Ceramic Capacitor Problems mike at mdbventures.com 17 May 2014 (revised 24 Mar 2019) Visit us at http://www.MDBVentures.com - Great prices on great tubes! Here is some trivia to wow your friends and bore your enemies. Hey, it is always handy to have something to pull-out to impress people. That or bore them so they will quit bothering you... Ceramic capacitors - the good, the bad and the ugly. The problem with high uF ceramic capacitors. Ceramic capacitors are wonderful devices, they don't wear out, the modern ones provide good capacitance in small packages and they are very robust to voltage spikes. This doesn't come without problems though. Changes in Capacitance With Applied Voltages: There is a problem with ceramic capacitors. Specifically large value surface mount monolithic (MLCC) capacitors which use certain types of dielectric material. They are affected by applied DC bias voltage and to some extent applied AC voltage and temperature. The main problem is the DC voltage sensitivity. For some situations, it can seriously affect the capacitance. One of the worst on the market is the Y5U dilectric material. When used in the 100uF rated capacitor, applying a 5VDC voltage on the cap will cause the capacitance to be reduced to 10uF. Which kind of makes it pointless to use it for a bypass cap on a 5VDC power supply. Might as well use a 10uF cap instead which is much cheaper, or use a 22uF and get more actual working capacitance. How to test this effect: Place two 22uF 16V X5R MLCC capacitors in series. Attach an LCR meter across the two series capacitors. Using a 9V battery, apply a 8V to 9V charge to the capactors by connecting the battery across one of the capacitors. (The LCR meter will provide the DC path to charge up the other capacitor.) Read the capacitance on the meter. Using a wire, short out one of the capacitors (The DC path through the LCR meter will discharge the other capacitor). Read the capacitance on the LCR meter. You should see a 40% diffeence between the two readings. My test setup: ----------- | | | 20uf LCR | Meter +-------+---sw1---- + | | | | | 20uf sw2 8VDC | | | | -----------+-------+----1K---- - LCR Meter is a B&K Model 878 LCR meter. 20uF caps are 1210 X5R 16Volt ceramic MLCC SMT capacitors. sw1 and sw2 are momenatary switches (actually just loose wire leads). 8VDC is a used 9V battery. The 1K is a 1K ohm resistor to limit the battery current from potentally damaging the LCR meter. With SW1 open, momentarily close SW2 to discharge any DC voltage from the capacitors. The LCR meter reads 10uF. With SW2 open Momentarily close SW1 to place the 8VDC charge on the capacitors (each will have 8VDC place on it (the LCD meter is a conducting path). Open up sw1 The capacitors will now read 6uF on the LCR meter because of the 8VDC bias charge present on the capacitors. With sw1 open, close sw2 momentarily to remove the charge from the capacitors. The LCR meter will now once again read 10uF. X7R, X5R, X5S and Y5V are the most common dilectrics with this problem., X7R is the least sensitive of them to the problem. Which is why it is the primary material used for high value caps. NPO/C0G is another very common material, but it doesn't have the problem. It is very stable over temperature, and applied DC and/or AC voltage. Unfortunately it is not usable for large value capacitors (too big). The problem is a factor of the thickness of the dilectric material. The thinner the material, the greater the problem. Thus high value low voltage caps show the problem the most. To combat this effect, use a physically larger and higher voltage capacitor. The thicker the dielectric and the farther away from the maximum working voltage that the cap is operated at, the less of an effect the DC bias voltage has on the capacitance. As a side note, C0G ceramic capacitors do not have this problem, they are vary stable and do not change capacitance when a DC bias is applied to the capacitor. High value tantalum capcitors do not have this effect either, although they do change value slightly with a change in temperature. Most capacitor manufacturers provide charts of this effect on their web sites. Capacitance Shift With Temperature: There is also an effect on the capacitance relative to temperature, although it is not as drastic as the problem with the DC bias voltage issue (about +-15% worse case). There is also a different effect when AC voltage is applied to the capacitor. Typically a slight rise in capacitance of about 5% as the AC voltage changes from 0.5RMS to 2VRMS. Unfortunately this variability with the DC bias and AC voltage can introduce significant distortion into the circuit when the capacitor is a part of the signal path. This is a key reason to avoid using high value ceramic capacitors in audio circuits. In most cases they are ok for power supply bypass work with the limitations noted above. Because of these affects, MLCC ceramic capacitors that use these dielectric materials are not suitable for use in filters, and in some situations may not be suitable for use as a bypass capacitor. This can create a potential problem since these dielectric materials are used in all the high value ceramic capacitors. Some design compromises may