Home | Glossary | Books | Links/Resources
|EMC Testing | Environmental
The importance of measurements, and the need for testing:
I suppose it’s true of any field of human endeavor that nothing goes right all the time, and that much care is needed even to make sure that things go right most of the time. Certainly, my own experience in the practice of electronics, and my observation, over the years, of the various problems which have been en countered by my colleagues in this field, has taught me to recognize the potential fallibility both of human beings and of the materials they choose to employ. In particular, however skilful the designer may think he has been in his initial concept, and however thorough he feels he has been in his calculations, there is always a possibility that he may have overlooked some relevant aspect of the system which he has designed. It’s also possible that he may have been aware of, but dismissed as unimportant, some factor which was more significant than he had realized.
Similarly, when a prototype is built to a design which is, in itself, entirely satisfactory, there is always some chance that a constructional fault or a component defect will cause a circuit malfunction. In most cases the existence of a problem will be immediately obvious, and investigations will be made to find its cause and put it right. However, if the fault, due either to design or to component error, is not a particularly conspicuous one, it may remain unseen, resulting in an unexpectedly poor performance of the system. The only really satisfactory way of avoiding the problem of unseen malfunctions is to take very little for granted, and always to make at least a few appropriate measurements of the static and dynamic characteristics of the circuitry. In this task the better the test equipment available, and the more skilled the engineer in its use, the easier it becomes to check whether things are working as they should do -- and, if they aren't, why not.
By static tests, I mean those which relate, in the main, to measurements made on the components in use, such as their resistance, capacitance or inductance (and if there are any doubts about component characteristics, it’s usually much easier to check these before they are installed in the circuit), or to the quiescent conditions which exist in the equipment under test, such as measurements of its DC voltage and current levels, and, perhaps, its operating temperatures.
Dynamic tests are those measurements which are made on the system when it’s in operation -- usually with a known input test signal -- and include output voltage or current swing, frequency and transient response, waveform distortion, voltage or current gain, input or output impedance as a function of frequency, signal to noise level, and, in the case of signal sources, output frequency and frequency stability.
Although I am sure that most 'proper' design engineers would frown on such a suggestion, the use of appropriate test equipment can save a lot of time in calculation, in showing which system works the best, if there is a choice between several, and also, at the prototype stage, in checking that the optimum component values have been chosen for the system. One must always remember, however, that component values may not always be what their markings indicate, and also that components will vary anyway, so the skilled engineer should always seek to design for a median value, to make sure that the system will still work well with component characteristics which depart some what from the ideal value. So, don't 'tweak' a system for perfection, if that perfection then rests on the edge of a performance precipice, or you may find that while the prototype works excellently, all its successors are duds.
Static testing (voltage, current and resistance measurement)
Like some of the other test instruments, these come in two basic kinds, 'analogue' and 'digital', and there is also a division between 'direct reading' and 'electronic' instruments -- digital meters will, obviously, be electronic in their nature. Each of these types have their own advantages and snags.
An ideal voltmeter should present an infinitely high shunt resistance when connected across the circuit under test. Similarly, an ideal current meter should cause a zero voltage drop when it’s inserted in a circuit to measure direct current flow.
Electronic measurement instruments can come much closer to this ideal than any direct reading meter, but require an external power source -- usually a battery, which leads to the disadvantage that it will run down if the meter is inadvertently left switched on --and may suffer from zero or calibration drift with time or changes in temperature.
Digital instruments can give a much greater degree of precision in reading (usually limited to +/-1 in the least significant digit), particularly in measurements of circuit resistance, and are particularly useful in determining the differential potential existing between non-zero points, such as across the load resistor (points 'x' and 'y') of the simple amplifier circuit shown in FIG. 1. However, they don’t usually approach the ideal zero voltage drop condition in current measurement applications as closely as, for example, an electronic analogue meter, and they are virtually use less for measurements of changing voltage or current levels, which makes them very awkward to use as a peak or a null detector during circuit adjustments.
There are, indeed, some digital voltmeters (DVMs) which have a ten or fifteen segment 'bar-graph' display, in addition to the digital read-out, but these offer nowhere near the delicacy of adjustment possible with even a cheap analogue meter.
