Philips 10.5w 800 lumen premature failure

[PhotonWrangler]

Well I just had a third failure on these bulbs. It was occasionally flickering for several weeks and yesterday it suddenly quit just like the previous two.
:shakehead :thumbsdow

 

[xolotl]

These are now EOL and have been replaced with a 9.5w 806lm model. No noticeable cosmetic differences but hopefully the lower power leads to less failures.

 

[PhotonWrangler]

Thanks for the tip. I'm hoping the lower wattage indicates a redesign of the power supply.

 

[SemiMan]

Possibly but not necessarily.

 

[PhotonWrangler]

Oddly the latest dead one has come back to life. There's got to be a cold solder joint in there somewhere.

Now a fourth one has started to flicker randomly. WHAT the...

**Update**

I've received an unsolicited email from Philips offering free replacements for these lamps. Very nice of them to step up to the plate on this issue. Stay tuned...

 

[PhotonWrangler]

I just lost another one of these lamps today. Same failure mode - occasional flickering, then poof. And Philips never followed up with that offer for replacements.

 

[Arilou]

I replaced these with Lighting Ever 810lm 3000K warm whites. Haven't had any of those fail so far.

 

[PhotonWrangler]

Thanks for the tip Arilou. I will look into these.

 

[Illum]

Unfortunately more copycats choose to use Philips brand name as their choice to brand clones than any other respected company.

The electrolytic capacitor is probably the component most vulnerable to degradation/failure over the long life of the LED lamp. However, if the lamp failed sooner, it was probably something else. It could have been an overheating transistor (resulting from a mistake in assembly, or overall design that did not adequately consider overheating of the circuitry). If the heat is adequately dissipated, and it is fed with a stabilized current supply, the LED chip itself should last nearly forever. It is the circuitry in the power supply that is likely to fail before then.

Few consumer electronics actually taken consideration the thermal dissipation beyond the high current resistors. Often the potting goop tossed in to insulate certain high frequency regions of the circuit and improve the vibration resistance often inhibits heat relief, perhaps even creates hot spots. If the Electrolytic capacitor did fail, or perhaps the tolerances are wide open, it may provide insufficient filtering for the circuit, causing the noise from supply to be more apparent on the output side. That's one possibility, but not the only possibility

I wonder if there is any circuitry design that could prevent all flicker (giving a smooth current supply to the LEDs) without an electrolytic capacitor.
(and yes, I realize that most LED bulbs do have electrolytic capacitors and still flicker)

Certainly, Voltage/current supervisor ICs exist, but for consumer products they are more tuned to price point targets, which electrolytic are a simple logistical solution. Besides that, what sort of "flicker" are you referring to? As in what frequency?


If I have to guess one of your TO-92 devices failed... or your MOV/MLV [blue one in the middle] is starting to fail.
Toroid core inductors, ceramic caps are almost impossible to be destroyed unless they incurred blunt trauma
Electrolytic cap, transformer and ferrite inductors may fail without warning and exhibit almost no symptoms unless it was a catastrophic failure. The transformer should never fail unless a twist of events caused it to saturate... but you'll smell it way before you notice the flicker.

There has to be a control IC somewhere under all that goop, and a bridge or an optoisolator that provides feedback to the chip. The failure of that is also unlikely. I have no idea how to test ICs, but the rest of passives should be relatively straightforward

 

[PhotonWrangler]

The flicker happens in short bursts. The bulb can be running just fine for maybe 20 minutes, then there are three or four quick flashes, then back to normal. The net effect is as if someone fired three or four flashbulbs in rapid succession. It happens at completely random times and is always the same pattern. I'm not sure that a failing electrolytic cap can cause this - it's more like the control IC is freaking out.

I just looked at the pcb that I removed from a dead one and I don't see a controller IC, just a couple of transistors, a handful of diodes and some passive components.

 

[SemiMan]

- The "goop" called potting compound is there primarily for heat transfer. It offers better conductivity versus air that has 0 movement. It may offer some sound reduction.

- Consumer electronics companies such as Philips most certainly consider thermal design and will even have failure models tied into warranty support cost models

- Supervisor IC's have nothing to do with flicker and to answer Ander's question, a capacitor will always be needed where the input is AC. You can eliminate electrolytics but it is not cheap and bulb size may require them.

- No bridge or optoisolator is required for feedback

- A MOV would need to fail short to cause flicker ... unlikely and that would cause flicker in other lights (short) ... and how would you ascertain it was a "TO-92" device that failed?

- Reality with that flicker is it could be anything ... bad solder joint, bad controller, insulation failure in the transformer, etc. etc.

- Control IC will be on the back of the board

Semiman

 

[PhotonWrangler]

At first I suspected cold solder joints, but none of the flickering bulbs responded to tapping on them. That doesn't completely rule it out but it adds to the mystery.

