1uf polarized vs non-polarized

[DavenPaget]

Electrolytic capacitors degrade by
(1) time when not polarized by a voltage; the oxide insulator dissolves over a decade more or less into the electrolyte
(2) heat; heat speeds up the chemical degradation of the oxide layer in (1) by about 2x for every 10C rise, as well as changing the conductive electrolyte

A polarized cap kept at the rated voltage by DC will last semi-forever. Left with no polarizing voltage, one to a few decades. Reverse polarized, minutes down to milliseconds.

Power filter caps have significant AC currents running through them. This AC current through the equivalent series resistance (ESR) heats the insides. This is why capacitors are rated for maximum ripple current - the internal self heating shortens their lives.

External heat sources can do much the same thing.

I was attempting to find a explanation about ESR heating and the maximum ripple current and that every 10C rise cuts their life by half .

[R.G.]

It's not true in every case, but in general, chemical activity speeds up by about 2x for every 10C rise in temp as a rule of thumb. Since the electrolyte decay is a chemical process, making the electrolyte hotter is much the same as aging it faster. This same principle is used in accelerated aging tests to figure out the useful life of parts.

The temperature aging effect doesn't much care where the temperature comes from. You can store them at high temperatures or work them at high temperatures and get the same thing. On the other hand, leaving them polarized by a voltage helps repair any damages to the insulating layer by regrowing the oxide (while the electrolyte continues to degrade). So leaving them polarized at low temps would be the way to preserve electrolytic caps for the longest shelf life. No one does that.

Internal heating happens any time charge moves in a resistive medium, which the conductive parts of capacitors are. Each time you push charge in by putting a voltage across the cap, some ohmic heating happens. Likewise when you drain the charge out. If it's an AC current, charge always going in and out like in filter caps, the best way to describe the heating effect is to do the math and calculate a root-mean-square current which is numerically equivalent to a DC current in terms of heating effect. You can then just multiply the square of the RMS ripple current times the ESR and figure out the internal heating.

[fpaul]

I had to replace 6 blown e-capacitors in an Nvidia video card and two in an emu1820 recording interface someone gave me.  They were all visibly blown. I don't think any of them were for the power supply.  I think they were both made around the time of the dreaded Capacitor Plague.

[thedefog]

I had to replace 6 blown e-capacitors in an Nvidia video card and two in an emu1820 recording interface someone gave me.  They were all visibly blown. I don't think any of them were for the power supply.  I think they were both made around the time of the dreaded Capacitor Plague.

Yup, I replaced dozens on Dell motherboards from that era.

[Paul Marossy]

Yeah, computers are a different story than audio. Whole different ball game...

[R.G.]

Actually, it's not. The differences are that there are *big* currents moving between the +V and ground wires (planes, actually) at the switching frequency. What happens is that switching tries to pull the grounds up and the power supply down driving the logic input loads. Those are primarily capacitive in CMOS logic (which is what all modern logic uses) and so there are *big* spikes of current at the clock frequency as everything switches at the same instant - measured in femtoseconds.

The currents can be truly phenomenal. "Ground bounce" is one old term for the "ground" rising by hundreds of millivolts or even volts on this transient. The local capacitors source/sink the currents for the hundreds of femtoseconds or few nanoseconds of the transient. And the ESR of the caps dissipate heat as this happens.

But the concept is the same, just a different venue.

[DavenPaget]

Actually, it's not. The differences are that there are *big* currents moving between the +V and ground wires (planes, actually) at the switching frequency. What happens is that switching tries to pull the grounds up and the power supply down driving the logic input loads. Those are primarily capacitive in CMOS logic (which is what all modern logic uses) and so there are *big* spikes of current at the clock frequency as everything switches at the same instant - measured in femtoseconds.

The currents can be truly phenomenal. "Ground bounce" is one old term for the "ground" rising by hundreds of millivolts or even volts on this transient. The local capacitors source/sink the currents for the hundreds of femtoseconds or few nanoseconds of the transient. And the ESR of the caps dissipate heat as this happens.

But the concept is the same, just a different venue.

+74

[boogietone]

First thing to do when there is whine is to throw a battery on it in place of the wall wart. I have more than once gone crazy looking for the source everywhere else before noticing that I have plugged into a particular PSA brick. Don't know why I even keep the thing. It bleeds through to everything.