Inside low-cost LED ac bulbs

[Dave_H]

YOU are guessing, I am not .....

I believe I corrected that, at least from viewpoint of Canadian requirements.

Small SMT resistors in volume are 1/10's of a cent each. Small ceramic (MLCC) capacitors can also be fractions of a cent each. During a 2nd-source LED exercise, small wLeds were priced, but nothing near 100 lumens. They were in the order of a few cents each, but not in huge quantities. Some of the LEDs in ac bulbs are 6-9 per package, looks like multi-chip and three LEDs per chip; I imagine these are a bit more expensive.

There are a bunch of other components I could go through, but not much point. I would be guessing at some of it. It is the overall costed BOM which matters. If you have any shareable cost information, perhaps you could do that. I think Ikea would not give us any of this.

Interesting side issue, in 2018 ISED tested a number of triac light dimmers, which all failed the ICES-005 requirements. Not that it affects most of the bulbs being discussed, which are not dimmible. So it's either the bulb or dimmer, or both.


Dave

 

[JustAnOldFashionedLEDGuy]

In volume 65lumen LEDs, $0.006 USD, crazy when you think about it.

w.r.t. Triac dimmers, did they fail with Incan's or LEDs? I am not that surprised they did, and expect they have always. I don't have huge amounts of triacs in my house any more though. Most of the LED bulbs suck at deep dimming. The 3-4" downlights are much better as are the controllable bulbs, but not all.

 

[jtr1962]

In volume 65lumen LEDs, $0.006 USD, crazy when you think about it.

I remember the days when 30 lumen Luxeons ran about $25. So prices per lumen dropped by 4 orders of magnitude since then. Amazing. And the consistency/efficiency of the LEDs is far better besides. 65 lumens back then-2+ watts of power, 2 watts of waste heat, more or less. 65 lumens now-1/3 watt of power, maybe 1/8th of a watt of waste heat.

 

[raggie33]

In my last home I had a few that when turned on I'd loose a few tv channels .

 

[JustAnOldFashionedLEDGuy]

I believe I corrected that, at least from viewpoint of Canadian requirements.

You believe you are right. I know you are not.

 

[Dave_H]

The usual going OTC retail price of non-prime brand LED bulbs here is around $2-$3 each in low quantity (1-4). Philips and GE etc. are substantially more, would be interesting to see how much obvious difference there is on the inside.

HD locally now has their 60W eq. EcoSmart bulbs at 96 cents again, "Special Buy" but normally in the same range as others.

A technical question, as linear regulation at 110vac requires at least 45 or so LEDs stacked in series, how is this done (or is it) for 220vac bulbs? With LED packages having up to 9 LEDs each, it would take ten of these per bulb, which is doable (I think).

Dave

 

[Dr. Jones]

I see some caps there that look like they will rupture after some time... killing the bulb.

True. Electrolytics don't last forever.

 

[Dave_H]

True. Electrolytics don't last forever.

May be true, but my question is, is this the primary failure in this kind of bulb? Is there meaningful data on this? No doubt there is some variation in quality, some I've seen are rated at 105C temperature, the Ikea's are 130C, but there's more to it.

Dave

 

[turbodog]

May be true, but my question is, is this the primary failure in this kind of bulb? Is there meaningful data on this? No doubt there is some variation in quality, some I've seen are rated at 105C temperature, the Ikea's are 130C, but there's more to it.

Dave

I hear you, but if I had a dollar for every dried out, swollen, and failed cap I've seen (compared to dead resistors/etc). I'd have a PILE of money.

Just replaced (3) 25v 2200uF in a power supply last week.

 

[Dr. Jones]

May be true, but my question is, is this the primary failure in this kind of bulb? Is there meaningful data on this? No doubt there is some variation in quality, some I've seen are rated at 105C temperature, the Ikea's are 130C, but there's more to it.

Dave

I think that it would depend upon usage; electrolytics are much more prone to failure from non-use, e.g., drying out, than from continuous use, all else being equal.

LED bulbs that are used every night will likely have other component failures prior to the capacitors going away; and of course the initial quality of the caps is a factor. Thankfully we're pretty much past the great Chinese capacitor plague.

 

[Dave_H]

Not exactly a cheap A19 bulb, here is inside a Maxlight PAR38 LED spotlight, 1250 lumens/14W with dusk/dawn on-off feature (no adjustment of sensitivity level, or disable).

Not a complete teardown, there is a small PCB in the base, appears to rectify ac line, some filtering (56uF 160v cap.). LED supply is around 135vdc, lower than usual for A19 bulbs, appears to be dropped to reduce load on regulator.

Total 30 LED packages (triples, vf=8.4v) appear wired as two series strings of 15 each. Close spacing of LEDs is not optimal for heat dissipation, more to do with forming the beam pattern.

8-pin SOIC is RM9033GC linear current regulator. Total LED current calculated to be 83mA, based on 0.1v across 1.2 ohms. At 135v that yields 11.2W which is somewhat close to claimed 14W power.

The light detector uses small 6-pin IC, powered by 4.7v dc from PCB in the base.

Dave

 

[Dave_H]

I think that it would depend upon usage; electrolytics are much more prone to failure from non-use, e.g., drying out, than from continuous use, all else being equal.

LED bulbs that are used every night will likely have other component failures prior to the capacitors going away; and of course the initial quality of the caps is a factor. Thankfully we're pretty much past the great Chinese capacitor plague.

Not doubting this, just that I've not seen many cases of visible electrolytic cap failure in consumer devices, test equipment, and telecom designs.

