2
\$\begingroup\$

I am trying to understand what type of output capacitor is a better option.

Suppose I require a 10uF output capacitor at the output of the buck converter, which one should I pick - aluminum electrolytic , ceramic or tantalum? (All SMD.)

I read that the aluminum has good ESR which would help in the stability of the DC-DC control loop, but it is big in size. So, size is the disadvantage over here.

Ceramic capacitors have 100 times smaller ESR when compared to the aluminum one, and they come in very small sizes. So, this is an advantage when size is a bigger constraint.

I couldn't find any big advantages or disadvantages of tantalum. Can someone tell me the benefits and disadvantages of using tantalum?

Please also tell me which cap would be ideal for the DC-DC converter output - aluminum, ceramic or tantalum?

\$\endgroup\$
7
  • \$\begingroup\$ Voting to close - this is a hypothetical question and there is no definite/ideal answer because there are many undefined design parameters (converter frequency, fixed- or variable-frequency, output current, output inductor size and load characteristics, converter size, fault tolerance, MTBF) which could make any of the options the best choice or the absolute worst choice. \$\endgroup\$ Jun 24 at 14:06
  • \$\begingroup\$ Hi-k ceramics have pretty well supplanted tantalum caps these days. Tants have better specs than electrolytics but the downside is tantalum is declared a conflict mineral and tants like to go up viciously in flames if you are not careful. Electros are cheap. Hi-k ceramics have less than ideal properties like dc bias derating and piezo electric properties. Otherwise they are small and have low ESR compared with tants and electros. Sometimes the low ESR is a curse and causes ringing. \$\endgroup\$
    – Kartman
    Jun 24 at 14:07
  • \$\begingroup\$ What is the meaning of Hi-k? \$\endgroup\$
    – Newbie
    Jun 24 at 14:11
  • \$\begingroup\$ It depends entirely on the chip's data sheet. \$\endgroup\$
    – Andy aka
    Jun 24 at 14:54
  • 1
    \$\begingroup\$ @AdamLawrence I disagree. I wouldn't close this question. Its not hypothetical and it may not have a precise numerical answer, but the answer to this question is definitely not speculation based. \$\endgroup\$ Jun 25 at 6:52
6
\$\begingroup\$

I would avoid tantalum unless you have no other option; one of their more common failure modes is to burst into flames.

I don't know where you read that aluminum electrolytics have good ESR compared to ceramic, but they just flat-out don't; they tend to have some of the highest ESR of any capacitor type. Even low-ESR aluminum caps can be multiple orders of magnitude higher ESR than comparable MLCCs or film capacitors.

For an application like this, I would use an MLCC since the exact capacitance value doesn't matter, and it isn't too large; 10 μF MLCCs are pretty cheap. If you needed much more than that, though, aluminum electrolytic would be the way to go.

I'd also like to note that you overlooked one very important type of capacitor: film capacitors. Film caps tend to be available in larger capacitances than ceramic, are more stable with temperature and applied voltage than high-κ ceramics, are available in higher voltage and temperature ratings than MLCCs or electrolytics, and have comparably low ESR to ceramics. It wouldn't be worth the price to use for just an output filter capacitor, but whenever you need something stable and high-performance, film is often the best choice.

\$\endgroup\$
8
  • \$\begingroup\$ Thank you for the answer. Could you tell me how Tant capacitors are prone to bursting in flames and what is the meaning of High-K ? \$\endgroup\$
    – Newbie
    Jun 24 at 14:13
  • 1
    \$\begingroup\$ Solid tantalums (MnO2 types) are damaged internally simply by going through the standard reflow process which can cause internal shorts from the tantalum slug to the MnO2 layer (the Ta2O5 dielectric gets cracked due to major differences between packaging and Tantalum CTE above Tg). Case sizes D and higher are most at risk. I have seen them fail when on a low impedance source at much less than the nominal rated voltage in a spectacular pyrotechnic fashion. \$\endgroup\$ Jun 24 at 14:23
  • 1
    \$\begingroup\$ @Newbie I don't know the exact mechanism, but tantalum caps are notorious for causing lots of damage to a circuit if they fail. Overvoltage or excessive temperature can cause failure. High-κ (that's the Greek letter kappa, not K) ceramics are type-II (or type-III, but you don't see that anymore) dielectrics, basically all the ones that aren't C0G/NP0. They trade off performance and stability (and a lot of stability!) for much higher capacitance due to their high dielectric constant κ. Their capacitance varies significantly with applied voltage and with temperature. \$\endgroup\$
    – Hearth
    Jun 24 at 14:25
  • \$\begingroup\$ Tantalums don't have self healing, so if an internal short develops, it will stay shorted. If it was a real low resistance short circuit, that would draw enough current to convince the DC-DC to go into foldback/hiccup current limiting, but... that's not usually what happens, since it's just a tiny short it draws enough current to make a lot of heat and fireworks, but not enough to bring down the power supply voltage. So you get a tiny volcano on your board. \$\endgroup\$
    – bobflux
    Jun 25 at 8:02
  • \$\begingroup\$ @bobflux Actually, tantalums do self heal if the fault current is limited. If you put a resistor (at least several ohms) in series with a tantalum cap, the local overheating of a damaged region will cause the material to turn into an insulator, stopping the leak. The bad thing about tantalums is that if the current is not limited, they undergo a runaway process where a small leak quickly propagates through the material, destroying the cap. And one can't typically afford making an output cap "really high ESR" by putting a resistor in series. \$\endgroup\$
    – TooTea
    Jun 25 at 10:24
5
\$\begingroup\$

