Fullscreen HTML5Med Res (??MB)HTML4

All About Capacitors Capacitors are probably the most misunderstood of the passive components, due to the many different types available, their many parameters and greatly varying performance. This article should give you an understanding of the most common types, how they differ, and how to choose the right ones for your design. By Nicholas Vinen C apacitors come in all shapes and sizes. Some are much smaller than a grain of rice, while others are huge and used in banks to launch aircraft weighing many tonnes into the air! Because there are so many different types, it can be very confusing trying to choose one. Even if you know what capacitance and voltage rating you need, there could be hundreds or even thousands of matching parts. Some of those might not work at all in your circuit, while others might work but not very well, and some will be very expensive. You need to narrow the choice down to just a handful and then pick one. We have tried to break the following descriptions into digestible sections, despite their complexity. If you find yourself overwhelmed, give yourself time to digest what you have read so far, then read the rest later. Capacitor dielectrics Fundamentally, a capacitor is just two conductors (originally flat plates) separated by an insulator (the “dielectric”). But because the area of the plates required for any significant capaci72 Silicon Chip tance is quite large, modern capacitors are typically arranged as many layers of smaller conductors and insulators connected in parallel, allowing for a more compact package. In some cases, the ‘plates’ are not even flat but instead are spiral coils, or 3D structures such as the etched surface of a metal foil or granular materials. Etched or granular materials have a much higher capacitance per volume, as capacitance is proportional to surface area and inversely proportional to the distance between the plates. This creates a tradeoff; thinner dielectrics give more capacitance, but have a lower breakdown voltage, so the maximum voltage applied to the capacitor must be kept lower. This is the main reason that a capacitor with a higher voltage rating, but the same capacitance, tends to be physically larger; its dielectric layer(s) need to be thicker. The type of insulating (dielectric) material used has a strong effect on capacitor behaviour, and for this reason, capacitors are mostly categorised by the dielectric type. Different dielecAustralia’s electronics magazine tric types have their own trade-offs in terms of capacitance, voltage ratings, linearity, current handling and more. Some widely used dielectric materials for capacitors are: • Ceramics (typically metal oxides) • Metal oxide layers (in electrolytic capacitors) • Plastic films • Mica • Paper • A Helmholtz plane of solvent molecules (as in ‘double layer’ super/ultracapacitors) The most common types of capacitors in use today are ceramic and electrolytic, followed by plastic film types. These three types of capacitors have important sub-categories which strongly affect their behaviour. One property of all dielectric materials is the dielectric constant (“K”). The larger this number, the higher the capacitance for a similarly constructed device. K can vary with temperature, voltage, age and other properties. While high K values make for greater capacitances in a small volume, there are significant penalties in other areas, as we describe below. siliconchip.com.au Ceramic capacitors Fig.1: the range of capacitances and voltages available in 3.2 x 2.5mm SMD ceramic capacitors today. Both larger and smaller sizes are available, extending the range of values down to 0.1pF (1.6 x 0.8mm) and up to 470µF (4.5 x 3.2mm). Note how some types of ceramic dielectric are available to higher working voltages, and others to a higher maximum capacitance. (original source: Wikipedia) If you look at the PCB of just about any modern electronic device, you will find it covered in ceramic capacitors. They are cheap, reliable, perform very well and are available in a wide range of capacitances and voltage ratings. Because modern ceramic capacitors are fabricated in bulk, they can have anywhere from one to many thousands of layers. This gives them a wide capacitance range, from fractions of a picofarad up to hundreds of microfarads, in a small package – see Figs.1-3. Ceramic capacitors are typically robust and long-lasting, and are not polarised (they can handle negative or positive voltages). Ceramic capacitors are available with voltage ratings from just a few volts up to several kilovolts. Ceramic capacitors with voltage ratings above 500V tend to use different types of ceramic to those below 500V, and have slightly different properties. The most common ceramics used are based on titanium dioxide (TiO2) or barium titanate (BaTiO3) with additives to tweak their properties. As there are so many different possible combinations, they are arranged in various categories based on their performance. The categories are based on the initial tolerance of the capacitor (ie, the variation of real samples from the rated value), how the capacitance changes with temperature (the temperature coefficient) and how it changes with applied voltage (the voltage coefficient). The most common