EEVblog #742 - Why Electrolytic Capacitors Are Connected In Parallel

[EEVblog]

Fundamental Friday
Dave explains why some designs have electrolytic capacitors connected in parallel. The answer is more in-depth than you might think.
9 reasons are given and explained, and then some thermal camera fun on the bench.
Rubycon capacitor datasheet: http://www.farnell.com/datasheets/1910904.pdf
NOTE: Video #741 has been skipped for a reason.

https://www.youtube.com/embed/wwANKw36Mjw

[radioFlash]

Nice video Dave, but there is one quibble: the heat dissipation will be dominated by conduction and convection with air rather than block body emission at these low temperatures.

[fusebit]

When do we see the video about the LM741?

[rs20]

Great video, I learned a lot! 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. But otherwise, surely there's a wide range of performance characteristics for both 470uF and 1000uF capacitors -- there must be 1000uF capacitors out there with a way better ESR than 2x some marginal 470uF capacitors. If the original designer looked at what was available in the catalogue and concluded that 2x 470uF was the way to go back in 1973, isn't it likely that what's in the catalogues has changed since then?  I'd almost (but not quite) go so far as to say that this is a question of component procurement and selection at the time of production, not something that belongs on a timeless schematic. Stating a max ESR seems like a far better approach.

Or am I wrong, is there some sort of universal accurate standard for how much ESR a "decent normal" cap of a given voltage and capacitance has?

[Barny]

What happened with the DIY Current sink at the End of the Video?
Is the 100mA Current raise to show the temperature diference better or is it drifted away?

[Refrigerator]

Dave you did it again ! Where did 741 go ?

[Barny]

Look at his twitter account:
https://twitter.com/eevblog/status/596802022814384128

He want to show a special video.

[EEVblog]

Nice video Dave, but there is one quibble: the heat dissipation will be dominated by conduction and convection with air rather than block body emission at these low temperatures.

Shh, don't tell the nerds who are busy trying to calculate it 

[EEVblog]

What happened with the DIY Current sink at the End of the Video?
Is the 100mA Current raise to show the temperature diference better or is it drifted away?

Oh, I upped them both to 300mA off camera but didn't do anything with that.

[EEVblog]

Great video, I learned a lot! 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. But otherwise, surely there's a wide range of performance characteristics for both 470uF and 1000uF capacitors

For sure, massive differences.
You'll see specific brand and models in the DC-DC converter datasheet like the one I showed, but without that you are left to guess why.

[EEVblog]

When do we see the video about the LM741?

No idea. We tried to build something novel but in practice it didn't work. So have to go with a whole new idea.
David2 etched a board and everything.

[errsu]

The datasheet gives longer livetimes for larger diameters. This goes in the opposite direction of the temperature reduction when splitting into multiple capacitors. To know which effect is dominant, the actual dependency of the livetime from temperature would be needed, in numbers. Hmmmm...

[rsu]

[Towger]

Dave you did it again ! Where did 741 go ?

The same place as 666.

[rs20]

The datasheet gives longer livetimes for larger diameters. This goes in the opposite direction of the temperature reduction when splitting into multiple capacitors. To know which effect is dominant, the actual dependency of the livetime from temperature would be needed, in numbers. Hmmmm...

An often-stated rule is that electrolytic cap lifetime doubles for every 10 degrees C reduction in temperature. Interestingly, some solid electrolyte capacitors claim more than a tripling of lifetime for each 10 degree reduction in temperature, which really stacks up if your capacitor is rated at 105 deg C and you're actually using it at, say, 65 degrees (theoretically, 2000 hour cap at 105 degrees --> 32,000 hour cap at 65 degrees for normal cap, 200,000 hour cap for solid cap).

So there you go, have a play with that model. Clearly, which effect is dominant depends on how hot the original single capacitor would get.

But thanks for pointing that out, I hadn't considered that all the surface area of all those smaller capacitors actually means more surface area for them to dry out and age (who knows if that's really the underlying cause, but seems like a nice idea in my head). Key point is, using multiple smaller capacitors might be a strongly bad idea in some circumstances if you don't keep lifetimes in check.

[Stupid Beard]

[EEVblog]

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

Yeah, I should have mentioned that one.
On a look at the video again there is a lot of stuff I left out actually.

[EEVblog]

The datasheet gives longer livetimes for larger diameters. This goes in the opposite direction of the temperature reduction when splitting into multiple capacitors.

If lifetime is your goal then you parallel up the best and longest life ones you can get, you wouldn't shoot for the smaller size just because you can.
Higher voltage also gives longer life, so rise and repeat your options again and again for each design criteria.
Messy business.

[bktemp]

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

Yeah, I should have mentioned that one.
On a look at the video again there is a lot of stuff I left out actually.

Do a follow up.

A follow up would be great. Then you can go a bit more into detail how to calculate/estimate the ripple current and power dissipation inside a capacitor.
Buy one of those, load it with the maximum specified current and measure the temperatur to show how not to design a circuit:
https://www.eevblog.com/forum/projects/adjustable-buck-voltage-regulator-aliexpress-exiciting-finds/

[NiHaoMike]

On a related note, there's a similar reason to parallel batteries.

[dentaku]

I love these educational videos.

[smashIt]

Shh, don't tell the nerds who are busy trying to calculate it 

you also didn't tell them to use the absolute temperature in kelvin
I guess they will be really pissed

[Tomorokoshi]

In parallel configuration, if the failure mode of one of the capacitors is lower capacitance then we have a nice way of constructing some redundancy into the circuit.

What if the failure mode is a short?

Can we rely on the increase in individual reliability to outweigh the increased chance in failure now that more parts are in the circuit?

I wonder where the cross-over point is?

[Alexei.Polkhanov]

I find that one big chunk of information about electrolytic caps is missing - Bipolar Capacitors. Where and why you may use them etc. Maybe it is a good subject for another capacitor video?

[c4757p]

In parallel configuration, if the failure mode of one of the capacitors is lower capacitance then we have a nice way of constructing some redundancy into the circuit.

What if the failure mode is a short?

Aluminum electrolytics generally do not fail short unless the failure is caused by overvoltage or reverse polarity.

[bookaboo]

I love these educational videos.

Agreed