As described in Lesson-2, a capacitor consists essentially of two conducting plates with an electrical connection to each. The dielectric, which both insulates the plates and keeps them apart, may be air or any of a range of substances which can both provide the necessary degree of insulation and also enhance the performance of the plates as a capacitor.
Good insulation is a prime requirement so that charge cannot leak from one plate to another and, of itself, this is not a difficult problem. However an applied voltage electrically stresses a dielectric and its resulting structural response not only enhances the storage capacity of the device but has an apparent effect on the insulating properties under ac conditions.
When an applied voltage alternates then the electrical stress on the dielectric alternates also and so the material is taken through a continuing cycle of work; work implies a consumption (a loss) of energy. When measurements are made on a capacitor any losses (the absorption of energy) appear as a resistive component and so any equivalent circuit for a capacitor appears as a slightly leaky device.
Clearly the actual loss over any period of time depends on the number of times the dielectric is cycled and so these dielectric losses increase with increasing frequency. Thus the choice of dielectric material is influenced by the use for which a capacitor is required and, in turn, this influences the technique used in its construction.
Large values of capacitance require a large surface area, close spacing of the plates and/or an insulating material with a very large dielectric-constant. The first of these poses problems in handling and packaging, the second poses problems both in manufacture and in maintaining good insulation while the third is a matter of materials research but has obvious limitations.
One other problem lies in the natural inductance which is inherent in any conducting material. Apart from the obvious limitations in the connecting leads the flow of electrons into and out of each plate means that electric currents flow in the plates themselves and so, the larger the plate surface area, so the larger must be the inherent inductance. Such inductive effects are inseparable from the required capacitive effects and so each capacitor is an LC circuit with a natural resonance frequency. At frequencies well below this self-resonance frequency the component does indeed behave like a capacitor; at frequencies well above this self-resonance frequency the component behaves like an inductor; at frequencies around the self-resonance frequency it may well present a low-value impedance but the phase-shift it may cause is a matter for conjecture.
The effect of self-resonance is inescapable and the only defence is to reduce the capacitive value as operating frequencies rise; this raises the self-resonance frequency. Fortunately capacitive-reactance falls with rising frequency and so the ploy is effective. Thus, in an audio amplifier, the decoupling capacitors are measured in µF; in a medium-wave receiver they are measured in nF; in vhf receivers and above they are measured in pF.
It is not uncommon to see large-value capacitors shunted by small value capacitors. In an audio amplifier for example the idea is that a 10 µF capacitor will deal with the audio signals but may appear inductive to intruding or self-generated rf signals; these are then dealt with by the smaller capacitor. The scheme can be made to work but the Designer needs to know what he is doing. Above the self-resonance frequency of the larger capacitor that component appears inductive; this inductance then forms a parallel-resonant (high-impedance) circuit with the smaller-value capacitor. There is a band of frequencies therefore over which the two capacitors in parallel are incapable of acting as a decoupling circuit - they form a band-stop filter. Typically a parallel combination of a 220 µF and a 22 nF resonate at around 5-MHz.
Very-large values of capacitance (1 to 10,000 µF or greater) with a range of working voltages up to about 500-volts are obtained by reducing the plate separation to molecular proportions. Their construction involves the use of electro-chemistry and they are known as electrolytic capacitors. In general a fairly large plate is folded inside a metal can (which forms the other plate) and the dielectric is a chemical mix. A direct voltage is applied between the plates and a very-thin insulating layer is built up by electro-chemical action. The performance of the capacitor is determined by the voltage used in its preparation and so it is essential that this voltage be maintained in use.
These capacitors therefore are polarised and must always be connected “the right way round”. Their insulation is poor and a direct current always passes between the plates. Their use is mainly in decoupling and smoothing (power-supplier) circuits (see Part-5) although they also find use in RC couplings in low-impedance circuits such as bipolar transistors.
An ac (non-polarised) electrolytic capacitor is also made which primarily is for series-operation with ac electric motors.
If only because of their poor insulation electrolytic capacitors are not suitable for use in circuits where stability is important (e.g. oscillators). This is the important point that is likely to arise in the R.A.E.
An important factor for Designers to consider with electrolytic capacitors is known as the ESR-value (equivalent series resistance) which, typically, may be around 1-ohm for a 220µF capacitor. This theoretical resistor is determined by measurement and it represents the dissipation of power when ripple current or any ac flows through the capacitor. The resulting rise in the capacitor’s internal temperature accelerates the drying-up process caused by evaporation and so determines the useful life of such a capacitor. Electrolytics should not be mounted near sources of heat.
