Puzzling LED behaviour

[Jeff Layman]

We have an en suite which uses three mains GU10 fittings (probably at least a dozen years old). One over the shower is IP65, and is switched separately from the other two, which are ordinary fittings. About 18 months ago I replaced the 50W halogens with 4W "warm white" leds (3000K

- 345 lumens). Each bulb had four leds.

From the start I was puzzled by the way they came on. The single shower led came on immediately I switched it on. The other two were not immediate - they took a fraction of a second - maybe a quarter or so - to come on. When switched off, they all went off immediately.

Recently, the two non-shower leds started to flicker. This didn't happen every day, and was unpredictable. Sometimes they'd flicker only once or twice, then worked perfectly. But a couple of days ago one of the bulbs dropped to a low output - perhaps less than a fifth of normal after a few seconds, and sometimes stayed like that or went back to full output.

When I switched the two non-shower leds off while one was in the low-output state, I was amazed. The normally-working led, as previously, went out immediately. The "low-output" led remained on for about half a second without any apparent change of output! Then it went out completely. I have been able to reproduce this effect, but how can it occur? Surely it would need something like a gold-cap capacitor to retain enough energy to keep a led like that lit, and I doubt there is one of those in the base of the bulb.

[The Natural Philosopher]

Ive got 14 GU10s and two 12V 50W spots on three dimmable circuits in my kitchen. All the LED GU10s are allegedly 'dimmable'. The halogens are driven by a dimmable electronic transformer. They work perfectly.

All the LED bulbs interact with each other to produce flicker.

Turning down one bank stops flicker in another. For a time. Once warm I need to slightly adjust the brightness again to stop flicker

The girl in Homebase displayed the stunning perceptiveness that marks an intelligent person who has no technical knowledge

'LEDS are a work in progress, arent they'

I think that is essentially the problem. There are half a dozens LED driver circuits and chipsets out there, and some of them are utter shit.

None are that good.

The only solution I have found is to try various brands until I find a set that work, and then buy a lot more.

I hant gote time to put RF chokes or capacitors in the wiring or whatever the issue is.

[Brian Gaff]

Yes I was just thinking the same, but the thing is one might hope that a design might be tested with other leds and different types as any circuit so unstable that it is affected by other lighting on the circuit sounds to me , not to be very well designed and possibly also means that at least one of the designs also has parasitic oscillations present causing radio interference far and wide, as if we don't have enough trouble from switch mode power supplies internet cabling over the mains an the like. Brian

[Jeff Layman]

Have you ever had any mains leds which look like they'll make their predicted 25000 hour life? I know a lot of people don't like them, and their light output falls over time, but I've had many CFLs last for years and years. They seem a lot more reliable than leds to me, but maybe it's because they use essentially 100+ years old technology to work, and they are no longer "a work in progress"!

[tabbypurr]

inbuilt capacitance keeps them lit when current draw is low. Flickering & similar erratic behaviour is usually due to a dying LED.

NT

[jrwalliker]

I have seen this effect. In one room there is a mixture of dimable and non-dimable mains LEDs from the same manufacturer. All other specifications are the same and the mixture happened because the shop didn't have enough of the non-dimable LEDs. The non-dimable LEDs light up immediately, but the dimable ones have a slight delay, just as you describe.

John

[Terry Casey]

I had a problem when I replaced some incadescant candle lamps in the wall lights with dimmable LEDs from Toolstation.

However, when set to the level I like when using the PC, there was a piercing high frequency whistle, so the original bulbs went back in.

Adam recommended changing the dimmer for a Varilight dimmer expressly designed for use with LEDs.

Since then no problems at all irrespective of light level.

[Jeff Layman]

Well, my energy calculations are pretty rusty, but let's assume the leds are using only 200mW for the half a second that they remain on. That's

0.1 Ws, or 0.1 joule. To get that sort of energy from a capacitor, you'd need a 10000uF capacitor charged to around 5 volts. There's no way that would fit in a GU10 base. Or have I got all that completely wrong?

Flickering & similar erratic behaviour is usually due to a dying LED.

No doubt. I've got a spare or two somewhere here.

[The Natural Philosopher]

Quite a lot. Its likley the capacitor is a mains one

at 400V, C= 0.1/2 x 400^2 = 0.3uF

Very doable

Not in my experience.

[tabbypurr]

NT:

I don't know where he gets 200mW from. It's obvious it's the capacitor that powers them for the half second or maybe less. If you know the capacitance & cutoff voltage you can approximately calculate the power used.

NT

[Jeff Layman]

I was being ultra-conservative in my estimation of the power being used by the "low-output" led. My OP referred to "perhaps less than a fifth of normal", which would have made it 800mW. In the end I went for a quarter of that for the calculation in case it made it more feasible.

There is an interesting table here

showing the physical size of a capacitor - whether electrolytic or metallised PP - which would be needed for storing some low hundreds of mW.s of energy. Unless there are some SMD capacitors now available which have that sort of storage, I still can't see how a those leds could remain lit for that time without any other source of power. When I've got the replacements I'll try to take the GU10 base off with a Dremel to see what's inside.

[The Natural Philosopher]

I fitted as fluorescent type low energy bulb to a lampholder, Itfalshed and I nearly dropped it. But when I fitted it properly, it was off?

I switched it on, and it worked.

Somwehere there was enough capacitance in the mains cable to cause it to switch on. It hadnt been used in years,

[tabbypurr]

A few uF at mains voltage. You took a wild guess on LED power - guess where your error is :)

NT

[Dave W]

It sounds to me like the non-shower leds are powered by a high voltage capacitor charged by the mains, driving a converter circuit starting up in that qusrter second after swirch-on. The normally-working one would quickly discharge the cap, but the dim one might be taking such a small current that the cap would supply enough for the half-second after switch off.

There are quite a few LED bulb circuits shown in Google images, not using converters. Many of them involve a capacitive dropper driving a bridge rectifier that charges an electrolytic across the string of LEDs, probably big enough to produce a dim output for half a second.

By my calculations your 4W bulbs would need a 0.2uF capacitor as the dropper, but the electrolytic could be any size.

[Terry Casey]

One of mine flickered at some levels but since I changed the dimmer, it's perfect - in fact, I can't even remember which one it was now!

[Jeff Layman]

Before I destroyed the bulb, I thought I could measure the current it was taking in normal and low-output mode. But how do I do this if it uses a capacitor as a dropper? I tried putting a digital multimeter in AC current mode in series with the bulb, but although the meter registered something, the bulb did not come on at all. If I use something like a 10 ohm resistor in series with the bulb, and measure the AC voltage drop across it, would that give me a valid way of calculating the current? Or does the series capacitor screw that up?

[tabbypurr]

You'd get a rough idea. You'd be able to calculate exactly if the lamp had a pf of 1 but it doesn't.

NT