Philips recalls potentially dangerous GU10 LED lamps

[Andrew Gabriel]

In case any of you have used these GU10 8W MasterLED retrofit spotlamps...

[Graham.]

"new " Google Groups interface, and very annoying.

[Lee]

Noticed a few LED spots on the RAPEX website over the last couple of months - inherent issue with this type of design?

[Andrew Gabriel]

>>

Yes.

These are made by companies which previously made CFLs. CFLs have no exposed metalic parts, except the lamp cap. Now they're making LEDs where the most difficult part is the thermal design, and this generally requires a significant proportion of exposed aluminium to dissipate the heat. They don't seem to have the experience required to do this safely, i.e. keeping the mains potential a safe distance from the exposed aluminium whilst still having a good thermal contact between it and the LED junction (often at mains potential). I was a bit surprised at Philips falling into this trap though.

[Adam Funk]

(I'll take the risk of asking a stupid question here.) Isn't that problem --- keeping the mains potential a safe distance from the exposed [metal] whilst still having a good thermal contact --- rather like designing kettle & immersion heater elements?

[alan]

Do Philips make anything these days or are they just re-branding electronics purchased from the cheapest source?

[Andrew Gabriel]

It's not a stupid question at all - it's the essence of the problem.

Yes and no.

It's like trying to make a kettle or immersion heater which requires no earth, i.e. it's double insulated, as there's no earth connection to a light bulb. This is not the case with immersion heaters and kettles, which are normally earthed.

Also, there's no requirement for good thermal contact between the resistance wire and the element casing. The wire gives off

3kW, and it will simply get hot enough to pass that 3kW across the electrical insulator, regardless of how thermally insulating it is (as long as the resistance wire remains below its melting point). It might well be running at 800C+ above the casing at 100C. That won't work with an LED where the challenge is to keep the junction at as low a temperature as possible, and the thermal conductivity must therefore be as good as possible, or both the efficiency and the life of the LED drops rapidly.

[fred]

In article , Andrew Gabriel writes

Spot on analysis IMV, a serious and surprising failure on Philips' part. Even if they self certify on safety I would expect the analysis and testing to have been carried out by a totally separate branch of the group who would dissect the design very carefully. I've certainly gained no favours in the past from in-house testers.

Given the compact dimensions perhaps we can assume reinforced rather than double insulation but that, in itself, should have called for more stringent analysis and testing. Manufacturing defect?

[Brian Gaff]

Completely impossible to navigate as bits are permanently hidden.

Brian

[Adam Funk]

:-)

Interesting, thanks. Out of curiosity, what is the stuff they use between the resistance wire & the casing of a water-heating element?

[meow2222]

snip

I expect the design and samples passed, but the manufacturer cut corners later, and got away with it for a while.

NT

[Andrew Gabriel]

Magnesium oxide is the traditional material for many decades, highly compressed as the element casing is rolled down to its final diameter though multiple pressure rollers.

Its downside is that it is hygroscopic (likes to absorb water), and if the outer casing is not perfectly sealed, moisture will get in and generate earth leakage, and eventually failure (even if the heater is not a submerged type).

It would not surprise me if some other suitable non-hygroscopic material had been found more recently, but I haven't heard of any such.

[Adam Funk]

Right, I remember that now.

Probably too expensive!

[John Williamson]

As long as it outlasts the guarantee, there's no incentive for the makers to find anything better. I've heard of very few elements failing within the guarantee period, so the technology used must be "good enough".