type codes are NP0 or C0G (different names for the same category), JB, SL0, X5R, X5S, X6S, X7R, X7S, X8L, Y5V and Z5U. To take three examples, NP0/C0G types have very close tolerances and no or minimal capacitance variation with temperature or voltage. They also have a low dielectric constant, so they are relatively large for a given capacitance value and voltage rating. As a result, they are also quite expensive. Fig.2: the structure of typical SMD and through-hole ceramic capacitors. SMD ’chip’ ceramics are made of many layers; throughhole disc capacitors may have a single layer construction, as shown here, or increasingly these days, a similar internal structure to a multilayer SMD capacitor. Multi-layer through-hole capacitors are usually encapsulated in epoxy, while the single-layer disc types can be encapsulated in ceramic. (original source: Johanson Dielectrics) Fig.3: the manufacturing process for multi-layer SMD ceramic capacitors. To keep the cost low, they are made in large sheets and after lamination is complete, the sheets are sliced up into individual capacitors. Those are then fired (similarly to ceramic pottery) and the end terminals are added, which provide a way to solder to the capacitor while also electrically joining every second layer. (original source: Johanson Dielectrics) siliconchip.com.au Australia’s electronics magazine March 2021  73 A set of ceramic capacitors ranging from 47pF to 2.2µF. Table 1: Class 2 capacitor codes First letter Middle number Last letter lower temperature upper temperature change in capacitance over given temperature range X = -55°C 4 = +65°C P = ±10% Y = -30°C 5 = +85°C R = ±15% Z = +10°C 6 = +105°C L = ±15%, +15 / -40% above 125°C 7 = +125°C S = ±22% 8 = +150°C T = +22 / -33% 9 = +200°C U = +22 / -56% V = +22 / -82% Fig.4: a cross-section of one layer of a standard aluminium electrolytic capacitor. The anode and cathode are both made from etched aluminium foil, for a large surface area. A thin layer of aluminium oxide is formed on the anode, which acts as the dielectric layer. The conductive electrolyte allows electrons to flow between the cathode right up against that oxide layer, so only the oxide layer separates the two halves of the capacitor, maximising capacitance per area. (original source: Wikipedia) Fig.5: like ceramic capacitors, electrolytics are made up of many layers to give higher capacitance, but they are typically wound in a coil rather than made from flat layers (with some exceptions). Once the leads are attached, the coil is inserted into a can with a rubber bung almost sealing it. We say almost because a small amount of airflow is needed to prevent the electrolyte from drying out. (original source: Wikipedia) 74 Silicon Chip Australia’s electronics magazine On the other hand, Y5V ceramics are very compact for a given capacitance value and voltage rating, but they have a very poor initial tolerance, and their capacitance is reduced dramatically at elevated temperatures and higher applied voltages. The benefit is that they are quite cheap to produce. Dielectrics like X5R and X7R are in between those two; they are larger than Y5V types for a given capacitance and voltage, but smaller than NP0/C0G. Similarly, their tolerances are intermediate, as are their production costs. Therefore, these capacitors are very popular as a reasonable ‘middle ground’. Note that NP0/C0G ceramics are almost ideal capacitors. They have very stable capacitance values over temperature and voltage, low ESL (equivalent series inductance) and ESR (equivalent series resistance), a low dissipation factor and excellent linearity. Their only real disadvantages are a low maximum capacitance value (due to their relatively high volume) and high cost. We’ll describe the meanings of those performance parameters in some detail later in this article. If the two/three-letter schemes given above look like gobbledygook to you, that might be because there are multiple different naming/categorisation schemes. The most common schemes are from the EIA, which consist of a letter, a number and a letter. But they don’t always mean the same things. For the most common type of ceramic capacitors (Class 2), the first letter and following number refer to the minimum and maximum temperature range, while the third letter gives the tolerance of the capacitance over this range – see Table 1. Better-performing capacitors are the ones with a smaller capacitance change over a broader range, like X7R. Electrolytic capacitors In an electrolytic capacitor, one plate is a metal foil while the other ‘plate’ is a conductive liquid or gel solution, known as the electrolyte. The metal foil is etched to increase its surface area greatly, and the liquid or gel is in intimate contact with this foil, separated only by a very thin oxide layer. Therefore, electrolytic capacitors have very high casiliconchip.com.au pacitance values for their volume (see Figs.4 & 5). For this reason, “electros” are typically used for ‘bulk’ bypassing or filtering applications. Their asymmetry, and the fact that the oxide layer is formed by electrons flowing from one ‘plate’ to the other, means that they are generally polarised. So one of their leads must always be at a higher voltage than the other (although there are ways