The concept of a load cap to a DCDC converter has two factors. Forward loss + bulk storage and effects on feedback error correction. The tradeoffs depend on load range , overshoot and ripple tolerances. The cap choices affect ESR vs frequency, losses vs size, quality and ESR*C time constant.

The ESR*C=tau determines the slew rate of voltage by the current it can handle such that the output is dV/dt= I * (1/C + ESR) = (1+ESR * C)/C. Low ESR e-caps have a tau=<10 us while ceramic may be << 1us but due to dielectric constant k will be smaller in C for the same size package so both ceramic and e-caps are usually chosen unless you choose many ceramic caps in parallel.

The other concept is that the bandwidth of voltage feedback for error correction is also determined by tau and this can introduce a phase shift which reduces phase margin in the loop, so partial derivative or slope compensation is required to restore stability in the loop.

The 3rd concept is the efficiency of storage and losses in ESR during loads which draw from the cap as a temporary storage device while the DCDC converter tries to charge up the caps and supply the load at the same time, so that the voltage error is minimal when then is a step load added or removed and the energy in the cap is sufficient to Buffer the voltage with a current swing so that the DCDC driver does not under or over charge the capacitor due to latency or under/over estimating the demand current and result in more error from amplifying the + or - error in bigger than necessary correction in the opposite direction.

So ultimately the feedback loop gain Kd, Kp must be examined and provide some tradeoff of output voltage ripple in order to remain stable if the output caps are “inside the control loop” . Isolating the caps with a ferrite bead outside the loop must also examine the Q of this filter for no load when Q is the highest with any LCR series filter.

How to achieve all these tradeoffs is somewhat complex and different for every design, but understanding these tradeoffs is the beginning towards intelligent DCDC designs by examining then driving impedance to reactive components and the feedback with a shorted output or a near shorted output due to the ESR of the caps.

Review answers by Basso for details ( verbal kint )

  • Tantalum are less favourable but some more expensive types can meet the requirements.

  • alum caps rated for ripple current @ 120 Hz are unlikely to be low ESR types at 100kHz. These have tau>100 us typically and used as bulk line rectifier caps.

\$\endgroup\$
2
  • \$\begingroup\$ Thank you for the answer. Let me understand and come back \$\endgroup\$
    – Newbie
    Jun 24 at 14:14
  • \$\begingroup\$ We used to rely on the output zero from the output capacitors (in current mode converters) for phase boost. With the ESR of surface mount ceramics being in the milliohm range, we can't do that any more. The solution is to create an output zero with a capacitor across the top resistor in the feedback path (positive regulators). Often known as a feedforward resistor. \$\endgroup\$ Jun 25 at 11:34
1
\$\begingroup\$

As others stated, main design considerations are not defined in your question.

  • What is the lifetime of the circuit (aluminum electrolyte caps are being used a lot as a "lifetime limiting component", ceramics and tantalum have much higher lifetime).
  • What temperatures do you expect for your device (aluminium electrolyte do not like high temperatures, ceramics can have a massive temperature influence, so tantalum can come in the picture when high temp range is needed).
  • Is it possible to mitigate the risks of tantalum (when in doubt, don't even consider them, fireworks are nice but you don't want it unexpected).
  • Is cost or size a factor? Higher quality needed? Vibration/shock requirements? Lots of other factors that can push your decision.
\$\endgroup\$
0
\$\begingroup\$

Ceramic and electrolytic caps have their advantages and disadvantages.

Electrolytic caps have a higher ESR than ceramic caps, but you get electrolytic caps with a ton of capacitance. They are good for bulk storage.

Ceramic capacitors have a very low ESR but don't come with a large capacitance (like 100uF, for example). Due to their low ESR, they are good for output noise suppression.

Designers use both ceramic and electrolytic capacitors in parallel to get the benefits of both types of capacitors. Electrolytic capacitors come with a large capacitance and take care of the bulk storage while ceramic capacitors are used to suppress noise at the output.

If all you need is 10uF of output capacitance, just use ceramic alone. You can use tantalum capacitors too, but there are some safety concerns. If you need a large amount of output capacitance (like 50uF, 100uF or more), use electrolytic and ceramic capacitors in parallel.

\$\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.