Multilayer ceramic chip capacitors (MLCCs) are reaching the market with ever-greater capacitance values without increasing in size. Unpleasant side effect: It is becoming more and more difficult to measure the correct capacitance value. The values determined by users are frequently too low, which regularly results in significant extra expenditure. However, this situation could easily be avoided if certain aspects were addressed.
In terms of quality assurance, users tend to measure more parameters more frequently, including the capacitance value of MLCCs; not just to check that the correct component is installed but also to ensure specified tolerances are met. Use of the popular HiCap MLCCs with C values of 1.0µF and higher and the associated very low ESR / impedance values, in particular, often result in incorrect measurements.
The reason for this is basically the characteristics of the applied materials, the base ceramics. These are barium titanates, which are in fact ferromagnet materials that ensure C value drifting as a direct result of influencing variables such as temperature, DC bias, AC voltage, and aging, which becomes even more profound the higher the capacitance value.
Some examples of this MLCC behavior are shown in graphic 1.
Correct measuring voltage
A correct measurement initially demands appropriate zero-point calibration, namely as "open compensation" and "short compensation". It is then particularly important to check the effective measuring voltage, as an insufficient voltage is the main source of erroneously determined C values.
Apart from a few exceptions, the nominal measuring conditions for all type 2 / class 2 ceramics (also known as X7R, X5R, X7S, ...Y5V, etc.) are defined in table 1 at a room or ambient temperature of +25°C:
The very low impedance of the capacitors leads to the measuring circuits of the devices being loaded to such a degree that the measuring voltage often drops. The result is C values that are well below the actual values. Specially design measuring devices featuring ALC (AutoLevelControl), e.g. Agilent 4284A, 4278A, 4268A, offer a remedy to this problem. The ALC ensures the measuring voltage at the capacitor achieves the actually preset measuring voltage. This is easy to check when using devices that indicate the voltage level on a display, otherwise a separate voltmeter is a good alternative.
If the capacitors are measured as part of an incoming inspection process, i.e. measured before actually being used, the phenomenon of aging also comes into play. This reduction of capacitance over time, so-called aging, will be greater or less depending on the age of components. A significant loss in capacitance value is experienced with HiCaps which can exhibit an aging rate of over 5% per decade hour.
It is imperative to compensate any deviations caused by a difference in component ages, especially when you need to be able to reference, i.e. compare, C values. Since aging is a reversible process, the MLCCs need to be re-heated. In other words, they are subjected to the industry's standard de-aging procedure. This involves heating the capacitor to a temperature above the Curie point, typically +150°C, for about one hour. In order to achieve the required level of stabilization for measurements, the components should be allowed to rest for 24 hours in advance.
De-aging procedure: Re-heating the capacitor to a temperature above the Curie point. (Source: Samsung)
This procedure is increasingly recommended by component manufacturers in their data sheets and catalogs.
Support from component manufacturers
In very rare cases, the determined C value is still too low despite all these aspects being considered. An analysis carried out by the respective capacitor manufacturer will then shed light on this issue. Certain details are necessary for this purpose: For instance, the measured value determined by the client for the component, as well as the preset and checked measuring signal and measuring device used, whether the ALC was active, and the type of de-aging that might have been achieved. If the batch number of the components is also provided, preferably as a role copy in order to view all the labels, the manufacturer can also check the date of manufacture and reference samples and use them for comparison. This is possible because manufacturers measure the C value of each and every component - usually twice - during production. Moreover, they carry out random sampling of components that have already been prepared for shipping.
MLCCs are available at www.rutronik24.com.