Frank, that is an interesting and very good analysis. Your obvious deep knowledge makes it "easy" for you to see the situation. On the other hand, not everyone has such a deep understanding. It might have been useful for the designer to model the system in a CAE tool that can handle multiphysics. The interaction of electrical and thermal effects is very important in electronics, and as you found out, often overlooked.
The design process for a switching power supply design does take temps and ripple current into account for the derating of the capacitor. The various switcher controller manufacturers have simulation and design automation software services available to make a design very effective even for a novice.
In my experience most all output filtering capacitor failures are one of two issues. Problem one and the most prevalent is poor capacitor performance due to electrolyte formulation, or counterfeit components. In one issue we had a supplier that attempted to replicate the Panasonic electrolyte with disastrous results. Not one of their capacitors met the lifetime rating listed in the datasheet. The parts were failing at 1/10 of the rating on the sheet. Needless to say this was a big issue costing the company a significant amount of money.
The second issue is that the environment experienced is not what it was designed for. We have had units placed in very high temp ambients, far greater than the posted ratings. A rule of thumb is that every 10C increase in temp results in a reduction in life by 1/2. Given that you bought the unit used, you cannot rule out if it was placed in a high temp environment.
Switching power supply design is a very mature specialty, which can be fraught with pitfalls just like any other activity. While bad designs are out there, sometimes the explanation is much more simple than the derating choices made by the engineer.
The use of a single diode past the switching stage points to a flyback design. If that is indeed the case, full wave rectification is not an applicable concept, and the assumpton that the charge/discharge current in the capacitor has a 50% duty cycle would not be correct except at a specific (and low) input voltage. Flyback designs work well and are quire common, but they do need a decent HF rated output capacitor.
Yes, there are capacitors that are designed for switching power supplies that have long life, but unfortunately Chinese makers of consumer products usually put in the cheapest part they can find. Panasonic has a full line of electrolytics for every application, but they are expensive. Most failure of consumer products are because of faulty aluminum electrolytic capacitors. It is the duty of the designer, unfortunately, to assume that cheap or couterfeit parts will be used. Even if the the designer approves the prototypes, the capacitors can be changed or substituted for counterfeits in production.
This power supply was not a flyback. It was used for isolation from the power line and had a two-winding transformer. I understand the compromises of using a full wave rectifier in that it takes a center-tapped winding or four diodes. You are correct that a flyback cannot use a full wave bridge and they usually spec low ESR capacitors for best efficientcy.
I do not remember the value, but I think that it was 1000 uf. I replaced it with a Panasonic capacitor or the same value. The replacement capacitor was intended for power supply use.
Mr. Karkota states what I have thought to be obvious. But as Naperlou, observes, that may not be true. To the considerations of pathologies of failure, let me add a few of my theories. The transformer output from a switching supply is more 'square' than the output of a linear transformer (translation: it has faster rise times). These faster rise times translates to higher surging charge currents. The inverse it also true; the faster fall times translates to higher surging discharge currents. The ESR plus higher surge currents equals heat; as Mr. Karkota states. But, the inductive component of the electrolytic capacitor plays a larger part as the frequency goes up. While there are electrolytic caps specified with a low-ESR, few if any are spec'ed with a low-ESL (effective series inductance). This ESL would cause uneven charge distribution down the length of foil in relation to the charge-discharge(C-D) frequency. All other things being equal, the amount of capacitance 'seen' by the circuit will diminish with increase in frequency.
To solve this problem; If 3000uF was needed for filtering on a switching supply; should one use a single 3000uF cap or three 1000uF caps? The 1000uF's are the better option because the C-D currents would be distributed across the three caps. A lower current is lower heat produced.
Additionally, derating the capacitance in relation to heat is obvious; but what about derating the cap voltage and 'uprating' the ESR?
Then there is how components are packed into smaller spaces creating higher temp operating environments.
It is often amazing that some piece of equipment, while having some specialized part of the circuit be well thought out and a good design, will then have a power supply that looks like it was dsigned in a real hurry by an inexperienced individual, or possibly designed by an accountant. And so under rated parts are tightly packed in an area with inadequate ventillation. And so the power supply fails while the balance of the system is OK.
Great piece - thanks. The multi-discpline detail is what makes electronic product development so much fun... and so painfull.
It struck me that were this a cable television or other set top box in the USA today you probably would not have been able to fix things quite so rapidly. (Guessing this applies in other continents too now). The Energy Star standard would make that front end voltage down-conversion somewhat less accessible visually. A bad design could still show bulging caps of course.
Also, the pesky unwanted noise emission issues arising from the new tighter energy standard today would likley mean that a senior engineer would have had to think perhaps just a tad more about the front end supply design these days. More thought... more components and of course all the more opportunity to accidetally screw things up.
Chances are you could still see a big fat non-SMT cap in that same position though.
Nothing will last forever. I am sure a designer can design a product to last for 50 years but the end cost for the consumer would be very high. We are used to paying cheap prices for stuff these days, even though we know it will only last a couple of years.
It has to do more with economics instead of engineering.
A new service lets engineers and orthopedic surgeons design and 3D print highly accurate, patient-specific, orthopedic medical implants made of metal -- without owning a 3D printer. Using free, downloadable software, users can import ASCII and binary .STL files, design the implant, and send an encrypted design file to a third-party manufacturer.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.