around this – described later). Electrolytic capacitors typically range from a bit under 1µF up to 100,000µF or more, with voltage ratings from a few volts to several hundred volts. Traditionally, the metal foil used was aluminium. However, other metals can be used, giving certain performance advantages. For example, tantalum, while more expensive, generally results in a capacitor which can handle more current and with a lower ESR (see Figs.6 & 7). Further refinements to the electrolytic capacitor came with the discovery that an organic polymer gel could be used as the electrolyte, giving a roughly ten-fold decrease in overall ESR (Figs.8 & 9). As such, organic polymer (‘solid’) aluminium electrolytic capacitors outperform standard tantalum capacitors, and solid tantalum capacitors perform even better again, approaching that of some ceramics but with a better capacitance-to-volume ratio (see Fig.10). Other, more exotic types of electros are hybrid polymer electros (which combine liquid and polymer electrolytics) and niobium polymer electros; niobium is cheaper than tantalum but performs similarly. Tantalum and solid electrolytics tend to occupy the space between traditional electros, which are still A set of electrolytic capacitors ranging from 10µF to 68mF (68,000µF). Note how the can-type electros have a stripe to indicate the negative lead, while the rectangular SMD types have a stripe showing the positive lead. siliconchip.com.au Fig.6: while tantalum electrolytics work on the same principle as aluminium types, the construction method is necessarily quite different due to the properties of tantalum. Tantalum particles are sintered to form a porous slug, which is then immersed in a manganese dioxide electrolyte. Graphite and silver in contact with the electrolyte form the cathode connection, while tantalum wire acts as the anode. (original source: Wikipedia) Fig.7: this gives you an idea of how the tantalum particles are sintered and attached to a tantalum wire lead to form the body of the capacitor. (original source: Wikipedia) Fig.8: polymer or ‘solid’ aluminium electrolytic capacitors use an organic polymer (plastic-like) electrolyte which has roughly ten times the conductivity of a wet electrolyte. This allows for more compact electrolytic capacitors with much higher ripple current ratings and lower ESR values. Other techniques like comprehensive lead stitching are often combined with the polymer electrolyte to maximise performance. (original source: Wikipedia) Figs.9(a) & (b): an alternative construction for polymer electrolytic capacitors which uses the same cathode construction as a sintered tantalum capacitor. This halves the number of dielectric layers, significantly increasing capacitance at the cost of a more complicated manufacturing process and more expensive inputs. Polymer tantalum capacitors are made much the same as regular tantalums, just with a different type of electrolyte. (original source: Wikipedia) Australia’s electronics magazine March 2021  75 Fig.10: an impedance vs frequency graph comparing four different types of electrolytic capacitor and a multilayer ceramic capacitor (MLCC) with the same self-resonant frequency. The tantalum-polymer capacitor comes closest to the MLCC in terms of performance at the resonant frequency, while being superior at lower frequencies, likely due to a higher capacitance value. (original source: Wikipedia) Fig.11: traditional electrolytic capacitors are wound with four layers: two metal foils and two paper separators which are soaked in electrolyte. Note the multiple tabs which ensure that current doesn’t have to flow too far to reach any point on the foils. Anode and cathode tabs are lined up together so they can be welded to the appropriate leads. (original source: TDK) Fig.12: for highperformance (eg, lowESR) capacitors, the lead tabs are stitched into the aluminium foils, with the metal of the lead tab and foil being joined at multiple points throughout the foil to provide a low-resistance, low-inductance path for current to flow. Electrolytic construction Fig.13: SMD electrolytic capacitors come in different forms, but the can style uses very similar construction to a through-hole radial capacitor. The main differences are that the can sits on a plastic platform with the leads bent horizontally under it, so that the capacitor sits on the PCB and the leads rest on their respective pads. (original source: Panasonic) 76 Silicon Chip widely used for bulk filtering, and ceramics, which are used more for local or high-performance supply bypassing. For example, tantalum or polymer electros might be used in switch-mode power supply circuitry, where very high pulse current handling and good filtering (low ESR) is critical. Non-polarised (NP) or bipolar (BP) electrolytic capacitors are effectively two polarised electrolytic capacitors connected back-to-back. You can create an NP electro from two polarised electros by joining either the negative or positive leads together, although the internal construction of an NP/BP may be somewhat different in practice. But the result is the same. This works because when one of the two capacitors is reverse-biased, it acts a bit like a diode, and the other capacitor does all the work. When the voltage reverses, the