An electronic digital clock which has been operating for decades started acting a bit erratically, gaining time for no obvious reason. Its unregulated dc power supply showed a lot of ripple. Replacing the electrolytic fixed the problem.

Life of components depends on environment and usage; the design, proper operating levels including margin; and quality of the components. For cheap LED bulbs, would be interesting to know which components are expected to fail first or early. I doubt Ikea or anyone else would publish reliability detail, other than expected lifetime however arrived at.

Some Ikea bulbs are rated for the number of on/off cycles, not operating hours. Based on a 1 hour minimum on-time (reminds of fluorescent bulbs) this may correlate.

Dave

 

[PhotonWrangler]

Not exactly a cheap A19 bulb, here is inside a Maxlight PAR38 LED spotlight, 1250 lumens/14W with dusk/dawn on-off feature (no adjustment of sensitivity level, or disable).

Not a complete teardown, there is a small PCB in the base, appears to rectify ac line, some filtering (56uF 160v cap.). LED supply is around 135vdc, lower than usual for A19 bulbs, appears to be dropped to reduce load on regulator.

Total 30 LED packages (triples, vf=8.4v) appear wired as two series strings of 15 each. Close spacing of LEDs is not optimal for heat dissipation, more to do with forming the beam pattern.

8-pin SOIC is RM9033GC linear current regulator. Total LED current calculated to be 83mA, based on 0.1v across 1.2 ohms. At 135v that yields 11.2W which is somewhat close to claimed 14W power.

The light detector uses small 6-pin IC, powered by 4.7v dc from PCB in the base.

Dave

View attachment 51035

Does the light sensor have it's own lens to prevent it from seeing the output of the LEDs and going into oscillation?

 

[Dave_H]

Does the light sensor have it's own lens to prevent it from seeing the output of the LEDs and going into oscillation?

Good question...light sensor is that black thing "S1" at the top right, a 3mm LED package angled somewhat away from the LEDs. It is outside of a centre lens for the LEDs so to that
extent is somewhat isolated. Sensor lens is dark grey tinted such as the case for some IR LEDs.

Maybe that's it...could be detecting IR from sunlight, but none from LEDs?


Dave

 

[PhotonWrangler]

Good question...light sensor is that black thing "S1" at the top right, a 3mm LED package angled somewhat away from the LEDs. It is outside of a centre lens for the LEDs so to that
extent is somewhat isolated. Sensor lens is dark grey tinted such as the case for some IR LEDs.

Maybe that's it...could be detecting IR from sunlight, but none from LEDs?


Dave

It seems plausible. That light sensor is probably selected for NIR wavelengths as opposed to longer "thermal" wavelengths. I'm betting that the LEDs put out a minimal amount of energy at wavelengths longer than 800nm.

What prompted my question is that we have a dusk-to-dawn lamp in an outdoor lantern style enclosure, and right around dusk and dawn it sees just enough of it's own reflection combined with just enough daylight to cause it to oscillate like crazy until the sun either rises or sets.

 

[PhotonWrangler]

It seems plausible. That light sensor is probably selected for NIR wavelengths as opposed to longer "thermal" wavelengths. I'm betting that the LEDs put out a minimal amount of energy at wavelengths longer than 800nm.

What prompted my question is that we have a dusk-to-dawn lamp in an outdoor lantern style enclosure, and right around dusk and dawn it sees just enough of it's own reflection combined with just enough daylight to cause it to oscillate like crazy until the sun either rises or sets. And that causes a ton of RFI while it's oscillating.

 

[jtr1962]

I'm betting that the LEDs put out a minimal amount of energy at wavelengths longer than 800nm.

Unless it's an IR LED, LEDs put out practically no energy at >800nm. Any energy they do put out there comes from heating of the die, not the p-n junction.

 

[kaichu dento]

...we have a dusk-to-dawn lamp in an outdoor lantern style enclosure, and right around dusk and dawn it sees just enough of it's own reflection combined with just enough daylight to cause it to oscillate like crazy until the sun either rises or sets.

Interesting side topic that has had me wondering why the photo sensors that I've installed myself have had this flickering issue, but the ones that are installed inside the bulbs themselves haven't. Though expensive, they seem to be lasting pretty long and proving very reliable.

 

[Dr. Jones]

Not doubting this, just that I've not seen many cases of visible electrolytic cap failure in consumer devices, test equipment, and telecom designs.

An electronic digital clock which has been operating for decades started acting a bit erratically, gaining time for no obvious reason. Its unregulated dc power supply showed a lot of ripple. Replacing the electrolytic fixed the problem.

Life of components depends on environment and usage; the design, proper operating levels including margin; and quality of the components. For cheap LED bulbs, would be interesting to know which components are expected to fail first or early. I doubt Ikea or anyone else would publish reliability detail, other than expected lifetime however arrived at.

Some Ikea bulbs are rated for the number of on/off cycles, not operating hours. Based on a 1 hour minimum on-time (reminds of fluorescent bulbs) this may correlate.

Dave

In my experience repairing primarily consumer-grade electronics in my spare time, electrolytics usually show little or no external signs of failure unless they are input filter capacitors, in which case they can be quite a mess.

 

[Dave_H]

In my experience repairing primarily consumer-grade electronics in my spare time, electrolytics usually show little or no external signs of failure unless they are input filter capacitors, in which case they can be quite a mess.

Makes sense, input capacitors may be exposed to more surges than other places in the circuit. Also, input capacitors on switching supplies may be subject to a fair amount of ripple current.


Dave