Most direct reading analogue displays are based on the use of a 'moving coil' indicating instrument, some times called a d’arsonval movement, after its inventor. The construction of these is shown, schematically, in FIG. 2. In this, a coil of wire, wound on a thin rectangular former, is suspended in the gap between a pair of magnetic pole-pieces, and pivoted so that it’s free to swing about its axis. A pointer is attached to this movable coil, and the whole movement is caused to return to some predetermined zero position by means of a pair of spiral wound springs. These are mounted in opposition, so that very little torque is required to cause the movement to swing about its axis. The displacement of the meter movement away from its zero position is due to the interaction of the magnetic field due to the permanent magnet and that due to passage of current through the coil to which the external meter circuit is connected. The resulting movement of the coil is related to the strength of the permanent magnet and the smallness of the air gap in the magnetic circuit. This is reduced by inserting a soft iron collar into the gap between the permanent magnet poles, and the profile of this can be adjusted, if required, to improve the linearity of the display. Ideally the instrument should be both sensitive and robust.
Unfortunately, these requirements are in conflict, in that high sensitivity of indication implies a light coil wound with fine wire, suspended by delicate springs on very low friction bearings in a narrow magnetic gap, while robustness demands the opposite. The skill of the instrument manufacturer lies in resolving these conflicting requirements, especially if, as is common, the movement is to be used in a multi-range test instrument which may have a rough life. A fairly common safety precaution with high sensitivity moving coil (m/c) meters is to incorporate a 'cut-out' mechanism which will interrupt the electrical circuit to the meter movement if the pointer travels beyond its full-scale reading, or if the acceleration of the pointer exceeds some predetermined value.
FIG. 2 The d’arsonval m/c meter movement If a moving coil instrument is to be used solely as a DC voltage indicator, a high resistance, multiple turn coil can be used, the upper limit to the number of turns, and the consequent coil resistance, being set by the lowest full-scale deflection (FSD) voltage likely to be required from the movement. If, on the other hand, it’s to be used solely as a direct current indicator, a low resistance coil would be preferred, with the lower limit to the number of turns being set by the minimum current required for a FSD display.
Typical commercial instruments are offered with FSD current ratings of 25, 50, 100, 250, 500µF, and 1mA, with display scale lengths of 3.8-15cm. The accuracy of indication is generally greater in those movements with longer scale lengths, and ranges, typically, from +/-2.5% to +/-0.25%.
To use a current indicating meter as a voltmeter, an additional resistance, called a 'multiplier', is usually connected in series with the coil, as shown in FIG. 3a. The value of this resistance is determined by the relationship:
…where Re is the total meter coil resistance, RM is the value of resistance required for the multiplier, and Rt is the total circuit resistance needed to satisfy the condition that:
…where I is the meter movement FSD current and V is the required FSD voltage display, in amps and volts respectively.
FIG. 3 Connection of multipliers in a moving coil voltmeter circuit
It’s apparent that the meter can be made to display as high a FSD voltage value as required by this means, provided that the multiplier resistors are able to sup port the voltage present between their ends. In normal practice, high voltage multipliers are arranged by connecting a number of lower value resistors in series, and if the value of the multiplier resistors can be altered by switching, as shown in FIG. 3b, a multiple range indicating instrument can be made.
In principle, the FSD current reading given by a meter can also be changed by connecting other resistors, known as shunts, in parallel with the coil, as shown in FIG. 4a. However, this method of operation tends to demand very low value resistors to give correct operation at high FSD current ratings. As a solution to this difficulty, the 'current multiplier' circuit layout shown in FIG. 4b is normally used -- particularly in multi-range test meters -- because this effectively adds a voltage multiplier to the meter circuit, at higher FSD current taps, and allows the shunt resistors to have somewhat higher values, which arc easier to make.
FIG. 4 Use of moving coil movement as current meter circuit.
In essence, a resistance meter (ohmmeter), is simply a current meter to which a voltage source, B1, can be connected by way of the 'unknown' resistor, as shown in FIG. 5a. An adjustable resistor, RV1, is also connected in the circuit to allow the full-scale reading to be correctly set for a zero resistance value. The higher the voltage provided by B1 and the lower the FSD current of the meter, the higher the value of the FIG. 5 Use of moving coil meter as a resistance measurement circuit unknown resistor for which a measurement can be made.