capacitors swap roles. Strangely, you can often get better performance by connecting two polarised electros back-to-back, at a lower cost than a dedicated NP capacitor! This is probably due to economies of scale; polarised electros are made by the squillions while NP capacitors are used in fairly specialised roles. Another thing to note about electros regarding polarity is that it is safe to apply a reverse polarity voltage longterm as long as it is low (ie, below the threshold where it starts to conduct). This means that polarised electros are quite suitable for use as AC-coupling capacitors even if the polarity of the voltage across them is not known, as long as that voltage never exceeds about ±1.5V DC. Australia’s electronics magazine Traditionally, electrolytic capacitors are ‘wound’. Two long strips of aluminium foil are chemically etched to increase their surface area, then a strip of paper (or some other porous insulator) soaked in electrolyte is sandwiched between them. Each conductive strip has one or more tags, for ultimately attaching leads, projecting from one side (see Figs.11-13). This sandwich is then wound into a roll, with a second paper layer to separate the anode and cathode. With the leads in place, the roll assembly is inserted into a can. More electrolyte is added, and a rubber bung to seal it. Current is passed through the casiliconchip.com.au Fig.14: SMD polymer electrolytic capacitors are available in various packages including cans like regular electros. The main difference is the use of a separator sheet impregnated with a conductive polymer instead of paper soaked in an electrolyte solution. (original source: Panasonic) pacitor to form the required insulating layer, up to a voltage somewhat above the desired rating (which determines the oxide layer thickness). The process is slightly different for tantalum and polymer capacitors; SMD tantalum and polymer capacitors in rectangular prism packages may be made in layers rather than wound. But the result is much the same: a metal conductor with a large surface area separated from a conductive electrolyte only by a very thin oxide layer (see Figs.14-16). If the leads were only connected to the conductive foils at one point each, the ESR and ESL of the capacitor would be poor, as current must flow along a spiral path to reach the inner and outer layers of the capacitor. For this reason, all but the most basic electros usually have extra conductive paths giving current a ‘short cut’ to move between the layers of the capacitor. Higher performance electros also have the tabs ‘stitched’ into the foil at multiple points to reduce resistance and improve conductivity between them, in addition to having numerous tabs spread throughout the roll, that all join to one of the two leads. Plastic film capacitors Fig.15: SMD tantalum electros are made similarly to through-hole types, but the sintered tantalum grains are formed in a rectangular prism shape to create a more compact and convenient package. (original source: Wikipedia) Fig.16: the same type of semirectangular SMD package can also be used to house polymer aluminium electrolytic capacitors. (original source: Wikipedia) Some through-hole tantalum capacitors ranging from 3.3µF to 47µF. For these capacitors, polarity is indicated by a plus sign (+) on one side of the body. siliconchip.com.au Australia's Australia’s electronics magazine Plastic film capacitors are not used as much commercially these days since ceramic capacitors are much cheaper and are available with very low ESR and ESL figures. However, in cases where linearity or safety are essential, plastic films are still widely used. Most plastic film capacitors have better linearity than all but the best (NP0/C0G) ceramics or mica capacitors, and they can be designed to fail gracefully (going open-circuit). Plastic film capacitors are available from a few dozen picofarads up to a few tens of microfarads, and have voltage ratings ranging from around 16V up to several thousand volts. The failure mode is vital in mains applications, where capacitors are connected between Active and Neutral or Active and Earth. If they were to fail short circuit, a fire could result, or it could be a shock hazard. While ceramic X/Y-class capacitors exist, generally, higher-value mains-rated (X/Y) capacitors use either polyester (PET), polypropylene (PP) or polycarbonate (PC) films. March 2021  77 Fig.17: plastic capacitor construction is similar to ceramic, with alternating layers of plastic (the dielectric) and conductive metal film or foil in between, staggered to create many capacitors in parallel (original source: Wikipedia). Like ceramic capacitors, the plastic dielectric used has a significant effect on capacitor properties. And similarly, the plastics with the lowest dielectric constants that result in the bulkiest capacitors (like polypropylene and polystyrene [PS]), tend to have the best performance figures, Fig.18: plastic capacitors can be made from stacks of sheets, similarly to ceramics, or from rolled-up strips, similarly to electrolytics. The stacking process tends to be more timeconsuming and expensive, but it can give better density, so it is used for some SMD plastic capacitors. such as good linearity factors and low dissipation