LNK305 and the likes mass dying... any reason for it?

[Neutrion]

So there are plenty, mostly consumer equipment where this, and similar parts are used, and they seem to be dying in such a huge masses, that for washing machines there are complete kits with fuse resistor, etc, including this IC  aviable to repair the boards.
Of course most of the boards are never get repaired.
Different appliance manufacturer are affected from different years, and these ICs are failing seemingly in a very short time, usually shortly after the warranty period is over.
What is the cause? I could not found any good explanation to it anywhere on the net. 
None of the manufacturer is able to properly mount it to give it a proper cooling? (It even has overtemperature protection.) Voltage spikes?
Bad batches of the parts?
It causes such a huge amount of equipment failures, that possibly tens of thousand tonns of waste is produced yearly only because of them.
And no, its not made by Apple, but it would fit well into their concept 

[BrokenYugo]

You're probably right, too hot or not enough protection. Just the usual penny pinching.

Keep in mind overtemperature protection is generally set closer to "Guaranteed to not desolder itself!" than "Will last a long time bouncing off the limit".

Also, aren't most of these appliance brands largely under the ownership of 2 or 3 companies now? This may even all trace back to a single design mistake that was simply never corrected.

[Kerlin]

Well ..... imagine you bought some things from someone who hates you, would you really be surprised if they often went faulty?

[Neutrion]

You're probably right, too hot or not enough protection. Just the usual penny pinching.

Keep in mind overtemperature protection is generally set closer to "Guaranteed to not desolder itself!" than "Will last a long time bouncing off the limit".

Also, aren't most of these appliance brands largely under the ownership of 2 or 3 companies now? This may even all trace back to a single design mistake that was simply never corrected.

If the design uses the part within its rated consumption, it should run way below the overtemp limit, and basically should last forever, especially as the IC has - in a case of a washing machine which is 1-2 yeas old - only 1-2000 working ours with the part working hard.

So either non of the manufacturer is able to solder the part in with its base properly connected to the board, or the something is not well documented in the ICs documentation regarding its surge protection.

But how can it be, that this broblem is ignored by almost all of the manufacturers for years?

[amyk]

These single-chip SMPS ICs seem to be pretty fragile especially to voltage surges.

[shakalnokturn]

Well ..... imagine you bought some things from someone who hates you, would you really be surprised if they often went faulty?

Are you trying to say that the appliance manufacturers hate the consumers in general? (Or just Neutrion?)

[Neutrion]

These single-chip SMPS ICs seem to be pretty fragile especially to voltage surges.

So at the moment we have 3 possibilities:
1.Bad mounting of the device, so it works on a higher temperature than it would be optimal.
2.Voltage spikes.
3.Bad general design of the IC.

Is there a way to find out which one applies here?
 

[wraper]

1.Bad mounting of the device, so it works on a higher temperature than it would be optimal.
2.Voltage spikes.
3.Bad general design of the IC.

Is there a way to find out which one applies here?

Measure the actual temperature.

[floobydust]

These IC's are built to a cost point and end up being a bit fragile, running 24/7 in many applications and subjected to mains transients.
The power mosfet and current rating are sized to the micro penny. Look at die pictures of LNK306GN , similar TOP250YN also shows the family of parts are just scaled up.

[mikerj]

It's possible the device could be sensitive to excessive input ripple, constantly shutting down and restarting every cycle.  Filter caps will likely be the bare minimum and not the highest quality parts.

[Neutrion]

1.Bad mounting of the device, so it works on a higher temperature than it would be optimal.
2.Voltage spikes.
3.Bad general design of the IC.

Is there a way to find out which one applies here?

Measure the actual temperature.

The one which we have in the washer is heavily modified for better cooling, and I can not undo it without much damage. And of course I am not sure if this board is affected as well so we would have to measure one it in a board which is definitely known for this problem.

These IC's are built to a cost point and end up being a bit fragile, running 24/7 in many applications and subjected to mains transients.
The power mosfet and current rating are sized to the micro penny. Look at die pictures of LNK306GN , similar TOP250YN also shows the family of parts are just scaled up.

Thanks for linking this topic! Amazing pictures, wanted to start to look for such pictures for a while.
Did anyone untill now pointed out the exact part which may be the culprit? The 306 should be similar to the 305, but I think  both are falling like flyes.
These are although connected to the 230V 24/7, but the actual working hours where they have to work at their rated level are very low numbers, like few thousend hours, so maybe not the power mosfet is the problem?
Anyway after the top of it is blown away, is it still possible to do an autopsy to point out the exact part which failed? I don't have any blown one  by the way.

It's possible the device could be sensitive to excessive input ripple, constantly shutting down and restarting every cycle.  Filter caps will likely be the bare minimum and not the highest quality parts.

I wish someone had a programmable surge generator with relative low energy level, and a high voltage scope probe to test what kind of voltage spike would cause any damage. But we don't know anyone with such equipment do we?  

By the way I just checked the website of the manufacturer Power Integrations, and now they seems to have a new series with number 2: like TN2 series of TN, or XT2 series instead of XT.
There is a small difference even in the schematics.