Electrolytic capacitors bad news for anything on full power for long-ish time

[bdunham7]

Thanks, ...I mean, if you search digikey for  “Aluminium electrolytic capacitor, 450v, 47uF”…..there are virtually none above 105degc.
If one operates an  El cap above quoted temp, then it can suffer  serious damage, significantly affecting its lifetime. I have a Cornell Dubilier doc which states this. The “10 dgec” law doesn’t work when you go above the quoted temp.
A 60w offline SMPS, totally enclosed in a plastic enclosure, will easily get above 105degC  (internal ambient) at 30degc  external ambient.

Here is such a power supply....one of many out there..
https://www.mouser.co.uk/datasheet/2/260/LPC-60-spec-1109552.pdf

This is why an alternative to El caps is well needed.

Are you saying the supply you listed will get above 105C (evidence??) or are  you saying that such a supply cannot be designed not to exceed 105C?

The part you listed has a max ambient of 70C, but is derated above 50C.  If you can't design it to stay under 105C, what is the Tj of the semiconductors within?  When does the case start to melt?  Does the part you listed(an economy LED-driver b/t/w) have el-caps?  Does that supply have a significant failure history?

Oh, here's a nice film capacitor you could use.

https://www.mouser.com/ProductDetail/KEMET/C4AEGBW5500A3LJ?qs=pqRVuuzkf6aQc8hsOFRzpg%3D%3D

Too bad it is only rated for 85C !!

[dmills]

Also, when you are buying Meanwell volumes, you can afford to either qualify parts yourself, or get your preferred vendor to qualify a part for off datasheet use if you really must.

It is quite possible to say to a vendor, we would like to use your part XXXXX-YYYY and have volumes of a million a year, but we need you to qualify that part for 125 degrees C, can you guarantee X hours at that temperature if you derate ripple current by 50%? And because you are suggesting a million a year, they might put a reliability engineer on it to run the numbers for you. 

Not that I think internal temps are getting that high on those stupid little line lumps.

[ocset]

Thanks, i believe in totally enclosed plastic at 30degc ambient, (with the stated power supply) you'd be looking at 105degc internal ambient....there's just nowhere for the heat to go.

Film caps  have generally a max temp of 105degc aswell. Though you need to derate voltage in most cases to  get this.
Eg the MKP1848C64580JY 800V film cap can do 10000 hrs at 105degc and up to  560VDC
However, if for a short interval, the film cap is taken above 105degc then it will be  ok,  whereas an El cap getting taken higher than datasheet quoted temperature  wont be ok. (as cornell say)
Also, at 105degc ambient, the internal temperature of a film capacitor will  be close to 105degc, because the esr is so low……eg, the above one is 5 milliohms ESR….whereas the El cap will tend  to have a higher internal “hot-spot” temperature due to its higher esr.

[dmills]

Lets do sums.

That 60W Meanwell specifies 85% efficiency (presumably at full load), so ~9W of heat to reject.

The derate curve says it can do this at 50 degrees ambient.
 
The case dimensions are a cube 162 * 42 * 30, so 0.026m^2 so the power flux density is ~350W/m^2, as I recall there is a product safety standard that limits the temperature of exposed surfaces, cannot remember the details, but lets assume say 70 degrees outer case temperature is the limit there, should be ballpark?

Lets further target an internal temperature of say 100 degrees, then we have 30 degrees across the case wall, so thermal conductance must be at least 11.6W/m^2K Lets see where that gets us?

Most plastics come in at somewhere around the 1w/mK, so as long as the wall is thinner then 8cm we will meet that target.
In practice the wall will always be thinner then that so the internal temperature will be lower, the actual limit being set by the limit on the outside surface temperature for product safety reasons, at the usual few mm thickness, the internal temperature and case external temperature will be very similar because the thermal resistance of the case is swamped by the crapness of the case as a heatsink.

I think that if the outer case is at 70c the internal air temperature is likely only a few degrees higher, so a 105 degree part is likely well within ratings.

Someone check my arithmetic.

You got to run the numbers, it trumps guessing wrong!

[Siwastaja]

0.2 W/mK for generic "plastic" will be closer. 1W/mK could be a special thermally-conductive plastic with fillers, even then it's a good number.

Otherwise than that, the problem is always hotspotting in the few heat-generating components, which almost always are a tiny fraction of the total product volume.

Your arithmetic assumes the heat is spread internally to the whole volume of the product. It would be nice if it was possible or easy to get even near to that!

This is why fan cooling is such a big thing in power supplies after maybe about 200W: many heat-generating components have odd shapes so that mounting heat spreaders or heatsinks is next to impossible: toroid inductors, transformers with the round windings, round electrolytic capacitors... Only the semiconductors are designed to be cooled using metals to transfer heat elsewhere.