For measurements on lower values of resistor, a simple design approach is to shunt the m/c movement so that it has a greater FSD current rating. Another method is to use the circuit shown in FIG. 5b, in which the unknown resistor is connected in the position of a current shunt. This layout has the advantage that the resistance reading is in the correct direction --i.e. with the zero resistance value at the LH end of the indicator scale.
As mentioned above, there is a conflict between the requirements of ruggedness and sensitivity in any simple direct-reading analogue instrument, simply be cause the amount of current which the instrument needs to draw, in order to generate the magnetic field needed to displace the pointer from its rest position, increases as the solidity of the coil and the strength of the springs is made greater. Recognizing this fact, the sensitivity of a moving coil voltmeter is normally expressed as Ohms per volt'. A low-sensitivity, and robust, instrument might well be specified as 1000 ohms/volt, which would imply that it was based on a basic 1mA FSD current meter. A10 volt range on such a meter would imply a total meter plus multiplier resistance of 10,000 ohms, similarly a 100V range would imply a total meter resistance of 100,000 ohms.
If a voltage measurement was made using such an instrument it would be inaccurate to the extent that the current drawn by the instrument, which would be in the range 0-lmA, would reduce the potential at the point of measurement. If the circuit being measured was a low resistance one, the error could be tolerable, but in a high resistance circuit the error could be quite large, and, perhaps, in a valve or FET circuit, with high circuit resistances, the change in DC potential which occurred, when the voltmeter was connected, might be so great that the circuit would no longer work correctly.
A more sensitive, and probably more delicate, instrument could well be rated as 20,000 ohms/volt, which would indicate that the fundamental meter movement was of a 50µF FSD type. With such a meter, a 10 volt range would present a meter circuit resistance of 200,000 ohms.
There are, of course, always snags and a basic 50µF movement would probably require 300mV to be applied to the coil to produce a full-scale deflection. This would determine the minimum voltage drop which would be produced by the meter in use as a current meter, which should, ideally be zero.
Moving iron meters
These are not very often found in small power electrical and electronic circuitry, but arc still moderately common in higher power systems. These consist basically of a fixed coil through which the passage of current generates a magnetic field which attracts a pivoted magnetic 'tongue' away from some spring loaded rest position. These movements don’t offer very high sensitivity, but are robust and can be made quite linear. They can offer comparable accuracy to m/c designs.
These instruments make use of the electrostatic attraction which exists between two oppositely charged plates -- or between a charged plate and an earthed plate -- to produce a deflection in a spring-loaded pivoted movement similar to that of a moving coil meter. However, in this case, instead of a coil twisting in a magnetic field, the pivot carries one or two suit ably shaped plates which are arranged so that they may be drawn, by electrostatic attraction, into the gap be tween an oppositely charged pair. The layout used is shown, schematically, in FIG. 6.
FIG. 6 Schematic arrangement of an electrostatic meter movement.
Meters of this type have the advantage that they draw a negligible amount of current from the circuit under test, and since they are insensitive to polarity they can be used to measure AC as well as DC. However, they are not very robust, and they are seldom available for FSD voltages of less than 500V (2.5kV 10kV arc more common ratings), and their meter scale is very non-linear, being very cramped at the low voltage end.
Electronic analogue meters:
A whole new range of test instruments can be made by combining an electronic amplifier with a moving coil milliammeter movement, and these can combine high sensitivity with robustness and freedom from damage through input overload. In the earliest of these instruments, a typical circuit design made use of a balanced long-tailed pair, of the kind shown in FIG. 7 for an arrangement based on discrete transistors. In this type of layout, the display meter is connected between the drains of the two J-FETs, Q1 and Q2, and the voltmeter input is connected, via a suitable voltage multiplier chain, to the base of Q1. A preset trimmer potentiometer connected between Q1 and Q2 sources allows a 'set zero' adjustment.
With contemporary components, a similar result could be achieved, at a much lower level of static battery drain, by the use of an opamp connected as a unity gain voltage-follower, as shown in FIG. 8a. In this circuit, the minimum FSD voltage will be that which would be required to drive the m/c movement to full scale, on its own, but with an input resistance which could be almost as high as one wished, particularly if one used an FET-input opamp type, such as a TL071 or an LF351.
Greater sensitivity can be obtained by connecting the opamp as a DC amplifier, as shown in FIG. 8b. In both cases, a meter zero adjustment can be arranged, if needed, by the use of the opamp 'offset adjust connections' (usually pins 1 and 5 in the 8 lead pin-out configuration) provided for this purpose.