factors. Other plastics used for capacitors include polyphenylene sulfide (PPS), polyethylene naphthalate (PEN) and polytetrafluoroethylene (PTFE). One interesting property of metallised plastic film capacitors is ‘self- healing’. This is where a physical defect or the application of excessive voltage might damage the capacitor, but it will not fail entirely; instead, a small area of the metallisation burns away, reducing the capacitance by (hopefully) a tiny amount – usually not enough to affect its function. Fig.19: the roll manufacturing process for metallised plastic capacitors. While metal foil can be used instead of metallisation, it tends to result in a bulkier capacitor with inferior properties. (Source: Wikimedia user Elcap) 78 Silicon Chip Australia’s electronics magazine siliconchip.com.au A set of film capacitors from 50pF to 1µF. This effect is taken advantage of to provide the higher safety margin required for mains-rated capacitors. Plastic capacitor construction Like ceramic capacitors, plastic film capacitors usually need many layers to give a usable capacitance value. However, they are not normally formed by deposition methods. Therefore, they must be either stacked or wound (see Figs.17-19). Stacked capacitors are made in the way you might guess: with alternating layers of conducting foils and dielectric films. The conducting foils are staggered so that when they are joined along each edge, they form interlocking ‘combs’ and thus effectively, many single-layer capacitors in parallel. Making capacitors that way is timeconsuming and expensive, though. Rather than using metal foils, the dielectric can also be coated with a film of conductive material which provides a thinner and more uniform layer, improving performance and allowing more layers to be packed into the same space for higher density. Wound plastic film capacitors are made similarly to electrolytics as described above. A sandwich is created with the dielectric film between two strips of metal foil, with the strips slightly offset. This assembly is wound up into a roll, and in most cases, the roll is squashed flat to better fit into a rectangular prism shape. A metallisation layer is sprayed onto the ends of the roll to connect the layers, and leads are attached. This is why the conductors are offset; each layer is only exposed at one end of the roll, or else the sprayed metal layer will short out the capacitor. After this, the capacitor is typically impregnated with silicone oil or another insulating fluid to prevent moisture ingress. The terminals are then attached and the capacitor is encapsulated, sometimes by being dipped, other times by being sealed into a pre-formed plastic case. To make plastic film capacitors with a voltage rating above 630V DC, partial metallisation can be used to effectively form multiple capacitors in series using the same basic techniques. This can extend voltage ratings up past 3000V DC (see Fig.20). As well as small-signal capacitors and those for mains filtering, plastic dielectric capacitors are also used in motor run applications. Many motor start capacitors are electrolytic types, but electros are not suitable for handling the continuous current and high voltages that motor run capacitors are subjected to. So they are typically made with polypropylene or similar plastic dielectrics and thick metal films to handle high currents continuously. Electrical double-layer capacitors (EDLs) Fig.20: the basic method of plastic film capacitor manufacturing doesn’t work very well for applied voltages above about 630V. Higher voltage ratings are possible, but the capacitor needs to be internally separated into several elements connected in series, so that the dielectric material only has a fraction of the applied voltage across it. (original source: Wikipedia) siliconchip.com.au Australia’s electronics magazine These are often known as supercapacitors or ultracapacitors. They are a variation on electrolytic capacitors, with extremely high capacitances but usually low voltage ratings, and often very high internal resistances (and thus low operating currents). We published an article on ultracapacitors in April 2008 (siliconchip.com.au/ Article/1793). EDLs use similar conductive polymers with a very high surface area for both the positive and negative electrodes, with a common electrolyte in contact with both. Anions and cations in the electrolyte form insulating Helmholtz layers in direct contact with the surfaces of both electrodes. These layers are only one atom thick, and as mentioned at the start of the article, capacitance is proportional March 2021  79 Fig.21: as the name suggests, a double-layer (EDL) capacitor effectively has two dielectric layers, one at the surface of the anode and one at the cathode, with a conductive electrolyte between the two. The advantage is that these layers are super-thin, just one molecule wide, giving extremely high capacitances in a small package. However, this thin dielectric layer results in a very low voltage rating, typically either 2.7V or 5.5V (source: Wikimedia user Elcap). to surface area and inversely proportional to dielectric layer thickness. You can’t get a much thinner layer than one atom (see Fig.21). Given the large surface area of the electrodes, EDL capacitors