28
\$\begingroup\$

I am building a power supply circuit, and the switching regulator (L4963) calls for a low-ESR output capacitor. The capacitor in question is C3 of the evaluation board circuit.

Figure 26 from the STMicro L4963 datasheet

What does "low" mean? How low?

Also, how do I find or calculate the ESR for a capacitor whose datasheet does not have a parameter called ESR?

\$\endgroup\$
  • 1
    \$\begingroup\$ The link to the datasheet of the IC you're using is broken. \$\endgroup\$ – Connor Wolf Jun 5 '14 at 8:57
  • \$\begingroup\$ The switching regulator in question appears to be a "L4963[D]", by exactly one Google result on the exact link. Current link is here. \$\endgroup\$ – user2943160 Jun 30 '16 at 0:42
  • \$\begingroup\$ Ceramic caps have much lower ESRs than even "low ESR" aluminum caps, so their datasheets don't even bother listing it. \$\endgroup\$ – TimH Jun 30 '16 at 5:12
18
\$\begingroup\$

If a datasheet just says 'low ESR' without specifying a value, you are usually fine with any style of capacitor with a relatively low ESR. All this really means that you should avoid cheap unrated aluminum electrolytic capacitors, since their ESR is terribly high and can be several ohms.

In this case it wants a 'low-ESR' capacitor for the 1000 µF output capacitor. I don't think I've ever seen a ceramic 1000 µF capacitor and a 1000 µF tantalum capacitorwould probably cost US$50, so you are going to have to track down a low-ESR aluminum capacitor for this application. The output ripple will decrease linearly with the ESR of the capacitor, so lower is better up to whatever price you want to pay.

As an aside, that is a ridiculously high required output capacitance for a switching regulator in that voltage range. You may want to take a look around for a regulator that meets your needs, but is stable without such a requirement. Don't get me wrong, usually the more capacitance the better, but 1000 µf is really high for a 1.5 A power supply.

\$\endgroup\$
  • 8
    \$\begingroup\$ I note that on the evaluation board, they use a 40V 1000uF electrolytic (claiming 'low ESR') and parallel it with a 1uF low ESR film cap. Putting caps in parallel will result in the resistance being smaller than the smallest resistance and the capacitance being higher than the highest capacitance. I agree that 1,000uF is ridiculous. \$\endgroup\$ – Kevin Vermeer Oct 27 '10 at 14:29
  • 5
    \$\begingroup\$ the caps in parallel will also do the job of absorbing higher frequency transients. \$\endgroup\$ – Kortuk Oct 27 '10 at 16:46
8
\$\begingroup\$

It's 'equivalent series resistance', and is somewhat frequency dependent. Basically it's the unavoidable ordinary resistance that comes along with the capacitor.

Lower ESR means that the capacitor is more like an ideal circuit element. Resistance just dissipates power, which results in heat, which is generally no good for capacitors, especially electrolytics.

Just speculating now - On that data sheet you linked, the salient parameter looks like 'tangent of loss angle'. If one assumes that 'loss angle' is the angle away from a purely capacitive reactance, then the tangent of that angle would be the series resistance divided by the capacitive reactance, in which case this number being low would imply ESR to be low.

\$\endgroup\$
  • 1
    \$\begingroup\$ I believe a typical cap actually behaves as a bunch of caps in parallel, each with a different ESR. At any given frequency, the "partial caps" with higher ESR won't charge and discharge as much as those with smaller ESR, and they'll thus play less of a role in circuit behavior; this effect will be more pronounced at higher frequencies. In axial electrolytics, the ESR of the "partial caps" would be pretty similar; in radial caps, the ESR of the portions nearer the leads will be much lower than that those further away. \$\endgroup\$ – supercat May 18 '11 at 19:39
  • 1
    \$\begingroup\$ For many power-supply applications, a cap with a non-uniform ESR may be preferable to one with a uniform ESR. For things like audio coupling and filtering applications, the reverse may be true. \$\endgroup\$ – supercat May 18 '11 at 19:40
6
\$\begingroup\$

As far as numbers would go: A low ESR cap for a switcher should usually have not more than a few 10 mOhm.

\$\endgroup\$
6
\$\begingroup\$

How "low" depends on the efficiency and reliability you are trying to achieve. For low ESR capacitors the manufacturer will supply the values.

Search for Nichicon low ESR capacitors and you will find parts that have a low ESR. The VR series is not a low series resistance capacitor. The PM Series is and the ESR is specified in the datasheet. Nichicon (which makes excellent capacitors) may have some newer series.

The ESR is critical to the life of the capacitor since as ESR increases the temperature of the capacitor will increase which will decrease its life.

\$\endgroup\$
5
\$\begingroup\$

ESR is frequency and temperature dependent. Most datasheets will list the ESR for a number of discrete frequencies, which may or may not be your switching frequency.

If you have an LCR meter you can connect up the capacitor and set the frequency and measure the ESR.

This is important for calculating the thermal loss inside your capacitor. It comes back to Ohm's law; for your switching frequency, there is an ESR, which is R, and you have current flowing into and out of the capacitor, which is I. Square it and multiply by R and you have the power loss inside the capacitor. The capacitor datasheet will also state its thermal resistance, so you can estimate the temperature you will be running your capacitor at. Select a suitable temperature rating for your application.

\$\endgroup\$
  • 5
    \$\begingroup\$ +1 because it is true that the best way to find out about the ESR of a certain capacitor will be a measurement with an LCR meter. The stupid thing is that only very expensive ESR meters work well with all types of electrolytic capacitors. Datasheet values are very (!) rough and tend to list only the ESR values that you get when all extreme circumstances play against the capacitor. Same is true for obtaining the ESR from tan(delta), at a certain frequency. You will usually get a value much too high. The best parameter from the datasheet is I_rms => The higher Irms, the lower the ESR. \$\endgroup\$ – zebonaut Oct 27 '10 at 10:36
  • \$\begingroup\$ good comment about the Irms \$\endgroup\$ – smashtastic Oct 27 '10 at 12:01
  • \$\begingroup\$ Presuming the ESR meter will be able to measure the low-value resistance values with any accuracy. \$\endgroup\$ – Peter Mortensen Jun 5 '14 at 5:23
4
\$\begingroup\$

In this case I think the low ESR is needed to get a low ripple on the output voltage.

The current through the inductor will have some ripple, and assuming a constant output current the capacitor will have to absorb or supply the difference to the output. Multiply this ripple current with the ESR and you get the ripple voltage.

You can measure ESR by charging and discharging the capacitor with a relatively high current switched by a function generator, and then measure the ripple voltage with an oscilloscope.

I've seen 170 mOhm in practice for a low ESR SMD electrolytic capacitor. If I remember correctly the voltage difference was 0.5V so the current ripple must have been 3A (limited by the power supply).

\$\endgroup\$
-2
\$\begingroup\$

ESR is the variation of resistance at different loads (current) and forced charge (current/frequency curves). thus with two or three (better) charge/discharge cycles and the angle-delta from their variations in capacitor "resistance" you can easily obtain the relative internal resistance. thus determine a coefficient called ESR which is the Z resistance at a multiple of charge/discharge speed cycles...

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.