EEVblog #742 - Why Electrolytic Capacitors Are Connected In Parallel

[VK3DRB]

A useful website for calculating the life out of a capacitor... http://www.illinoiscapacitor.com/tech-center/life-calculators.aspx

Dave hit the nail on the head over TV capacitors.

To cut costs, many CRT TV brand names often used no-name branded capacitors. Typically in the vertical driver stage they were used to block DC, allowing the AC vertical signal to pass only to the vertical output amplifier (they were often 100uF to 470uF in size), and the AC currents (and heat) caused the electrolytic capacitors to dry up. The effect was a reduced vertical image size.

Another area where dried caps were commonplace was in medium to low end audio amplifier power supplies. The symptom was a loud mains hum in the audio. In cheap audio amplifiers often found in radios and low end consumer amplifiers, electrolytic capacitors were used to block any DC from the power amplifier output getting into the speaker. Often, these capacitors passing a large AC current into an 8 or 16 ohm speaker would dry up and the effect was lack of bass in the audio. In real bad cases the entire volume would diminish or the sound would be very tinny.

[VK3DRB]

Dave nicknames 100uF capacitors "100 mike caps". Maybe that is a Sydney term... I have no idea. I don't recall technicians and engineers using that term here in Victoria. Nothing wrong with it of course, because it is clear and not ambiguous.

Here, 100uF capacitors are nicknamed "100 muff caps".
100pF capacitors are nicknamed "100 puff caps".
100nF capacitors are nicknamed "100 nano caps" or "Point one muff caps".

[Dino KL0S]

Well, at least here in the U.S. WAY BACK in the day when we still called them "mmF" (micro-micro Farads) they were nicknamed "mickey-mikes"; "puff" remains in use but never heard of "muff" or "Point one muff".  Although folks use the term "nano", that in between range never rings true with me and I have to mentally convert up or down since standard values are most often referred to as in terms of uF or pF.

[station240]

mike is short for micro, as it's micro-farad.

[EEVblog]

Well, at least here in the U.S. WAY BACK in the day when we still called them "mmF" (micro-micro Farads) they were nicknamed "mickey-mikes"; "puff" remains in use but never heard of "muff" or "Point one muff".

Same here, first time I've ever heard it called "muff". Muff means, err, a part of the female anatomy 
Sometimes "mu" or simply "you"
Likewise nano can get shortened to just "n", like kilo gets simply "k"

[Votality]

Well, at least here in the U.S. WAY BACK in the day when we still called them "mmF" (micro-micro Farads) they were nicknamed "mickey-mikes"; "puff" remains in use but never heard of "muff" or "Point one muff".

Same here, first time I've ever heard it called "muff". Muff means, err, a part of the female anatomy 
Sometimes "mu" or simply "you"
Likewise nano can get shortened to just "n", like kilo gets simply "k"

Hahahah thanks for the laugh Dave .. Being from oz. I agree with your definition of a muff.

Also am I paranoid or what or do tv manufacturers seem to stick all the power supply caps above the bulk heat sinks in the lcd's thus nicely warming them and reducing their operational life... Haven't they ever heard of convection.

Planned obsolescence?.

[Fungus]

Well, at least here in the U.S. WAY BACK in the day when we still called them "mmF" (micro-micro Farads) they were nicknamed "mickey-mikes"; "puff" remains in use but never heard of "muff" or "Point one muff".

Same here, first time I've ever heard it called "muff". Muff means, err, a part of the female anatomy 
Sometimes "mu" or simply "you"
Likewise nano can get shortened to just "n", like kilo gets simply "k"

Hahahah thanks for the laugh Dave .. Being from oz. I agree with your definition of a muff.

They call it that in the UK as well.

PS: "Puff" can mean bad things in the UK. There's no escape.

[donkey77]

I wasn't the only one to notice this then!

[DanielS]

Dave hit the nail on the head over TV capacitors.

To cut costs, many CRT TV brand names often used no-name branded capacitors.

No-name and low-cost Chinese caps are in many modern consumer electronics too. How many people have LCDs and PC power supplies that failed due to Fuhjyyu, SamXon and other Chinese brand capacitor failures or the power supply simply exceeds the capacitors' ripple current rating?

[Paul Moir]

If electrolyte drying is the age failure, I would think that the ratio between the aluminium part of the can to the rubber stopper would have a big impact on lifespan, assuming that gross leaks in the seams aren't the real culprit.  I wonder if that doesn't partly drive the trend towards tall narrow packages?

[DanielS]

If electrolyte drying is the age failure, I would think that the ratio between the aluminium part of the can to the rubber stopper would have a big impact on lifespan, assuming that gross leaks in the seams aren't the real culprit.  I wonder if that doesn't partly drive the trend towards tall narrow packages?

The "bung" is intended to be permeable to hydrogen so hydrogen gas which may form during heavy load and fail to recombine can percolate through it instead of blowing the casing off the bung or bursting the pressure relief valve which would create a much worse leak and potentially more spectacular failure.

