Capacitor Voltage Question

[mike7877]

To get an idea for the power levels here, imagine this is the pre- section of a headphone amplifier, and the design has the op-amps supplied with 15V.

My understanding for the +20% capacitor voltage rating is to account for spikes from induction. So, say this op-amp's 15V supply is from a regulated SMPS, but the supply is quite far away so there's still a 220uf capacitor required a couple inches away. There's never going to be enough induction from the op-amp's changing power requirements during the audio signal for the voltage to move around more than a few 10s of mV (unless the trace was as thin as a hair...). So, can this standard advice be forgotten in this case? Or does the cap's rated voltage really need to be increased from +7% to + 67%?

I remember reading about electrolytic capacitors, and over voltage damage is supposedly caused when sparks jump the insulator, burning/compromising it (and possibly the metal itself).
So with no real possibility of going over 15V during operation... shouldn't a 16V capacitor be safe?

Or am I missing something?

I'd much rather have larger, lower ESR caps for the same price/size (if cap size is one of a design's constraints). A 25V 220uf cap is about the size/cost of a 16V 330uf cap, with the 16V cap likely having a lower ESR. At the very least, >1/3 of the size/weight/cost could be eliminated. Don't want to go on too much lol

 

[MaxwellsEq]

To get an idea for the power levels here, imagine this is the pre- section of a headphone amplifier, and the design has the op-amps supplied with 15V.

My understanding for the +20% capacitor voltage rating is to account for spikes from induction. So, say this op-amp's 15V supply is from a regulated SMPS, but the supply is quite far away so there's still a 220uf capacitor required a couple inches away. There's never going to be enough induction from the op-amp's changing power requirements during the audio signal for the voltage to move around more than a few 10s of mV (unless the trace was as thin as a hair...). So, can this standard advice be forgotten in this case? Or does the cap's rated voltage really need to be increased from +7% to + 67%?

I remember reading about electrolytic capacitors, and over voltage damage is supposedly caused when sparks jump the insulator, burning/compromising it (and possibly the metal itself).
So with no real possibility of going over 15V during operation... shouldn't a 16V capacitor be safe?

Or am I missing something?

I'd much rather have larger, lower ESR caps for the same price/size (if cap size is one of a design's constraints). A 25V 220uf cap is about the size/cost of a 16V 330uf cap, with the 16V cap likely having a lower ESR. At the very least, >1/3 of the size/weight/cost could be eliminated. Don't want to go on too much lol

It depends on the environment. It's generally accepted that you can get away with marginal over voltage if the capacitor is in a cool environment and there are few variations (such as in-rush spikes). If the capacitor is in a normal operating environment, e.g. a few degrees above ambient in a case, then you need more margin on the voltages to ensure a long working life.

 

[wwenze]

You mentioned your SMPS is 15V, so look at what voltage capacitor your SMPS is using.

 

[solderdude]

Lifespan of a capacitor is partly dependent on the voltage across it.
Using a voltage rating very close to the applied voltage is not the best idea.

For 15V best to use 20V or 25V rated capacitors and look at the temperature limits/lifespan.

On the outputs of switched mode PS best to check for lifespan (temp rating) and ESR.

For LPS it is best practice to ensure the voltage rating is one 'step' above the present voltage range.
With steps I mean 10V-16V-25V-35V-50V-65V-80V-100V etc.

For manufacturers that want to make a profit usually they skimp on quality of components, especially capacitors.

 

[DVDdoug]

but the supply is quite far away so there's still a 220uf capacitor required a couple inches away.

That shouldn't be necessary. Are you having some kind of problem?

...Some people with high-power car amps add super-capacitors so the voltage doesn't sag on loud parts but a headphone amp doesn't use much current so the voltage shouldn't drop significantly and if it does drop the headphone amp will probably work fine at 12V or less.

And if this only for the "pre section" it requires even less power/current than the headphone output stage.

I remember reading about electrolytic capacitors, and over voltage damage is supposedly caused when sparks jump the insulator, burning/compromising it
So with no real possibility of going over 15V during operation... shouldn't a 16V capacitor be safe?

16V should be OK. Especially since the power supply is regulated and should never hit 16V. BUT a higher-rated capacitor doesn't cost much more so I'd go with at least 25V for some extra safety margin.

And if I was adding a capacitor I'd probably go with 1000uF or more. But that's mostly based on my experience with "linear" power supplies and switching supplies can use smaller capacitors.

 

[Philbo King]

When I designed circuits for aircraft electronics (designed to last >50 years) the design guidelines required using caps with a minimum of 2x the peak voltage it could be exposed to. So it depends on whether you'd like the device to last a while or prefer saving fractions of a cent per component on parts. Engineering is always a series of compromises...

 

[AnalogSteph]

You can technically use 16V rated caps on a 15V supply, and potentially for quite a long time as well, but it still remains a gamble on capacitor quality and QC. As the parts age related to temperature, leakage current is up, and the risk of the dielectric layer degrading past the point of no return during extended downtime increases. Plus, the thinner the dielectric layer is to begin with (i.e. the lower the voltage rating), the easier it is for weak spots to become critically thin. I would also expect particularly small series to have less margin.

If you want to have a decent idea of what to expect, measure what leakage current settles down to long-term. When doing that for a range of voltages, you should find a near-ohmic relationship towards the lower end that'll start to increase markedly beyond a certain point. A bit like a zener diode. It's up to the manufacturer where on this curve the nominal voltage is chosen, also depending on sample variability of course.

Your best bet for seeing shorted electrolytics is in devices with small series caps of modest voltage headroom that made it to fairly high hours and were then stored unpowered for an extended period of time. For tantalums, voltage headroom and age may be the dominant factors (16V types on 12V rails seem to be going up in smoke quite predictably after 40 years).

 

[Sokel]

Rule of thumb for decades is double the expected voltage for trouble free operation.