Question on ESR and Capacitor Selection.

[TheLaw]

I've been trying to figure out this ESR thing for a while now, but it's not making too much sense. I know that 0 ESR would be a perfect capacitor with no resistance whatsoever. But we all (most of us), know that all capacitors, ceramic or electrolytics, have resistance.

What I don't understand is how this affects a power supply or a circuit.

If you have less ESR, how does this affect a power supply?

I'm trying to build a stable linear power supply and although I know that certain components should be low ESR, and I know how to achieve that, I don't understand what ESR itself actually does.


Also,

Usually the main big filter cap before the regulator IC does NOT have to be low ESR. On PSUs with PFC circuits, the bulk capacitor does not have to be low ESR. Why is this?

I'd appreciate any answer. Big or small.

Thanks.

[Slakky]

Re: Question on ESR and Capacitor Selection.

You can type all of these terms into Wikipedia and get some answers. ESR is Equivalent Series Resistance, which is the same as a resistor in series with the capacitor. The exact value of the resistor changes with frequency though, and it's dominated by tan(delta) at low frequency and Z at high frequency. If you type "loss tangent" into Wikipedia, you can see an explanation of the relationship.

Usually when they say "Low ESR" caps, they mean "fit for high frequency use", which is basically any cap that bothers to list Z@100kHz in its data sheet. Most of the time general purpose caps are just a cost cutting or space saving measure, unless the circuit is built with a specific high ESR in mind like with LDO regulators.

The reason why you have to have low ESR caps on the secondary side of an SMPS is because the switching frequency noise they have to filter out is around 25~100kHz, so they have to perform well at that speed. The big primary cap doesn't have to be low ESR because it's just providing hold-up time at 50~60Hz. The same goes for the big cap before the regulator in LM317 and LM78xx/79xx circuits.

[Trinite]

Re: Question on ESR and Capacitor Selection.

From what I've understood, the higher the ESR, the longer a capacitor takes to charge and discharge. Which means a high ESR capacitor works fine for simply storing electricity, but would do poorly for filtering, since the slow charge/discharge limits its ripple capacity. The effect of ESR on ripple capacity would climb proportionally with current frequency, i.e. the ESR of a capacitor has a more pronounced effect the higher the frequency of the current you are filtering.

[TheLaw]

Re: Question on ESR and Capacitor Selection.

You can type all of these terms into Wikipedia and get some answers. ESR is Equivalent Series Resistance, which is the same as a resistor in series with the capacitor. The exact value of the resistor changes with frequency though, and it's dominated by tan(delta) at low frequency and Z at high frequency. If you type "loss tangent" into Wikipedia, you can see an explanation of the relationship.

Usually when they say "Low ESR" caps, they mean "fit for high frequency use", which is basically any cap that bothers to list Z@100kHz in its data sheet. Most of the time general purpose caps are just a cost cutting or space saving measure, unless the circuit is built with a specific high ESR in mind like with LDO regulators.

The reason why you have to have low ESR caps on the secondary side of an SMPS is because the switching frequency noise they have to filter out is around 25~100kHz, so they have to perform well at that speed. The big primary cap doesn't have to be low ESR because it's just providing hold-up time at 50~60Hz. The same goes for the big cap before the regulator in LM317 and LM78xx/79xx circuits.

But the capacitors after the LM78XX regulator shouldn't be low ESR? This is what has been confusing me. They technically aren't switch-mode converters but they do function similarly in theory.

And in regards to low drop out regulators, you want higher ESR?

Thanks for the detailed answer.

[TheLaw]

Re: Question on ESR and Capacitor Selection.

From what I've understood, the higher the ESR, the longer a capacitor takes to charge and discharge. Which means a high ESR capacitor works fine for simply storing electricity, but would do poorly for filtering, since the slow charge/discharge limits its ripple capacity. The effect of ESR on ripple capacity would climb proportionally with current frequency, i.e. the ESR of a capacitor has a more pronounced effect the higher the frequency of the current you are filtering.

Simply put! Thank you.

[Slakky]

Re: Question on ESR and Capacitor Selection.

But the capacitors after the LM78XX regulator shouldn't be low ESR? This is what has been confusing me. They technically aren't switch-mode converters but they do function similarly in theory.

No, the LM78xx don't care what you put after them. The caps after a switching regulator are part of a pi filter (Wikipedia that) which turns a high frequency rectified square wave into flat DC. Linear regulators have no analog to this.

And in regards to low drop out regulators, you want higher ESR?

LDOs want certain amounts of capacitance and ESR on their output. National has a pretty good writeup on the topic and how linear regs work in general if you're interested.

[PCBONEZ]

Re: Question on ESR and Capacitor Selection.

From what I've understood, the higher the ESR, the longer a capacitor takes to charge and discharge. Which means a high ESR capacitor works fine for simply storing electricity, but would do poorly for filtering, since the slow charge/discharge limits its ripple capacity. The effect of ESR on ripple capacity would climb proportionally with current frequency, i.e. the ESR of a capacitor has a more pronounced effect the higher the frequency of the current you are filtering.

No.
The charge and discharge you are thinking about is DC.
ESR -only- affects AC.
.

[PCBONEZ]

Re: Question on ESR and Capacitor Selection.

This might help.

[PCBONEZ]

Re: Question on ESR and Capacitor Selection.

Z² = ESR² + (XL - XC)²

There is a point where XL = XC.
At that point Z = ESR.
It's roughly around 100kHz for most caps we talk about here.
.

[Trinite]

Re: Question on ESR and Capacitor Selection.

No.
The charge and discharge you are thinking about is DC.
ESR -only- affects AC.
.

Don't most applications of electrolytic capacitors involve DC? For example, output filter caps deal with already rectified current, so it is going to be DC. Since ripple is simply variation in the voltage and not a cycling between positive and negative, how would ESR play a role in DC filtering then?

[PCBONEZ]

Re: Question on ESR and Capacitor Selection.

It removes the Ripple.
The Ripple is shunted to ground through the cap leaving the DC 'clean'.
.
Remember, AC goes THROUGH a cap.
DC does not. It charges it.
.
The Ripple's point of reference [zero] is the DC voltage.
That's why it is said to be 'floating' on the DC voltage.
Since DC is it's reference it IS cycling between positive and negative.
.

[PCBONEZ]

Re: Question on ESR and Capacitor Selection.

Case 1.
Cap has +5vdc on it and 100mV Ripple on it.
- What does the cap do with the Ripple?

Case 2.
Cap has 0vdc on it and 100mV Ripple on it.
- What does the cap do with the Ripple?

The cap's responce to Ripple doesn't change with the amount of DC.
That's why you disregard the DC when looking at the Ripple.
.

[PCBONEZ]

Re: Question on ESR and Capacitor Selection.

You have to stop thinking like the AC and DC are mixed.
Caps don't treat them like they are mixed.
That's why they are useful as filters.
.

[Trinite]

Re: Question on ESR and Capacitor Selection.

Ah, that seems to make sense. I was thinking in terms of reference to ground, rather than DC the voltage present.