[tooki]

The most life-threatening accessory I owned was the genuine Apple charger cord that came with my iPad.  After a year or two, the outer sheath began to come off, and it began to throw off sparks before it caught fire.  A genuine replacement cord (that was barely long enough to be useful) was over $50.  I went to the gas station next door and bought a new $3 cord that worked perfectly for years.

No, the genuine Apple cord didn’t cost $50. They cost around $20 for the standard 1m length. High enough that you don’t need to resort to fabricating hyperbolic prices to make your point.

[ocset]

lets further target an internal temperature of say 100 degrees, then we have 30 degrees across the case wall,

Thanks very much for this, i think the maths looks all good. However, i am not sure about the assumption of 100degc internal ambient.
Also, the "70 degc Max  external case temp" i am not sure about.....i mean light bulbs are well above 70degC externally and people can touch them.

[dmills]

0.2 W/mK for generic "plastic" will be closer. 1W/mK could be a special thermally-conductive plastic with fillers, even then it's a good number.
Otherwise than that, the problem is always hotspotting in the few heat-generating components, which almost always are a tiny fraction of the total product volume.

Granted, but 0.2 Vs 1.0 just changes the wall thickness to make the assumption hold (to a still stupidly thick), 16mm.

The point I was getting at is that the limit is usually surface touch temperatures on the outside of the line lump, not bulk internal temperatures.
If you actually open some of those up you often find metal heat spreaders to lower the hot spot temperatures, at least on the sand, by more efficiently coupling the heat to the case. I have even seen the bulk cap with what is effectively a heat spreader!

Telecordia GR-63 for non metals in short term contact specs 70c as maximum temperature, 60950-1 (2005) gives up to 95c for plastic or rubber external surfaces of equipment that may be touched, but UL seems to say no more then 45 degrees above ambient, so 85c if assuming a 40 degree upper ambient limit (This is a bit of a case of going down the rabbit hole to identify and I may have misinterpreted something).

[Connecteur]

I hope capacitors don't follow the same fate as ball bearings.  Years ago, ball bearings lasted nearly a lifetime. Some unscrupulous manufacturers made them to fail prematurely so that they could sell a lot more of them.  It took several decades, but those manufacturers gradually bought out all the long-life bearing manufacturers, and once they had a virtual monopoly they could shorten the lifespan of bearings with impunity.  That's why nowadays, bearing replacement is routine and frequent.

Old machinery with original ball bearings are the best example of this.  They were made to last, and they did.

[ocset]

Years ago, ball bearings lasted nearly a lifetime. Some unscrupulous manufacturers made them to fail prematurely so that they could sell a lot more of them.

Thanks, now that's interesting...the "phoebus cartel"  phenomina...again.
In cooling fans, the biggest cause of failure is bearing failure....said to be due to use of cheap grease, but as you mention it, yes, its probably to do with the poor bearings themselves.

[Gyro]

Years ago, ball bearings lasted nearly a lifetime. Some unscrupulous manufacturers made them to fail prematurely so that they could sell a lot more of them.

Thanks, now that's interesting...the "phoebus cartel"  phenomina...again.
In cooling fans, the biggest cause of failure is bearing failure....said to be due to use of cheap grease, but as you mention it, yes, its probably to do with the poor bearings themselves.

You would expect the bearings to be the first thing to fail, they are the only 'wear item' in a fan... Oh wait, there's an Electrolytic capacitor in there! 

These days, better quality, quieter fans use FDBs.


Edit: In the old days bearings tended to have lubrication points, 'sealed for life' bearings have put convenience above ultimate life to some extent (heavily dependent on whether the owner stuck to the specified lubrication intervals, cleanliness and lubricant specifications).

[Kjelt]

There is something called economic lifetime of a product and TCO.
In these days companies don't want to buy a machine that will last 100 years costs 8x as much than a machine they can write off in 5 to 10 years and replace them pre-emptive or even lease machines.
It makes no sense.

Us hobbieists should do the same when we replace a fan of a benchmultimeter and know we only will use it 4 hours a week on average we don't need an overexpensive fan with bearings that will last decades.
On the other hand if you have a company and it is on 16hrs/5days a week you might.

[Connecteur]

I saved my money for six months to buy the best video card I could afford for my computer.  I was sorely disappointed when it got fried as a result of the cheap cooling fan failing less than a year later.  I swore off buying that brand for the rest of my life.

It must be the farmer in me, who were notorious for holding grudges for decades against brands that weren't made to last.

[Kjelt]

I saved my money for six months to buy the best video card I could afford for my computer.  I was sorely disappointed when it got fried as a result of the cheap cooling fan failing less than a year later. 

That is still within warranty you should get a new one.

But that fast failing products happens with the best products also.
QA people call it infant mortality , look it up under bathtub curve.

[Connecteur]

I saved my money for six months to buy the best video card I could afford for my computer.  I was sorely disappointed when it got fried as a result of the cheap cooling fan failing less than a year later. 