 

[Illum]

The "goop" called potting compound is there primarily for heat transfer. It offers better conductivity versus air that has 0 movement. It may offer some sound reduction.

So why is the goop often tossed over parts where very little thermal relief exists? Like right on the pcb bearing no traces? I see the goop quite often in ATX power supplies whereas the region between heatsinks and the PCB is completely devoid of it?

Consumer electronics companies such as Philips most certainly consider thermal design and will even have failure models tied into warranty support cost models

I never said Philips didn't, but sandwiching all these components into an enclosed space just under the space limitations of a socket will undoubtedly put strain on the components. Such strains amplify the variances from manufacturing of those components

- Supervisor IC's have nothing to do with flicker

Very well, I stand corrected, what then would you suggest from an IC perspective that can eliminate flicker?

- No bridge or optoisolator is required for feedback

We haven't figured out what is driver is yet, trouble shooting isolated power supplies once awhile I run into optoisolators:duh2:

- A MOV would need to fail short to cause flicker ... unlikely and that would cause flicker in other lights (short) ... and how would you ascertain it was a "TO-92" device that failed?

From experience it seems TO-92 transistors fails most often while exhibit no symptoms from looking at it... at least from the low end Made in China products. I've lost count on how many I've replaced now. I didn't ascertain it, I speculated it may be
When an MOV fails, its quite exciting... most if not all the MOVs I've replaced from UPSs, strip sockets, or switchboards have failed quite spectacularly. However that kind of failures probably bears no relevancy to this.

- Reality with that flicker is it could be anything ... bad solder joint, bad controller, insulation failure in the transformer, etc. etc.

If it runs smoothly until a certain time then started to flicker, I like your thought of insulation failure. Bad solder joint doesn't need to heat up to show its bad, but insulation certainly could.

- Control IC will be on the back of the board

pardon for my asking, but where do you see it?
https://farm4.staticflickr.com/3705/9002025847_b8924e4e61_k.jpg
All I see are diodes

 

[PhotonWrangler]

https://farm4.staticflickr.com/3705/9002025847_b8924e4e61_k.jpg
All I see are diodes

They are indeed diodes and they all checked out fine.

 

[SemiMan]

Looked at the parts list posted and makes sense why there is no IC now. The two HV transistors are likely set up in a push-pull configuration and use the transformer inductance and other passive parts to form an oscillator. This was common in fluorescent lamps. This is the non-dimmable one right? No need for anything very complex. Saves the money of the driver IC.

In a bulb you need the potting compound for heat transfer. In a bit ATX supply with forced air that is not much of an issue, but you do have large caps and even inductors that you need to prevent moving and breaking solder joints. You may have isolation you need to improve as well.

All circuits are susceptible for tolerance issues .. that is just good design practice. Size the heat are not the big issue ... cost is.

Nothing from an IC perspective is going to solve flicker. You need storage plain and simple and the smallest and cheapest way is an electrolytic cap.

 

[PhotonWrangler]

Looked at the parts list posted and makes sense why there is no IC now. The two HV transistors are likely set up in a push-pull configuration and use the transformer inductance and other passive parts to form an oscillator. This was common in fluorescent lamps. This is the non-dimmable one right? No need for anything very complex. Saves the money of the driver IC.

I think you're spot on here, Semiman. The two TO-92 transistors are located on either side of the transformer and the traces look like they do operate it in push-pull. And yes, this is a non-dimmable lamp so there's no need for fancy control circuitry.

Just to be clear, these lamps don't normally produce visible flicker, so the big electrolytic at the DC output seems to be good enough. However when they go into failure mode, they flash on and off rapidly a few times, then they settle down again.

 

[gearhead1972]

I just got two of these, my first LED A19 purchases. I will keep you informed of any failure. They both will get frequent daily use, one on the front porch in a table top lamp, the other in the basement in a pull string overhead socket. The second one is not facing straight down though, it is in a y splitter so it is at a 45 degree angle.

 

[EngrPaul]

I've had a Samsung LCD TV and a Samsung Monitor from work both go bad because of blown capacitors.

I stumble across this article: http://en.wikipedia.org/wiki/Capacitor_plague

 

[PhotonWrangler]

I've lost a couple of motherboards due to this capacitor plague. Wish I know if this is why the lamps are failing.

 

[amd20x6]

My Linksys WRT-610n router died because of bad Samxon caps about a year after I bought it. I replaced them with Nichicon aluminum polymer caps of about 1/2 the rated capacitance. It's been fine for about 4 more years and counting.

Their failure mode was high ESR. This was confirmed with an ESR meter.

I saw Samxon in an LED bulb teardown somewhere... I don't remember which one.