Does anyone know why it is that when White Power L.E.D.S fail they tend to give a flashing effect?
This appears to be a regular flashing unlike the random flashing that you used to get with a worn out Fluorescent Tube.

This phenomenon only occurs with LEDs that operates from a switching power supply. It occurs when the driver can't light the LEDs.

This phenomenon only occurs with LEDs that operates from a switching power supply. It occurs when the driver can't light the LEDs.

I could imagine that the switching transistor is not able to give a continuous oscillation and instead it charges and discharges the transformer core in a low frequency.

In all the repair videos I've seen of a 'pulsing' of a SMPSU, the culprit is a something in the bootstrap circuit for the main switching IC. A common example follows:

ICs don't work at 120/230V - not directly, anyway. The supply voltage needs to be 10s of volts at most. When voltage is first applied to a SMPSU, this supply rail is trickle-charged by a few high value resistors. This is enough to wake it up and begin switching. After the switching has begun, an auxiliary winding is rectified and fed into the supply rail instead. If something in this supply rail fails (A zener for clamping, a diode for rectification, or a cap for smoothing most commonly), the aux rail cannot feed the supply rail for the main switching IC. Only the trickle-charge bootstrap resistors can.

The flashing is because the resistors are charging the IC until it begins switching, then the switching activity pulls the rail down again until the resistors charge it again.

Thanks for that folks!
I was thinking more of the regular flashing effect that you get when White Power L.E.D.S fail in simple Torches etc where the L.E.D.s are running on D.C. via a series Resistor and it is definitely something to do with the L.E.D. itself and complicated Switch-Mode Power Supplies and Driver Circuits are not used though.

It depends how the flashing pattern looks like.
If it is brief full power flashes with longer dark pause, with regular repetition, it could the auxiliary supply described above.
But if it is irregular flashing with duty ratio around 1:1, itcould be some bond wire connecting the LED die to something had broken off and has an intermittent contact, the whole thing cycling thermally. If some LEDs in the assembly are either dead or lighting normally or cycling between higher and lower brightness while others are cycling between dark and alight and mainly if it needs some few secondsbefore it start acting, it is most likely the broken bondwire fault.
If the leds are just flashing dimly, the chain has high resistivediscontinuity fault somewhere, so exhibits too high voltage drop, triggering the load overvoltage protection in the control ic.
Or the auxiliary capacitor may have failed and it is not able to smoothen out the supply for the control part.
Or with some designs the main filtering capacitor has dried out, unable to cover the mains voltage zero crossings, causing the ic loosing supply and to reset (if the auxiliary capacitor is of low value on purpose to reduce startup delay or dried out as well).

Statistic wise most frequent are bonding faults (the bond wire detaches from the die) and electrolytes drying out.
The first depends on led design for reliability (there is known cure for high temperature bonding reliability: A Ni-Pd-Au coating of the bond pads, separating the bond wire material from the Al pad, so they can not diffuse into each other and weaken the bond), but that increases the led cost a bit (even when very tiny amount of gold is needed, it is expensive material after all).

Dried out electrolytics are the same thing like with CFLs. There are LEDs not using them in order to eliminate this fault, but then they flicker on the double of the mains frequency.

I’ve seen a LED street light flickering at an unpredictable rate, hasn’t gone EOL yet. Every time I drive by it always has a different flashing pattern.

I’ve seen a LED street light flickering at an unpredictable rate, hasn’t gone EOL yet. Every time I drive by it always has a different flashing pattern.

There seems to be an industrial trend making the high power LED modules as the aluminum core PCB carrying not only the LEDs, but the last stage of the ballast, all that supplied by 24, 36 or 48V DC voltage source supply. Then the lantern has somewhere tucked a brick converting the AC mains towards this DC supply. Important is, these final ballast stages do feature a thermal shut down feature. Because on a common aluminum core PCB, they share the same temperature. So when that PCB isnot adequatelly cooled (clogged heat sink air flow, not properly attached to the heat sink,...), the thermal shut down feature responds and shuts the section down. It thencools down, the TSD restarts it, so it starts cycling. Because the sensor is part of the converter IC and that IC does generate someheat on its own and has certain thermal resistance to the heat sink but small mass, often this cycling is in the rate of a second or so, so may become a flasher. Now when multiple converter chips are on a single LED board, the thermal cross coupling and slight differences among the channels may together form rather wild flashing effects.

My thanks to all who responded.
The Bond Wire coming adrift suggestion seems the most plausible.
It seems odd to think that it is thermally cycling though when there is so little electrical power in the circuit.

I love this, it's like something out of a dream.

Well that wasn't quite my intention.
I did consider asking for opinions on Street Lamp Interference which I have seen occur.
This is currently out of reach of human understanding and so paranormal.
I wonder what you would have thought about that?

I thought I sent that reply to a completely different thread (the one about a working streetlight in the middle of the woods).