11
\$\begingroup\$

Is there any important reason why I can`t use high voltage mulilayer ceramic capacitors for low voltage scenarios? For example 50V or 100V rated MLCCs for 3.3V DC applications?

I have bad eyes, and have trouble soldering 1206 or smaller packages, and higher rated caps always have much bigger (and easier to solder) packages. I can't use THT/DIP components everywhere because of the PCB layout.

According to this note from Maxim, it seems that capacitance is always more stable for lower voltages. But what about ESR, leakage and so on? Could there be any issues?

Note: My typical scenarios are mostly decoupling caps - using OS CONs for SMPS power. Anyway I could have still trouble finding big SMT 15pF caps for a XTAL ... :(

\$\endgroup\$
  • 2
    \$\begingroup\$ Definitely yes! It will also make your life slightly easier since you don’t need to care about DC bias dependence on the capacitance. \$\endgroup\$ – winny Aug 5 '18 at 15:28
  • \$\begingroup\$ For decoupling you need to worry about the inductance of the package. Look for capacitors with the solder terminals on the long dimension \$\endgroup\$ – Elliot Alderson Aug 5 '18 at 15:47
  • \$\begingroup\$ @ElliotAlderson Thanks for mentioning importance of inductance! Almost forgot that multilayer ceramics also have it (unlike single layer ceramics). MLCC is still quite new technology for me :) \$\endgroup\$ – HeliTux Aug 5 '18 at 19:35
13
\$\begingroup\$

There is no danger using a capacitor rated for much higher voltage than you required, nor is there any performance loss for doing so. In fact quite the opposite.

Certainly if you have a DC bias across the capacitors (e.g. decoupling), you want the voltage rating to be much higher than your DC bias, else you need to choose a capacitor with higher capacitance rating than needed. This is because the capacitance of MLCCs drops off significantly as you apply a DC bias.

Most MLCCs don't get anywhere near their rated capacitance at their rated voltage. For an X5R dielectric, typically by the time you get to half the rated voltage, the capacitance has already dropped way below half the rated value. X7R dielectrics fare slightly better - you might expect to still retain 70% rated capacitance by the time you reach half the rated voltage, but even those will drop off.

Most manufacturers don't supply this data, however some including TDK and Murata do give these test results, and you can pretty much expect the same trends to apply to other manufacturers as the technology is practically the same.

As a simple example, this one is a bog standard 10uF 10V X7R MLCC in an 0805 package. At its rated 10V DC bias, the actual capacitance is only 4uF. With a 5V bias it fares slightly better, achieving 7.5uF. In fact you have to be at less than 2V bias (1/5th of the rated voltage) to actually achieve the 10uF capacitance rating. This is shown in the graph below.

Capacitance Vs Voltage Bias

This is why you typically want for X7R the rated voltage to be >2x the required DC voltage. For X5R, you probably want to be >4x the required DC voltage. The higher the better.

The only downside for going with a larger rating, is typically the size is required to be larger. However for low capacitance values (sub-100nF), this isn't too much of an issue, and you can easily find high voltage ratings in small packages. For very low capacitances (sub-1nF), you'd probably be hard pressed to find one with a low voltage rating anyway.

\$\endgroup\$
  • 2
    \$\begingroup\$ Thanks for very complex answer. Its now very clear to me how to use MLCCs. I always prefer biggest SMD parts as possilble (never have problems with PCB space), so it seems that MLCC doesnt have any downsides for me. \$\endgroup\$ – HeliTux Aug 5 '18 at 19:41
  • \$\begingroup\$ Probably with one exception: decoupling capacitors. \$\endgroup\$ – Simon Richter Aug 5 '18 at 21:26
  • 1
    \$\begingroup\$ You could mention that you don't need ">2x the required DC voltage" - you can just take the decreased capacity into account and take 30% higher nominal capacity. And in decoupling caps absolute values don't matter a lot. \$\endgroup\$ – asdfex Aug 6 '18 at 11:22
  • \$\begingroup\$ @asdfex See paragraph 2, I mentioned that. The final couple of paragraphs were more just a summary. \$\endgroup\$ – Tom Carpenter Aug 6 '18 at 13:58
2
\$\begingroup\$

It would surprise the heck out of me if you could find low value capacitors rated only for low voltage. Typical SMD ceramic capacitors are rated at like 50V.

High value ceramic capacitors are more often available with lower rated voltage, but that is because they are voltage sensitive - a particular part may be capable of operating at a higher voltage, but it won't reach the specified capacitance above the rated voltage.

Larger parts have higher inductance, so they won't be as good as smaller parts for high freqencies.

I see on a quick check that even small valued parts (12pF) are available in 1206 size.

You can use the larger sized parts, but if you get into high frequency stuff, or stuff where the self resonance frequency of the part matters then you will need to take extra care if you stick with the larger parts.

For decoupling uses on typical hobby stuff it shouldn't matter.

\$\endgroup\$
  • \$\begingroup\$ "but it won't reach the specified capacitance above the rated voltage" - Most MLCCs don't get anywhere near their rated capacitance at their rated voltage under DC bias. Typically by the time you get to half the rated voltage, for an X5R, the capacitance has already dropped below half the rated value. For X7R, you might get 70% rated capacitance by the time you reach half the rated voltage. \$\endgroup\$ – Tom Carpenter Aug 5 '18 at 16:25
  • \$\begingroup\$ Just as an example, TDK give lots of data on their caps. This one is a bog standard 10uF, 10V, X7R cap. At 10V DC bias it is down to 4uF, at 5V it is 7.5uF. That's why you typically want the rated voltage to be over 2x the required DC voltage. \$\endgroup\$ – Tom Carpenter Aug 5 '18 at 16:29
  • 4
    \$\begingroup\$ RE "Typical SMD ceramic capacitors are rated at like 50 V". Not if you move to smaller sizes (0402 and lower) and values above 1 nF. Right now I'm doing part selection and working hard to find 10 V ratings rather than 6.3 or even 4.something volts. \$\endgroup\$ – The Photon Aug 5 '18 at 17:50
  • 3
    \$\begingroup\$ The limit on operating voltage for low voltage MLCC caps is actually from capacitance decrease, not breakdown voltage. Analogous to saturation current being limiting for inductor peak operating current. As Photon says, 6.3 and even lower are common. \$\endgroup\$ – Spehro Pefhany Aug 5 '18 at 20:28

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.