can have values exceeding one Farad in a package not much bigger than a can 19mm in diameter and about 16mm tall. The fact that the current must pass through two polymer layers plus an electrolyte, neither of which is especially conductive, is why the current delivery of EDLs is generally limited. The extremely thin dielectric layer is the reason why voltage ratings of only 2.7V or 5.5V are common. Both of these problems can be mitigated by connecting many EDL capacitors in parallel (to improve current handling) or series (to increase the voltage rating, at the expense of capacitance). Higher voltage EDLs usually have multiple internal EDLs in series. You might be using an ultracapacitor without realising it; Mazda introduced its i-ELOOP system in vehicles from 2011, and it is now in many vehicles. This system recovers kinetic energy during braking to rapidly charge an ultracapacitor, then uses that energy to charge the vehicle battery over a longer period. Other types of capacitor Silvered mica capacitors unsurprisingly use mica, a type of mineral, as the dielectric. Mica was chosen both for its good dielectric properties and because its crystalline structure makes it very easy to cleave into super-thin sheets; just what you need to achieve a decent capacitance. A thin layer of silver is applied to each side, and voila, you have a capacitor with excellent linearity and low leakage. An example of an 806pF 300V mica capacitor. The 1% rating means its actual value will be in the range of ~798-814pF. Mica capacitors have mostly been replaced by ceramic or plastic film types, as both are significantly cheaper to manufacture and achieve similar performance. Some still value mica caps for audio circuits. Besides good linearity, another property of mica capacitors is that they usually have tight toler- ances due to their predictable thickness, measurable surface area and low temperature coefficient. Another type of non-polarised capacitor that was widely used but is now far less common (although still available) is the paper capacitor (sometimes known as an MP [metallised paper] capacitor). These have also mostly been supplanted by ceramic or plastic film capacitors. The main disadvantage of paper capacitors is that they can absorb moisture from the air and fail; older types have also been known to catch fire! Modern capacitors usually combine paper and plastic (usually PET or polypropylene) to overcome these disadvantages. Their main advantage is low cost. One benefit that paper capacitors retain is that they usually have zinc metallisation compared to the aluminium metallisation of plastic capacitors. This provides better ‘self-healing’ capabilities due to its lower-energy evaporation process. Variable capacitors Variable capacitors work either by varying the amount of overlap between two sets of metal plates, or by chang- Fig.22: the simplest type of variable capacitor, used in many vintage radios, is just two sets of interleaved metal plates where the amount of overlap can be adjusted. Air is the dielectric. Miniature trimmer caps tend to use a plastic or mica dielectric and bring the two plates closer together or further away to vary the capacitance. 80 Silicon Chip Australia’s electronics magazine siliconchip.com.au Some varcaps (variable capacitors) from various older radios. ing the spacing between two plates (possibly separated by a plastic or mica dielectric). In many cases, the dielectric is air. As such, the range of capacitance for a typical tuning capacitor is generally from a few picofarads up to a few hundred picofarads (see Fig.22). They are typically used as part of an RF oscillator or filter circuit, so low picofarad values give the required time constants with reasonable values for other components (usually inductors). Oddball types The capacitors described above probably cover 99% of the capacitors you might come across, but other types exist. That includes those with a glass or silicon dielectric, or even a vacuum! Capacitor parameters In addition to its construction/dielectric (ceramic, aluminium electrolytic, tantalum electrolytic, plastic film, plastic foil etc), a capacitor is described by its capacitance, tolerance, voltage rating(s), ripple current rating(s), leakage current rating(s), operating temperature and expected lifespan. Furthermore, each capacitor type has several associated performance metrics, which may be fixed or vary with parameters like temperature, applied voltage, signal frequency, age etc. These include the ESR (equivalent series resistance), ESL (equivalent series inductance), dissipation factor (DF or delta [Δ]), temperature coefficient (tempco), voltage coefficient, linearity and more. We’ll describe all of these, starting with the parameters which typically form part of a model or part code. siliconchip.com.au Capacitance: the nominal capacity of the device, measured with no or little charge (typically around 0.51.0V across the capacitor) and at room temperature. For very low-value capacitors (fraction of a picofarad to a few picofarads), the measured capacitance can be affected by the connected PCB tracks/ pad or wires, or even the device’s lead length. Tolerance: how close you can expect the capacitance to be compared to the nominal value. If you have a 10µF ±10% capacitor, if its value is less than 9µF or more than 11µF, then it would be considered faulty. However, during actual use, its capacitance could vary outside this range, as explained below. Voltage rating(s): the applied DC voltage across the capacitor terminals can safely vary from 0V up to this figure. For non-polarised types like ceramic or plastic film, it can also be negative, meaning the full range of operating voltages is effectively doubled (ie, -50V to +50V for a 50V capacitor). Some capacitors have a higher ‘surge’ voltage rating which will not damage them if applied for a limited period, but that is less common these days. Note that you sometimes need to keep the voltage below the rating for good performance; more on that shortly. Ripple current rating(s): all capacitors have some intrinsic resistance and therefore heat up as current passes through them; current flows through a capacitor during both charging and discharging. For example, if a capacitor is used to filter the output from a bridge rectifier turning AC to DC, it supplies the full load current most of the time Australia’s electronics magazine (when the bridge is not conducting). But it also must absorb large pulses of current to recharge when the bridge comes into conduction, 50 or 100 times per second. Such circuits must be designed to avoid exceeding the RMS ripple current rating of the filter capacitor(s) or else they can rapidly overheat and fail. There are generally different figures given at low (50/100Hz) and high (100kHz) frequencies, due to the changing impedance of the capacitor, from both its capacitance and its ESL (see below). The ripple current rating is usually higher at higher frequencies. Leakage current rating(s): some capacitors (eg, electros) can have a fairly significant leakage current through the capacitor even when the voltage is steady. This is usually proportional to the applied voltage. Ceramic, plastic film and mica capacitors also have leakage currents, although they are generally very low and are often (but not always) of no concern. This is important in some applications, like sample-and-hold buffers, where the voltage across a capacitor must remain stable for relatively long periods. Operating temperature: critically for electrolytic capacitors, this is the maximum temperature at which the capacitor is guaranteed to meet the stated performance figures. It is also the temperature at which the expected lifespan (if given) is calculated. Capacitor lifespan roughly doubles for each 10°C below the rated temperature, and halves for each 10°C above it. Typical ratings are 85°C, 105°C and 125°C. We recommend using 105°Crated capacitors with an expected lifespan of at least a few thousand hours to avoid early failures. Lifespan: usually stated in thousands of hours MTBF (mean time between failures), with 1000 hours at the lower end and about 10,000 hours at the upper end. If you can find a capacitor rated to last for 10,000 hours at 125°C, it’ll probably outlast the rest of the circuit! Performance metrics Equivalent series resistance (ESR): this is a crucial metric for most capacitors as it has a strong effect on how well the capacitor can smooth DC voltages, and how much heat is generated at higher currents. March 2021  81 You can think of a real capacitor like an ideal capacitor with a resistor in series; the lower the value of that resistor, the less it is ‘isolated’ from the circuit it is connected to. Typical ESR values are a few ohms for a low-value electrolytic capacitor, down to a few milliohms for a large, low-ESR electrolytic, tantalum, polymer or ceramic capacitor. Equivalent series inductance (ESL): just like ESR, you can imagine that all capacitors have a low-value inductor internally connected in series with the capacitance. This has little effect at low frequencies, but can make the effective impedance of the capacitor so high that it is useless at higher frequencies. ESL is critical for applications like bypassing multi-GHz ICs such as CPUs and RF devices. Smaller capacitors generally have a lower ESL, and certain construction methods can dra- Fig.23: by definition, the temperature coefficient of an NP0 ceramic capacitor is zero (or very close to it). On the other hand, Y5V ceramic capacitors vary in value wildly with temperature. Z5U is a little better at lower temperatures, but still poor at high temperatures. X5R and X7R are the ‘go to’ ceramic dielectrics because they are cheaper and more compact than NP0 capacitors, but have a much more modest temperate coefficient than Y5V or Z5U. (source: Wikipedia & Johanson Dielectrics) Fig.24: the variation with temperature of the dielectric constant, K, for several ceramic materials. You can see that X5R and X7R have a much higher K than C0G/NP0, making for higher capacitances in a smaller volume, with only a slight variation over the temperature range. The vast variation for Y5V and Z5U makes them unattractive. While they give a high capacitance at room temperature, at very high or low temperatures, the K value drops below that of both X5R and X7R. (original source: Digi-Key) 82 Silicon Chip Australia’s electronics magazine matically reduce the ESL of larger capacitors. The combination of capacitive reactance, ESR and ESL produces a characteristic