Often the ESR value may increase without any corresponding change in the value of capacitance and so a simple check of capacitance value alone may not reveal that the capacitor has become lossy.
These solid-Tantalum dielectric capacitors are polarised and provide capacitances in the range 1 to 500 µF with working voltages up to about 35-volts. They are intolerant of both reverse voltages and voltage-surges.
Reasonably large values of capacitance (around 0.01 to 2 µF) can be obtained by using plates with large areas. These take the form of long strips of metal foil which are insulated by a similarly-shaped strip of material such as waxed paper, polystyrene, polyester or polypropylene film. Waxed-paper capacitors are seldom encountered today except in elderly equipment.
Except for the polystyrene types the plates seem to be formed by metallising both sides of the dielectric strip. The entire assembly is then rolled or folded and packed into tubular or rectangular cases and sometimes encapsulated. Each metal plate extends only to one edge of the insulating strip so that disc-connections to either is made at each end of the rolled assembly. By thus making connection throughout the length of the strip-plate the inductive effect is largely eliminated. In the early waxed-paper types it was not uncommon for this connection to fail with the result that capacitors lost part or most of their capacitance although it is not clear how much of the change in performance was due to the resulting inductive effect.
Polypropylene capacitors offer low dielectric losses with the ability to withstand high alternating voltages. They give excellent high-frequency performance and can withstand high-voltage pulses with fast rise times. They also offer a negative temperature-coefficient of 200 parts per million per °C which means that their capacitance falls by 200/1,000,000 (1/5,000) for each degree-centigrade that the ambient temperature rises.
Polyester capacitors are more lossy and provide lower insulation but their temperature coefficient is around +200 p.p.m/ °C. (A correct combination of these two could yield an overall capacitance that does not change appreciably with temperature).
Polycarbonate capacitors have a performance similar to that of polystyrene types but their temperature coefficient is ± 60 p.p.m/t. These capacitors drift less with temperature but their uncertain direction would not be too satisfactory in an oscillator circuit.
Polystyrene capacitors are less tolerant of temperatures greater than 70 °C but they have excellent high-frequency performance, very good insulation properties and a negative temperature coefficient between 70 and 200 p.p.m./°C.
8.1.2.4 Small-value Capacitors
Small-value capacitors are formed by techniques such as metallising opposite faces of a slab of solid dielectric material or by layering multiple plates and dielectric; the whole is then encapsulated. Very high quality - and expensive - capacitors are made using metal plates separated by a thin plate of mica. Once again of course their properties depend on the material used to form the dielectric:
Silvered mica capacitors offer very low losses, high insulation and excellent stability with a positive temperature-coefficient between 30 and 75 p.p.m./°C. They are very useful in oscillator circuits or any circuit where frequency stability is important.
Ceramic dielectrics offer very high dielectric coefficients (symbol K) which enables the construction of very compact units. In general however they have low stabilities and should not be used in oscillator circuits except those designated as Low-K. Low-K types are specifically made for compensating temperature drifts in oscillator circuits and offer temperature-coefficients which are selectable over the range 0 to -750 p.p.m./°C.
Medium-K types offer small size for use where small changes of capacitance with temperature are not important and some losses can be tolerated.
High-K types offer small size with high relative capacitance but their performance in terms of stability and losses leave much to be desired. They are useful in coupling and decoupling circuits.
These are specifically designed for decoupling purposes to be used where a feed line passes through a screen or through the wall of a screening box or container. They have a coaxial-type construction in which a central conductor is mounted in an outer tube in a ceramic cement. The outer is flanged so that it can be pushed through a hole in the screen and then soldered into place. The result is a useful mounting post which penetrates the screen and with capacitance provided between the conductor and earth; by joining such devices with small chokes or resistors it is possible to build very effective filter units to prevent signals either entering or leaving compartments via the supply lines.
These are low-loss capacitors which are used in transmitter P.A. circuits where large circulating currents are encountered. They are usually of ceramic construction with fixed capacitance. Clearly if an unsuitable unit were to be used in a tank circuit it would overheat and suffer destruction.
Variable capacitors take a variety of forms which depend on the purpose for which they are used.