FIG. 7 Simple electronic voltmeter:
A somewhat more elegant, and certainly more versatile, circuit layout is shown in FIG. 9, in which the m/c meter movement is connected in the opamp negative feedback path. In this case, the internal resistance of the m/c movement is almost irrelevant, pro vided that its FSD current demand is less than the maximum permitted output current of the opamp., and the meter FSD is determined solely by the chosen setting of the preset potentiometer RV1. This situation arises in this type of circuit layout because the feed back loop will operate so that the potential difference between the + and -- (non-inverting, and inverting) inputs will be the opamp output voltage (the voltage required to drive the meter to its appropriate scale deflection) divided by the opamp. open-loop gain.
For example, if the opamp has a DC gain of 100,000, and the meter requires 300mV for FSD, the differential input voltage required between the +in and -in connections will be 300/100,000 mV, or 3µF, which is negligibly small.
FIG. 8 Improved electronic voltmeter circuits FIG. 9 Unity voltage gain
In addition to the ability of such a circuit to operate as an exceedingly high input impedance voltmeter, this kind of layout would also allow the construction of a test instrument which had an exceedingly low voltage drop when used as a current meter. For example, the circuit would require an input voltage differential of only 1mV for a meter FSD, if one employed a 500mA m/c display movement and RV1 was set to 20 ohms.
This kind of circuit arrangement would also allow the construction of a very high sensitivity ohmmeter, when used in either of the circuit layouts shown in FIG. 5.
In general digital instruments supplement rather than supplant their analogue predecessors, since both systems have advantages and snags, as noted above. A variety of techniques are used to derive a numerical read-out proportional to the input DC voltage, of which the two most common are the 'double ramp' system, and the 'successive approximation' method.
A typical digital voltmeter circuit layout using the double ramp system is shown, in schematic form, in FIG. 10. In this, the input (analogue form) signal is amplified, as required, by an input buffer stage, and used to generate an output current which is directly proportional to the input voltage. This current is then caused to charge a high quality low leakage fixed capacitor for a precisely controlled period of time. At the end of this time, an electronic switch is actuated which causes the ramp output voltage to fall linearly towards zero, through a constant current load circuit.
This switch also resets the pulse counter so that it will display the number of clock pulses which occur during this downward portion of the voltage ramp, in the time between the ramp voltage starting its downward slope and reaching the zero level once more. A single cycle of operation is shown in FIG. 11.
% Input buffer^ Xtalc Voltage |to current! converter! Oscillator Pulse counter and display! Time interval generator Switching logic
Capacitor discharge period
Counter reset; %2 Reset
FIG. 11 Operating cycle of double ramp digital voltmeter
FIG. 10 Schematic layout of digital voltmeter
Because such digital signal conversion systems usually have a fixed full-scale input voltage range, such as 0.1999V or 1.999V, and the use of input voltage gain stages will increase the risk of zero drift, digital current meters usually have a greater insertion voltage drop, sometimes called the voltage burden, than electronic analogue instruments. However, with this proviso, it’s just as simple to construct a multi range digital volt/amp/ohm meter using a digital display meter as it’s to do so with an analogue one, using the types of circuit arrangement shown above for analogue meters. The AC bandwidth of inexpensive digital multimeters (DMMs), is usually limited to a few kHz by the characteristics of the coupling trans former/rectifier circuit used to derive a DC output voltage from the AC input signal.
The successive approximation technique uses the internal electronic 'clock' to generate a binary coded sequence of decreasing magnitude voltage steps, as shown in FIG. 12. At the end of each step, the output from a voltage comparator is used to sense whether the sum of the 'step' voltages is greater or less than the input voltage, and then either to reject (logic 0), or to accept (logic 1), the last binary encoded voltage step from the binary staircase sequence. This provides, directly, a digitally encoded transform of the input voltage level at the instant of sampling.
FIG. 12 Operation of successive approximation type of digital voltmeter
Inductance and capacitance measurement:
Various techniques are used to make these types of measurement, depending largely upon the relative importance of convenience or precision. Of the various systems which have been proposed or used, the most common, in the past, was one or other of the 'bridge' arrangements shown in FIG. 13, in which the resistance or impedance of one or more of the arms of the bridge would be adjusted to give a 'null' reading on some adequately sensitive indicator mechanism. If the adjustments of the reference arms are precisely calibrated, bridge measurement systems can give high accuracy, but achieving a null reading may require the sequential adjustment of several controls, which is usually time consuming.
