It's no mystery that increasing capacitance and competitive pricing are helping multilayer ceramic capacitors (MLCCs) displace tantalum capacitors from areas they have long dominated. The onset of delivering decreased cost, higher performance and advanced developments in high-capacitance MLCCs provides an attractive alternative for designers who have traditionally used tantalum and who are now looking at the new high-capacitance MLCC range parts as a serious contender for new circuit designs.
Over the past 5 to 10 years, capacitance per unit volume for both MLCCs and solid tantalum capacitors has increased between 10 and 100 fold. As a result, the application areas for MLCCs have gradually encroached on those formerly dominated by tantalum capacitors and these, in turn, have extended into even higher capacitance application areas formerly occupied by, for example, electrolytic capacitors.
The two technologies now compete in the area of overlap between 1.0 microfarad and 100 microfarad. Many circuit designers who traditionally use tantalum capacitors might well find that their needs are now better served by the new generation of high-capacitance MLCCs. What's more, the ever-higher production levels of MLCCs (currently running at 600 billion annually compared with 25 billion for tantalum capacitors) coupled with new developments such as base-metal electrodes have greatly reduced their cost.
Comparing MLCCs and tantalums in smoothing applications
By 1990, Y5V type MLCC prices had reached the tantalum level; and as of 1995, the X7R types had reached the price levels of the 1.0-microfarad tantalum capacitors. These significant changes enabled MLCCs to directly compete with tantalum capacitors in many applications.
Because the requirements of electrical performance are highly application dependent, it is necessary to make a comparison per functional application area. Two of the major areas are smoothing and decoupling. One of the most common areas for smoothing is found in switched-mode power supplies (SMPS), which cover a very broad spectrum of output powers and ripple currents. Even so, design constraints such as maximum operating temperature, capacitance required and high-frequency switching requirements of modern SMPS designs often make the choice of capacitor technology clear. In areas where the choice is not clear, it is prudent to make a cost-performance analysis.
This can be done using what is known as a spice model to compare the smoothing performances of both types of capacitors. This makes use of a simulation circuit consisting of a full-wave rectifier bridge that is "smoothed" by either an MLCC or a tantalum capacitor. The capacitor models are RLC-models with the values properly chosen as a function of frequency, temperature and bias voltage. This approach is valid due to the circuit being simulated for a single frequency.
Based on a cost index, the smoothing performance is translated into a cost-performance diagram. Figure 1 is a typical example in which cost and performance are given relative to the tantalum 1-microfarad level. Good performance corresponds to low ripple index.
Fig.1 Cost-performance diagram at 40 degrees C, 100 kHz, 5 V bias for smoothing
For all three series, the performance improves with increasing capacitance. However, the figure clearly shows that within the MLCC range, a cheaper solution can be found for all performance levels. Moreover, for low and medium performance levels, a Y5V MLCC is the most cost-effective solution and for high performance levels (including the best level) a X7R MLCC is the best choice. The competitive position of X7R MLCCs becomes even better at higher frequency (for example, 1 MHz) due to the tantalum capacitor's reduced effective capacitance and high equivalent series resistance (ESR).
When looking at local decoupling, it is necessary to assess the amount of charge required by an IC to perform its stated change without causing excess voltage ripple (which can seriously degrade IC performance). This means the capacitor has to respond, in effect, as a low-impedance charge source. As with smoothing, performance improves with increasing capacitance.
Fig.2 Cost-performance diagram at 40 degrees C, 100 kHz, 5 V bias for decoupling. Results of assessments of 0.1 to 10 microfarad MLCCs and 1 to 100 microfarad tantalum capacitors in the 100 kHz to 1 MHz frequency range
In Figure 2, it is clear that at 100 kHz and 40 degrees C, a 1 microfarad MLCC offers a more cost effective solution than a 1 microfarad, 16 V tantalum capacitor. The 33 microfarad in case size C and 100 microfarad in case size D achieve even better performance levels but at considerably greater cost. What's more, since the size of these components is much greater, a designer may well consider using more than one MLCC in preference to a larger tantalum capacitor.
The results indicate that MLCCs in both X7R and Y5V dielectric offer cost-effective alternatives to equivalent tantalum capacitors over a wide frequency and temperature range.
In both technologies, developments continue towards further increase of capacitance per unit volume and smaller sizes. What's more, both technologies are introducing products with lower ESR levels. With MLCC technology, in particular, developments such as the new high-capacitance series, increased numbers of layers and greater effective capacitance are leading to drastically reduced ESR levels.
First indications lead us to the expectation that high-capacitance MLCCs with their intrinsically superior properties, smaller size and lower costs, will become more attractive to designers who formerly favored tantalums, forcing manufacturers of tantalum capacitors to strengthen their position in the high capacitance area (currently dominated by electrolytics).
Dave Ritchey is a senior application engineer for Yageo America Corp. He can be reached at firstname.lastname@example.org.