That is still within warranty you should get a new one.

But that fast failing products happens with the best products also.
QA people call it infant mortality , look it up under bathtub curve.

It wasn't covered. I got the FU treatment from the seller.

[SilverSolder]

Pretty poor design, that the card didn't throttle itself back when the fan failed.

[Connecteur]

Pretty poor design, that the card didn't throttle itself back when the fan failed.

Same thing happened to the HP laptop I had when the air vent was covered.  It wasn't covered, and I've never bought an HP product since.

[cliffyk]

An artifact of the Asian manufacturers cranking out ultra-cheap electronic devices has been a steady degradation in the quality of electrolytic capacitors. They are perhaps the most commonly failing components in contemporary consumer electronics. Cheap test equipment and controls are not immune to this...

[CatalinaWOW]

An artifact of the Asian manufacturers cranking out ultra-cheap electronic devices has been a steady degradation in the quality of electrolytic capacitors. They are perhaps the most commonly failing components in contemporary consumer electronics. Cheap test equipment and controls are not immune to this...

This has been a team effort.  The drive to produce really cheap consumer electronics has created a market for these low quality components.  And producers have responded to that market.  And designers of further upscale products have been lured by low prices into using them, reducing the market for better quality products.  Which further reduces their sales.  It is a vicious cycle with blame in many directions.

[cliffyk]

An artifact of the Asian manufacturers cranking out ultra-cheap electronic devices has been a steady degradation in the quality of electrolytic capacitors. They are perhaps the most commonly failing components in contemporary consumer electronics. Cheap test equipment and controls are not immune to this...

This has been a team effort.  The drive to produce really cheap consumer electronics has created a market for these low quality components.  And producers have responded to that market.  And designers of further upscale products have been lured by low prices into using them, reducing the market for better quality products.  Which further reduces their sales.  It is a vicious cycle with blame in many directions.

Oh, absolutely!

It's like the "squawk" about Walmart (or pick any evil big box store) "moving in and destroying local businesses..." Uh--NO--it's the local buyers who chose to trade at Walmart instead of their supposedly cherished "local shops" that kill off "mom and pop". If the community ignored Walmart it would go away or have not been built in the first.

I am as guilty of this as anyone--it is fundamental human nature: I want to be "fairly" paid as much as I can get for my KSAs (Knowledge, Skills and Abilities), and of course shell out no more than i have to for the labours of others...

[Kjelt]

Same thing happened to the HP laptop I had when the air vent was covered.  It wasn't covered, and I've never bought an HP product since.

User obstructed Air vents are user mistakes and imo rightfully not covered by warranty.
A failing fan within the warranty period should.

[Connecteur]

Same thing happened to the HP laptop I had when the air vent was covered.  It wasn't covered, and I've never bought an HP product since.

User obstructed Air vents are user mistakes and imo rightfully not covered by warranty.
A failing fan within the warranty period should.

Perhaps, but the on-board thermal protection had been deliberately disabled by HP, so that they could sell more new computers.  Every other computer I've owned would either throttle back performance or shut down when CPU cooling failed. It resulted in my never buying an HP product again, and my story was told dozens of times to people who asked my advice.  I doubt it was a successful strategy for HP either.

[SilverSolder]

Same thing happened to the HP laptop I had when the air vent was covered.  It wasn't covered, and I've never bought an HP product since.

User obstructed Air vents are user mistakes and imo rightfully not covered by warranty.
A failing fan within the warranty period should.

Perhaps, but the on-board thermal protection had been deliberately disabled by HP, so that they could sell more new computers.  Every other computer I've owned would either throttle back performance or shut down when CPU cooling failed. It resulted in my never buying an HP product again, and my story was told dozens of times to people who asked my advice.  I doubt it was a successful strategy for HP either.

It is pathetically poor engineering when a device that can easily be foreseen to overheat has no protection against it.

[Connecteur]

Same thing happened to the HP laptop I had when the air vent was covered.  It wasn't covered, and I've never bought an HP product since.

User obstructed Air vents are user mistakes and imo rightfully not covered by warranty.
A failing fan within the warranty period should.

Perhaps, but the on-board thermal protection had been deliberately disabled by HP, so that they could sell more new computers.  Every other computer I've owned would either throttle back performance or shut down when CPU cooling failed. It resulted in my never buying an HP product again, and my story was told dozens of times to people who asked my advice.  I doubt it was a successful strategy for HP either.

It is pathetically poor engineering when a device that can easily be foreseen to overheat has no protection against it.

I'm sure the engineers wanted to keep the thermal overload protection, but the MBAs in the corner office got them to disable it.

[Connecteur]

I dated a girl in university who's dad was a top engineer in Detroit. She claimed he invented the original digital dashboard.  She said he was forced into retirement because he wanted to build things to last, but his bosses always wanted him to engineer shorter lifespans.  Good business I suppose, but so is stealing if you don't get caught.