FIG. 13 Some of the common bridge circuits used for inductance and capacitance measurement --- Wien Schering; Maxwell Owen; Hay Resonance.
Alternatively, the value of the component under test can be measured by connecting it in a circuit whose performance is determined by the inductive or capacitative impedance of the unknown component, perhaps as part of the frequency determining network in an LC or RC oscillator, from which the value of the L or C can be discovered by noting the resultant resonant frequency of the system.
However, for a simple indicating meter, the most straightforward approach is just to use the kind of layout shown in FIG. 5a, but using an AC current measurement circuit, with a source of AC voltage in place of Bl. One can then make use of the fact that, for any input frequency, there will be a precise relationship between alternating current flow and the value of the capacitor or inductor, in Farads or Henries.
Appropriate AC current meters are shown in the section covering dynamic measurements.
Measurement of Q:
A particular type of measurement which is of interest where inductors are intended for use in resonant circuit applications, more particularly at high frequencies, is that of the circuit magnification factor, or Q. A variety of techniques have been proposed for this purpose, but many of these are variations of the simple layout shown in FIG. 14, in which the magnitude of the AC voltage developed across a tuned circuit, with a standard value, high quality capacitor, in series with the unknown inductor, is determined when the system is forced into resonance by an input signal from a constant output voltage AC signal generator.
FIG. 14 Simple Q meter
The measurement of LC resonant frequency will also allow the values of the capacitor or inductor to be determined, by calculation, from the relationships ...
... where f_r is the frequency of resonance, and L and C are the values of capacitance and inductance of the components under test, provided that the signal generator output frequency, and either the value L or C is known.
Chopper type DC amplifier systems:
Dynamic testing Because of the difficulties associated with voltage drift in electronic amplifiers, due to thermal effects and component ageing, the measurement of very low voltage levels presents particular difficulties. A commonly used method for avoiding or minimizing this problem is the use of a chopper system of the kind shown, schematically, in FIG. 15. In this circuit, an input switch, SI, is used to connect the input of a gain stabilized AC amplifier alternately to the input DC signal voltage and to the signal OV line. If the input switch is rapidly reciprocated between these two input points, the incoming DC potential will be converted, at the input to the amplifier, into an equivalent alternating rectangular waveform, and this can then be amplified. If the amplified AC output signal is then rectified, it will provide a DC output which is directly proportional to the input DC voltage, but unaffected by any drift in the amplifier DC levels. In early systems, the chopper mechanism was commonly used to operate switches working in synchronism at both the input and the output of the amplifier, in order to provide the necessary output AC to DC conversion, but, with contemporary equipment, electronic input switching systems and AC/DC conversion techniques -- see below -- are more commonly employed.
FIG. 15 Simple chopper-type DC amplifier.
The simple system shown in FIG. 15 would only be able to operate with steady state, or only slowly varying, input signals -- depending on chopper speed.
This has encouraged the development of chopper stabilized systems, in which a conventional opamp. is used as the amplifier, but with a 'nulling amplifier' used to monitor and cancel the main amplifier DC drift. Because of the great value of such chopper stabilized amplifiers for low level DC signals, complete circuits of this kind are now available as IC packages, such as the LCT1052, the ICL7650 and the ICL7652 devices. These offer effective input DC drift levels as low as 100nV/_/month.
AC voltage and current meters
Direct reading instruments:
The simple d’arsonval type moving coil current meter can be used to indicate the magnitude of an applied AC voltage or current by rectifying the input voltage waveform, using one or other of the rectifier circuit arrangements shown in FIG. 16, but all of these arrangements will suffer, to some extent, from the fact that all solid-state rectifiers have the less than ideal kind of conduction characteristics shown in FIG. 17. The principal problem which results from this type of forward/reverse conduction characteristic are that a finite forward voltage must be applied before the rectifier will conduct at all, and that there will also be some reverse conduction.