These have a set of fixed accurately-spaced "vanes” which are insulated from the main frame by which the unit is mounted — i.e. the "stator" usually carries the signal and is said to be “hot”. A second set of equally-spaced vanes is mounted on a spindle so that, as the spindle is rotated, so the “rotor” is moved into and out of mesh with the stator. The technique therefore is to vary the effective area of the capacitor’s plates. The vanes of the rotor are shaped to achieve any required law - i.e. the manner in which capacitance changes with the spindle rotation.
Many such units say be found mounted on the same spindle thus producing a “ganged” capacitor. This is used to tune several circuits simultaneously.
The rotor is invariably electrically a part of the mounting of the unit so that it operates at earth potential. This is mainly to avoid change-of-capacitance effects when a hand is placed on the spindle but it is also a necessary safety precaution where the capacitor carries high voltages (dc or rf). Often it is required that such a variable capacitance should be “hot” on both plates and so that it becomes necessary to mount it on an insulating plinth — in this event it is important to use an insulating shaft-extension.
These are used to equalise stray capacitances in ganged variable-tuned circuits thus ensuring that, as the tuning control is rotated, so the several circuits vary their tuning in step. Trimmers are used also in circuits such as bandpass-coupled pairs (or triples) to adjust each part of the circuit to resonance. The requirement here is that the capacitance value should be adjustable but, once set, it will not be further adjusted in normal use.
The construction of trimmers varies according to frequency-range, use and cost. Many are but miniature forms of the tuning capacitor described above except that the spindle is cut short and a screwdriver slot provided for adjustment. Others, known as compression trimmers, consist of two small plates separated by a thin sheet of mica; the upper plate acts also as a spring and it is pressed down by a small screw which penetrates the assembly to screw into the main frame (i.e. it is the earthy connection).
There are other types also but, in general, they all employ the techniques of either variable-spacing or variable area.
Not truly a capacitor but not to be forgotten is the variable-capacitance diode which permits the capacitance in a circuit to be adjusted electrically by varying the direct-voltage applied in reverse to a semiconductor junction-diode.
These are manufactured specifically for use in reducing radiated and conducted noise signals in mains circuits. They come in two Classes X2 and Y and it is important that they are not wrongly used.
These capacitors are for use in mains-suppression circuits where their failure would not cause electric-shock hazard to a user; e.g. connected directly across the mains supply (live and neutral). They seem to come in two main types namely:
(a) metallised polyester film and encapsulated in flame-retardant plastic or
(b) epoxy-resin impregnated paper.
Both have self-healing properties.
While these do much the same job they are intended for use where their failure could cause electric-shock hazard; e.g. connected between the live and earth wires of the mains supply. They use either metallised polypropylene film or epoxy-impregnated paper and are encapsulated as before in flame-retardant cases. They are manufactured to a much higher specification with greater resistance to the ingress of moisture and cost up to twice as much as Class X2 types.
Three-terminal Class Y “capacitors” are intended for RF filtering and consist of a capacitor to ground (via one terminal) with small inductors each side in the through connections; these take the form of small ferrite beads. They are in fact not truly capacitors but T filter networks.
The three-terminal Delta Class X2 “capacitors” consist of three capacitors connected one between each of the three terminals and are intended for use in suppressing small motors; i.e. they provide a capacitor between the mains-supply leads and one from each lead to earth. This would appear to contravene X2 regulations?
There are several other types available including 3-terminal ferrite-bead capacitors, which provide greater noise attenuation when compared with standard 2-terminal capacitors; motor-vehicle suppressor capacitors; varistor capacitors which, in 3-terminal form, remove fast surges; capacitors and capacitor-networks specifically designed as Contact Suppressors.
All the above capacitors are likely to change their value with temperature changes and, as already mentioned, some increase their capacitance as temperature rises (a positive coefficient) while others decrease their capacitance as temperature rises (a negative coefficient). By suitably mixing the two types it is possible to obtain an overall capacitance which remains substantially-constant. Temperature drift is not normally a problem in, for example, the rf stages of a receiver but its elimination in a Master Oscillator is important especially where tuning diodes are incorporated.
Apart from the correct choice and mixing of capacitor types negative-feedback is useful in those circuit arrangements known as automatic frequency-control (a.f.c.); however frequency-stability then becomes important in the error-sensing circuit which develops the a.f.c. signal.
Back to Top of PageEND OF INFORMATION SHEET 1