Lacking fundamentals: Understanding electrolytics

[hirada]

Hello,

I am having a bit of an issue with understanding electrolytics. The standard polarized, not the less common bipolar ones.

What strikes me, when looking at a datasheet, the capacitance or ESR are often rated at a certain frequency. Like 120Hz or 100kHz. Latter maybe because of switching regulators.

But maybe we are not talking kHz AC, but pulsating DC (ripplevoltage)? Would make more sense after rectification as well.

However, asides the datasheet, I've seen quite a few designs, where electrolytics are used in the audio path to decouple DC. Probably because low frequencies require larger capactities. However (decoupled) audio is clearly AC or bipolar. How does that go together? Clearly you'll have the electrolytic reverse polarized for a moment. A short moment, but highly repetetive.


The final issue I do not really grasp is ripple current, also metioned in the datasheets. As opposed to ripple voltage, which occurs because the capacitor discharges, the current is determined by the load.
Sure, if the voltage drops, the current drops as well, assuming ohms law here. So I would understand a maximum current rating, but where comes the ripple into play here?

The ripple current seems to be different to the the current the capacitor is able to deliver, which is my main problem to understand.

Further, if there happens to be a regulator in between, then the load gets a constant voltage anyway and the drawn current is constant, just the regulator has to take care of the ripple voltage. So what is this about?

Quite obviously I am missing something very fundamental here and would appreciate some help to get back to the right path.

Thanks

[tggzzz]

You need to consider the AC and DC components independently. For example, if you have a 10Vdc with a superimposed 2Vac (peak-to-peak to make this example's arithmetic easier, but it will be RMS in datasheets), then the voltage varies from +9V to +11V.

The ripple current will be determined by the 2Vac, the frequency, the capacitor's ESR, and the load.

[Cerebus]

Hello,

I am having a bit of an issue with understanding electrolytics. The standard polarized, not the less common bipolar ones.

What strikes me, when looking at a datasheet, the capacitance or ESR are often rated at a certain frequency. Like 120Hz or 100kHz. Latter maybe because of switching regulators.

But maybe we are not talking kHz AC, but pulsating DC (ripplevoltage)? Would make more sense after rectification as well.

The rating at a spot frequency is done simply because the full picture, a graph of impedance and phase versus frequency, would take up too much space, be too onerous to produce, and would give too precise an impression for products that can have tolerances like ±20%.

You're quite right in your implication that the spot frequency chosen is often related to the intended usage: 100Hz for smoothing caps for linear supplies, 10 kHz or 100 kHz for SMPSs, etc. etc.

However, asides the datasheet, I've seen quite a few designs, where electrolytics are used in the audio path to decouple DC. Probably because low frequencies require larger capactities. However (decoupled) audio is clearly AC or bipolar. How does that go together? Clearly you'll have the electrolytic reverse polarized for a moment. A short moment, but highly repetetive.

Take note that there is usually a DC bias present as well as the AC signal, if the (positive) DC bias is greater overall than the (negative) excursion of the AC signal, then there will always be some (positive) DC bias. You can obviously swap (positive) and (negative) here and the same scenario applies. For cases where there isn't an adequate DC bias present, that's when the bipolar electrolytics get rolled out.

The final issue I do not really grasp is ripple current, also metioned in the datasheets. As opposed to ripple voltage, which occurs because the capacitor discharges, the current is determined by the load.
Sure, if the voltage drops, the current drops as well, assuming ohms law here. So I would understand a maximum current rating, but where comes the ripple into play here?

The ripple current seems to be different to the the current the capacitor is able to deliver, which is my main problem to understand.

Ripple current is current flowing into and out of the capacitor. Into the capacitor as it charges from the supply, out as it discharges toward the load.  There's a constant trade-off between the current contributed to the load by the supply and the current contributed to the load by the capacitor, mediated by the instantaneous voltage of the supply and the instantaneous charge on the capacitor. The ripple current rating matters because this current has a heating effect as it flows through the parasitic equivalent series resistance of the capacitor.

Further, if there happens to be a regulator in between, then the load gets a constant voltage anyway and the drawn current is constant, just the regulator has to take care of the ripple voltage. So what is this about?

A linear regulator has to 'burn off' any excess voltage between its input and output. The output will be at a constant voltage, the input voltage will vary with the (transformed and rectified) mains, the smoothing capacitor before the regulator tries to resist the change in voltage by charging and discharging, and that gives rise to the ripple current into and out of the smoothing capacitor. The ESR of the capacitor, along with the source impedance of the supply, turns this whole arrangement into a low pass filter and residual voltage ripple, rather than a flat DC signal, is the result of the passband/stopband of this filter. In practice, it's easier to think in terms of the current the capacitor has to supply during its discharge phase, rather than of the filter's characteristics, when actually designing a supply.

Quite obviously I am missing something very fundamental here and would appreciate some help to get back to the right path.

Thanks

[grifftech]

output from bridge rectifier attached