FIG. 16 Meter rectifier circuits
Those types of rectifier, such as the silicon junction diode, which have low reverse leakage currents, at voltages below their reverse breakdown potential, re quire a forward potential of 0.55-0.6 V before they will conduct at all, while those rectifying diode systems, such as germanium P-N junction diodes, or cop per/copper oxide rectifiers, which have relatively low forward threshold potentials, suffer more from reverse leakage. This means that any simple meter/rectifier circuit will tend to be rather nonlinear in its response to low voltage inputs: a problem which can be minimized by including a resistor in series with the meter movement, of a value which will swamp the variation in the rectifier resistance with voltage in its forward (conducting) direction.
A further problem with any alternating voltage indicating instrument is that its indication will be influenced by the waveform of the input signal -- as shown in FIG. 18 -- a circumstance which will be of little importance with a symmetrical sinusoidal input waveform, but will influence the reading quite a lot if the input signal is non-sinusoidal: a factor which must be borne in mind when making measurements.
Averaged DC output:
FIG. 18 Influence of input waveform on average DC output voltage
A simple half-wave rectifier circuit such as that shown in FIG. 16a, will, with a sinusoidal input voltage waveform, give an output which is approximately half the RMS value. If the meter is calibrated to take this factor into account, it would give an erroneous reading -- either too high or too low -- if the input voltage waveform was non-symmetrical. The particular circuit illustrated would also suffer from the problem that any DC component, associated with the AC voltage to be measured, would also affect the meter reading, as would also be the case in the full wave rectifier circuits of FIG. 16e, and 16f.
This problem can be avoided by using an input DC-blocking capacitor, as shown in FIG. 16b and 16c, but with the penalty that the meter would lose sensitivity at lower measurement frequencies, unless C1 was made adequately large. Since an alternating input voltage would be applied to the rectifier circuit, a polar (i.e. electrolytic) capacitor type would be unsuitable, and large value non-polar (i.e. polyester or polycarbonate film type) capacitors are bulky.
The current transformer input arrangement shown in FIG. 16d provides a better solution to this problem, in that, in addition to isolating the meter/rectifier circuit from any incoming DC potentials, if it’s connected as a voltage step-up component, it will allow the voltage present across the meter/rectifier circuit to be increased, thereby reducing the importance of rectifier nonlinearity. Such an input transformer will, of course, also impose some limits on the bandwidth over which the meter will give an accurate reading.
Full-wave rectifier systems such as the voltage doubler layout of FIG. 16c, and the bridge circuits of FIG. 16e, and 16f, tend to be less affected by input waveform asymmetry, because they tend to average the readings from the two halves of the wave form. However, while a sinusoidal input will give a meter reading which approximates to a true RMS value (0.707Vin) in the case of a sinusoidal signal; a rectangular waveform will give a peak value reading, and both triangular and sawtooth waveforms will give readings which approximate to half of the peak value.
All of the circuits shown in FIG. 16 can be made to give an approximate peak voltage indication by putting a capacitor, C2, across the meter, as shown in FIG 16a.
Because of the need to insert a swamp resistor in series with the meter movement, to lessen the nonlinearity of the scale reading due to the rectifier characteristics, it’s common practice in commercial instruments to use the layout shown in FIG 16f, where a pair of swamp resistors are used in place of the other half of the rectifier bridge, either as shown or, more commonly, in combination with a DC-blocking input transformer, as shown in FIG. 16g.
Electronic meter systems:
The use of electronic amplification techniques pro vides an elegant means of increasing meter sensitivity, and minimizing any nonlinearities due to meter rectifier characteristics, as well as avoiding errors due to the presence of DC components in the signal being measured.
A typical AC voltmeter circuit based on an opamp and a moving coil meter/bridge rectifier combination is shown in FIG. 19. In this circuit arrangement, the fact that the meter/rectifier combination is placed in the feedback path of the amplifier means that the closed-loop negative feedback arrangement acts to generate a feedback voltage across R2 and C2 which is closely similar to the input voltage present across R1.
During those parts of the voltage swing across the meter/rectifier combination in which the silicon junction diode rectifiers are open circuit the negative feed back loop around the amplifier is disconnected, so that the amplifier output voltage (as seen at point '?') is driven, with the full opamp open-loop gain, to 'jump' across the rectifier open circuit voltage band, as shown in FIG. 20a. The voltage developed across R2 is as shown in FIG. 20b and that across the meter as shown in FIG. 20c.