As for the tall form factor, I think the main drive for it is lower PCB footprint cost: almost everything inside PSUs is getting smaller except inductors and capacitors. Once your board size is basically dictated by the size of your electrolytic capacitors, the only way to shrink the PCB is to use taller caps. In most PC PSUs I have opened up, the output caps are tightly clustered, packed about as close as their footprint will allow. If the original caps were 8mm diameter, there rarely is enough room to fit 10mm caps. At least not neatly.

[Paul Moir]

Thank you very much DanielS.  That is very interesting:  it explains why more hermetic systems are not used on electrolytic capacitors.  I had presumed it was for cost savings since the packaging appeared simpler to make.

[Votality]

Dave hit the nail on the head over TV capacitors.

To cut costs, many CRT TV brand names often used no-name branded capacitors.

No-name and low-cost Chinese caps are in many modern consumer electronics too. How many people have LCDs and PC power supplies that failed due to Fuhjyyu, SamXon and other Chinese brand capacitor failures or the power supply simply exceeds the capacitors' ripple current rating?

But not only that why in so many designs are the caps orientated directly above the heatsinks in the convection path.

[DanielS]

But not only that why in so many designs are the caps orientated directly above the heatsinks in the convection path.

The simplest answers to that would be:
1- because that's where the output rectifiers are, especially in flyback converters where you need to minimize stray inductance
2- layout constraints (connectors, heatsinks and other mechanical or structural components have predetermined locations on a predetermined board size)
3- the final orientation was unknown and could not be taken into consideration for board placement
4- it simply makes sense to put them there from a PCB routing flow point of view
5- they simply put them where they had spare space
6- they simply didn't care

I'm sure there are many more possibilities.

[LabSpokane]

http://www.chemi-con.co.jp/e/catalog/pdf/al-e/al-sepa-e/001-guide/al-technote-e-140701.pdf

Nippon Chemi Con's app note on electrolytic capacitors.

Incidentally, one of Chemi Con's aluminum foil factories is located in Moses Lake, Washington about an hour and a half away. I'm hoping I can score a tour some day. (We have the world's lowest power rates here, so we traditionally have had a lot of aluminum plants in the area.)

[errsu]

An often-stated rule is that electrolytic cap lifetime doubles for every 10 degrees C reduction in temperature.

Interestingly, this rule is what I remember from school for the general speed of chemical processes.

So, in Daves example, the diameter-dependent livetimes are 5000 and 7000 hours. The temperature difference where smaller gets better is then 10*log2(7000/5000)=4.85 K. A little more than in the example, but probably a common case in high-power setups.

[rsu]

[free_electron]

I wasn't the only one to notice this then!

could have been worse ... wwwank ... world wide wa... let's not go there.

[m100]

What never ceases to amaze me is the failure rate of cheap no recognisable brand caps in so called quality consumer equipment that clearly runs at relatively low temperatures, supplied by external linear or switch mode power supplies.  They seem to degrade to useless in a few years when operated at something not much above UK room temperatures.

Despite the drive for lower and lower costs I'd have thought that by now everyone in the business of manufacturing caps would have got some idea how to manufacture something that could withstand life outside the component drawer.   It feels like the failure rate now is an order of magnitude or two above what it was in say the early 80's

[djQUAN]

I think it has something to do with engineered obsolescence or whatever it is called.

[Fungus]

An often-stated rule is that electrolytic cap lifetime doubles for every 10 degrees C reduction in temperature.

Interestingly, this rule is what I remember from school for the general speed of chemical processes.

We used to call it Q10

[Len]

I think it has something to do with engineered obsolescence or whatever it is called.

I think so too. In the last few years I've had a few devices that failed due to bad electrolytic caps, but they were all past their warranty period. (Some not very far past.) From the manufacturer's point of view, that's proper value engineering.  Sigh.

[tszaboo]

One thing that still irks me is when I see a schematic published/shared that just shows "2x 470uF". If it said, "2x Panasonic 470uF model XYZ", or "1000uF with < 0.1 ohm ESR" then that would be somewhat rational.

Because that information doesn't belong to the schematic, that is the Bill of Material.

One thing which should have been mentioned, is that you cannot just mix capacitors and hope for the best. Meaning if you are using 5 electrolytic for 1000uF, they have to be the same type (unless very carefully investigated) they cannot be like 4x 100uF and 1x 560uF. ESR is different, ripple current is different and so on.
Otherwise, excellent reference video.

[Smokey]

...
David2 etched a board and everything.

Did you make him hand tape the layout as well?  Carve the BOM in stone tablets? 

[rs20]

One thing that still irks me is when I see a schematic published/shared that just shows "2x 470uF". If it said, "2x Panasonic 470uF model XYZ", or "1000uF with < 0.1 ohm ESR" then that would be somewhat rational.

Because that information doesn't belong to the schematic, that is the Bill of Material.

Which one? Do you mean this?:
-- Schematic should say "1000uF with < 0.1 ohm ESR"
-- BOM should say "2x 470uF"

[GoldSrc]

Could you eliminate the ripple at almost any load if you use a lot of capacitors in parallel to get close to 1 Farad?