]

The capacitor you added charges to the peak voltage of the AC which is 1.414 times higher than the rectified RMS voltage before so the current in the LEDs is also much higher. The current is probably 30mA to 50mA, not very dim 0.5mA.
Cheap LEDs are cheaply made with bits of string and chewing gum to hold the connections together. Not reliable.

 

[Les Jones]

Assuming that the resistors are the same value as the originals and that the input to both circuits is straight from the mains. (I.E. no other components in series with the mains input to the circuits.) then I can't see why the modification should cause the LEDs to fail.
Edit.
I have just read AG's comment about the capacitor charging to the peak voltage of the mains. I originally thought this could be the case but then I assumed that the resistor between the diode and the capacitor was quite a high value (Assuming it was one of the original resistors.) so the capacitor would not charge to the peak voltage of the mains. It would be interesting to know the value of the resistors.

Les.

 

[Diver300]

The resistor connected to the rectifier diode is the original one. The one between the LEDs and the capacitor is much smaller. The capacitor is only about 35 V and runs at about 20 V. The current taken is basically controlled by the original resistor and it is only about 0.5 mA average, and there were no other components, except the switch between what I have shown and the mains.

My modification has reduced the average current slightly, as there is a little more back voltage from more LEDs and the additional resistor. The time constant on the capacitor is such that there is hardly any ripple current in the LEDs. As the old circuit was only half-wave rectified and not smoothed, the peak current would have been around 3 times as big, but still very small for the size of LEDs.

If this type of LED is prone to leakage, there would be less variation in light had the current still been simply half-wave rectified. I am wondering if the lack of smoothing was deliberate in order to increase the peak current and reduce the effect of leakage.

I still have no idea why there were two strings not one.

 

[Les Jones]

It would be nice if you could tell us the values of the resistors.

Les.

 

[Diver300]

The original resistor, the one that is now connected in series with the rectifier, is 200 kΩ. The additional one between the capacitor and the LEDs is 4.7 kΩ. The capacitor is 10 μF, and it runs at about 12.5 V.

 

[alec_t]

I wonder if the 200k resistor was inadequately rated for voltage and was breaking down? That would pass excessive current through the LEDs and shorten their lives. Having two parallel strings of LEDs originally makes no sense to me, since that doubles the required resistor current.

 

[Diver300]

The 200 kΩ resistor is quite large. It is an axial leaded component, 8 - 10 mm long, so I doubt it has any problem with 240 V ac.

I agree that two strings of LEDs makes no sense. The only possible explanation is reliability, but mains indicators are hardly the sort of things to need redundancy.