be required as a result. In some cases, a smaller but higher voltage rated capacitor may actually have a higher operating capacitance in the real world application. Capacitance Fade Over Time: Another problem is that ceramic capacitors fade over time if they are not used. the ceramic crystal structure will relax making it less attractive to electrons. An X7R will lose about 10% capacitance over life. Y5V will lose about 30%. It is a bit more complicated than that, but you can use the noted values as a generalized rule. After 10000 hours of no use, an X7R will typically drop it's capacitance by 10% to 20%. Luckily the fall off seems to be logrithmic, meaning that it doesn't fall off much more than that. Interestingly they can be restored by putting them in an oven at 120C (250F) for 4 to 8 hours. Of course if they are installed in a circuit, the heat may damage other parts that cannot handle that much heat. Ceramic Capacitor Noise Problems: Ceramic capacitors are ceramic devices. The higher the dialetric constant, the greater the problem with piezoelectric effects. If the capacitor cannot be allowed to have this problem, restrict your design to NPO or Poly caps which don't show this problem. X7R and higher dialectrics will have this problem. In some applications electrolytic or tantalum capacitors can be used, however these type of capacitors are polarized and need a DC bias across the capacitor for best operation. An X7R and other high dialectric constant ceramic capacitors will vibrate if an AC voltage is applied. The higher the voltage, the greater the vibration. The capacitor will also produce voltage if it is physically vibrated. And finally, an X7R ceramic capacitor will change capacitance in all these situations as well as with a change in temperature and more significantly with a change in bias voltage. If you need to minimize the effects, do not use anything higher than X7R. For a rule of thumb calculation a 1uF cap with an X7R dialectric will exhibit a 3uV output when you thump a typical fiberglass circuit board (assuming no bias voltage). If the bias voltage is 10V, the output goes up to 100uV. Capacitors with higher dialectric material will generate even higher output voltages. A 10uF X7R capacitor with a 10V bias will generate about a 1mV spike when tapped with a probe. The actual output value is highly variable depending on too many factors to be able to specify what you will encounter. The output is dependant upon the stiffness of the board, the impedence of the circuit, and the type and construction of the capacitor as well as other physical and electrical factors. Final Notes: An advantage of ceramic capacitors is that they can handle peak voltages much better than other capacitors (up to 10x the rated working voltage. Tantalum capacitors on the other hand will self destruct with as little as 1.5x the rated working voltage. Another disadvantage of ceramic capacitors is that they are very fragile and are easy to break if mounted on a board that can flex. To minimize this, use small outline thick capacitors. This minimizes the exposure to flexing and provides added thickness to minimize flexing of the capacitor to what board flex is encountered. Some Example Measurements: Below is a table showing the measurments I made for various capacitors I had handy. You can see how the 100uF capacitor drops off to only 26% of its rated value when a DC bias voltage is applied to the cap. You can also see how changing the frequency has a significant effect on the capacitance. Cap = rated capacitance Volts = rated voltage Tol = rated tolerance Size = surface mount footprint Die = dielectric type Bias = applied DC bias voltage Freq = LCR meter measurement freq selection Value = value measured with B&K LCR meter (LCR measurement 150mVrms at 120Hz and 1KHz) Cap Volts Tol Size Die Bias Freq Value 100uF 6.3V 20% 1210 X5R 0V 1KHz 63uF 100uF 6.3V 20% 1210 X5R 8V 1KHz 26uF 100uF 6.3V 20% 1210 X5R 0V 120Hz 73uF 100uF 6.3V 20% 1210 X5R 8V 120Hz 28uF 22uF 16V 10% 1210 X5R 0V 1KHz 20uF 22uF 16V 10% 1210 X5R 8V 1KHz 14uF 22uF 16V 10% 1210 X5R 0V 120Hz 21uF 22uF 16V 10% 1210 X5R 8V 120Hz 11uF 10uF 16V 10% 1210 X7R 0V 1KHz 10.0uF 10uF 16V 10% 1210 X7R 8V 1KHz 9.0uF 10uF 16V 10% 1210 X7R 0V 120Hz 10.1uF 10uF 16V 10% 1210 X7R 8V 120Hz 9.1uF 2.2uF 25V 10% 0805 X7R 0V 1KHz 2.00uF 2.2uF 25V 10% 0805 X7R 8V 1KHz 1.55uF 2.2uF 25V 10% 0805 X7R 0V 120Hz 2.11uF 2.2uF 25V 10% 0805 X7R 8V 120Hz 1.80uF 1uF 50V 10% 1210 X7R 0V 1KHz 0.97uF 1uF 50V 10% 1210 X7R 8V 1KHz 0.95uF 1uF 50V 10% 1210 X7R 0V 120Hz 0.98uF 1uF 50V 10% 1210 X7R 8V 120Hz 0.96uF 1uF 50V 10% 1206 X7R 0V 1KHz 0.98uF 1uF 50V 10% 1206 X7R 8V 1KHz 0.93uF 1uF 50V 10% 1206 X7R 0V 120Hz 0.99uF 1uF 50V 10% 1206 X7R 8V 120Hz 0.94uF 0.33uF 50V 10% 1210 X7R 0V 1KHz 0.328uF 0.33uF 50V 10% 1210 X7R 8V 1KHz 0.328uF 0.33uF 50V 10% 1210 X7R 0V 120Hz 0.320uF 0.33uF 50V 10% 1210 X7R 8V 120Hz 0.320uF 0.33uF 50V 10% 0805 X7R 0V 1KHz 0.350uF 0.33uF 50V 10% 0805 X7R 8V 1KHz 0.328uF 0.33uF 50V 10% 0805 X7R 0V 120Hz 0.353uF 0.33uF 50V 10% 0805 X7R 8V 120Hz 0.333uF 0.1uF 50V 10% 0603 X7R 0V 1KHz 0.0978uF 0.1uF 50V 10% 0603 X7R 8V 1KHz 0.0933uF 0.1uF 50V 10% 0603 X7R 0V 120Hz 0.0990uF 0.1uF 50V 10% 0603 X7R 8V 120Hz 0.0960uF