valley-like impedance curve for most capacitors, as shown previously in Fig.10. Dissipation factor (DF): also known as tan(δ), is the reciprocal of the ratio of the ESR and capacitive reactance, and as such, is typically a number close to but slightly less than one. It can be easier to tell how close to ideal a given capacitor is by looking at the DF rather than ESR, and it can also make it easier to compare the performance of capacitors with different values of capacitance. Temperature coefficient (tempco): how much the capacitance changes with temperature. In an application where the capacitance is used to define a frequency (eg, in an oscillator or filter), this must be minimised to prevent frequency drift with temperature. Hence, NP0/C0G ceramics or plastic film capacitors are generally preferred in those roles (see Figs.23 & 24). This is very important to keep in mind with high-K dielectrics like Y5V ceramics; they can lose 80% or more of their capacitance at elevated temperatures! That’s made even worse by… Voltage coefficient: how much the capacitance changes as the capacitor is charged. This causes two main problems. One is a loss of effective capacitance; combined with the poor tempco of Y5V ceramics, a 10µF 6.3V Y5V capacitor at 80°C, charged to 5V, might have less effective capacitance than a 1µF 50V X5R capacitor under the same conditions! (See Figs.25 & 26). This is why we steer well clear of cheaper Y5V ceramics. The other problem with a high voltage coefficient is that it dramatically impacts linearity. So, perhaps unintuitively, standard aluminium electrolytic capacitors are better for audio coupling than most high-quality multi-layer ceramic capacitors. NP0/C0G ceramics are the exception, but they are huge and horrendously expensive at the sort of values required for most audio coupling applications. Linearity: this is not something you will find on most data sheets, and there is no standard way of representing it. But it is a real effect that varies sigsiliconchip.com.au nificantly between different capacitor types. The easiest way to measure it is to form a simple RC filter (low-pass or high-pass) with a relatively low-value, linear (thin film) resistor and a capacitor. You then feed a very pure sinewave into the filter, at a frequency near the -3dB point, and measure the distortion figure of the voltage across the capacitor. The resulting % THD is inversely proportional to the capacitor’s linearity. A very linear capacitor like a polypropylene or NP0/C0G ceramic capacitor will introduce an unmeasurable level of distortion (below 0.0001%). Other plastic film types like polyester are slightly worse, resulting in a measurable but not worrying level of distortion (say 0.0005%). Other capacitors like high-K ceramics, electrolytics and so on could give distortion measurements of 1% or more, reflecting the fact that their I/V curves are not straight. In some cases (eg, electros), the curves can even have hysteresis, meaning they are a different shape for charging and discharging. This is most important in audio circuits, although other circuits (eg, RF) might be sensitive to linearity too. Note that electros are fine for audio coupling, even though they are not terribly linear, as the applied AC voltage in that role is so small that it isn’t a significant effect (unlike the voltage coefficient of many ceramics, which makes them unsuitable for that role, despite probably being more linear than electros). Ageing: capacitors can change value over time, usually decreasing due to degradation of the dielectric. Typically, those with a tighter initial tolerance will tend to maintain their capacitance better over time (see Fig.27). This is apart from an actual failure of the component, which might manifest as a much lower capacitance, higher ESR, higher leakage (especially at voltages approaching the rating) or some combination of the three. Further reading • • • • • • • Types: Ceramic: Electrolytic: Tantalum: Polymer: Film: Supercaps: https://w.wiki/q86 https://w.wiki/q87 https://w.wiki/q88 https://w.wiki/q89 https://w.wiki/q8A https://w.wiki/q8B https://w.wiki/q8C SC siliconchip.com.au SMD Capacitor Actual Size: 0603 0805 1206 1210 1812 Fig.25: the measured value of a range of 4.7µF X5R and X7R capacitors with the application of a range of DC voltages. Note how the physically larger capacitors tend to retain their capacitance better as they are charged to a similar voltage. (original source: Maxim) Fig.26: the change in capacitance over voltage for several different ceramic dielectrics. While the 100V and 400V capacitors seem to perform poorly, consider that the X-axis is a percentage of the voltage rating. Due to this effect and the temperature coefficient, Y5V or Z5U capacitors can easily fall below 10% of their rated values, and below even 5% at temperature extremes! (Original source: Wikipedia) Fig.27: if you need another reason to avoid Y5V ceramics, here is a comparison of the loss in capacitance due to aging with the more robust X7R types. According to the originators of this graph, Johanson Dielectrics, the value of Y5V capacitors drops at roughly three times the rate of a comparable X7R capacitor (original source: Johanson Dielectrics). Australia’s electronics magazine March 2021  83