FIG. 19 Electronic AC voltmeter circuit
FIG. 20 Voltage waveforms in circuit of FIG. 19
Because there is no DC path through R2 and C2, the circuit will be insensitive to DC voltages present across Rh even were these not, in any case, blocked by C1. In this circuit, the meter FSD will be ImR2, where Im is the meter FSD current, so that, for example, if R2 = 100 ohms, and Im = ???µF, a full scale deflection of the meter would be given by a voltage of 10mV. The sensitivity of the circuit can be increased, as required, by increasing the overall AC stage gain of the circuit. The basic layout of such a multi-range AC millivoltmeter, is shown in FIG. 21. If the design specification for such an instrument was that it should have a bandwidth of 3Hz 1MHz, at its -3dB points, and a scale range of lmV-100V FSD, this order of bandwidth and sensitivity could be achieved with the basic meter/rectifier circuit without difficulty. However, there are problems which must be resolved in the design of the input attenuator, in respect of the desired input impedance (which should be high, to limit the error caused by connecting the meter to the circuit under test), and the (input open circuit) noise level at zero input signal, which is due to the thermal noise of the input attenuator resistors, and should obviously be as low as possible.
FIG. 21 Multi-range AC milli-voltmeter.
The designer has two options in this circuit, of which the first is to use a low resistance input circuit, such as the constant impedance attenuator network of the kind shown in FIG. 22, which offers an identical input and output resistance at all range settings, and however long the attenuator chain. In this kind of circuit layout, if K is the attenuation, Rt is the necessary value of resistance required to terminate the line, at both ends, and Rc is the required characteristic resistance of the attenuator (i.e. the value of resistance which will be seen by an ohmmeter between the 0V rail and any tapping point on the network), then the value of a will be given by ...so that if the desired attenuation from one step to the next was 10, and Rc -- the input impedance of the meter circuit -- is to be 10kohms, then a will be 99k, b will be 12.22k, and Rt will be 11.0k.
FIG. 22 Constant impedance attenuator network.
Assuming that this level of input resistance was acceptable in a general purpose multi-range instrument, and it’s somewhat on the low side, nevertheless it would still be associated with a background thermal noise voltage, over the 1MHz instrument bandwidth, of some 12µF. -- equivalent to just over 1% of the full-scale reading, which would cause a constant zero offset on all voltage ranges. This would probably be just about tolerable since the offset could be trimmed out by the meter zero adjustment, and the consequent scale error would not be great. On the other hand, increasing the characteristic resistance of the attenuator network to 100k, which would be more suitable for a general purpose instrument, would increase the background meter reading, due to thermal noise, to some 40.6µF, or 4% of FSD, and this would certainly not be acceptable.
Using the simple resistor chain attenuator arrangement shown in FIG. 23, in which the input resistance at the maximum sensitivity position was 1k, would reduce the zero-input-signal noise threshold to 4µF, or 0.4% of FSD, but this circuit layout would need some very high resistance values -- difficult to obtain with any precision -- for the 10V and 100V input ranges. Also, if it’s required that the basic HF -3dB response of the meter remains the same, regardless of the range setting, then care must be taken to keep the stray capacitances of the attenuator network as low as possible, so that their reactive impedances (Zc = 1/(27ifC)), are high in relation to that of the attenuator multiplier resistors. This would be feasible with the low impedance attenuator arrangement of FIG. 22, but not with that of FIG. 23.
FIG. 23 Resistor chain attenuator circuit for 100V-lmVrange
A practical compromise, easily achieved with electronic meter systems, is to use an input attenuator to select the required FSD sensitivity over the ranges 1mV to 100mV, and then to adjust the gain of the amplifier feeding the meter/rectifier circuit to provide the 1-100V range settings, as shown in the final instrument design illustrated in FIG. 24. The only penalties with this arrangement are that on the 1mV FSD range, the meter will have a low (1k) input resistance and that the low frequency -3dB point is increased to 30Hz on this range. On all range settings above 100mV FSD the meter would accept an input voltage overload of up to 300V without damage. The variable resistor, RV1, is used, when the instrument is initially assembled, to set its FSD sensitivity to the required full-scale values, using an AC voltage reference standard.
FIG. 24 Wide range milli-voltmeter with two stage attenuation.
Waveform distortion measurement:
It’s often desirable that the waveform at the output of an AC signal processing or amplification system is the same as the waveform at the input. If the signal is non-sinusoidal, this kind of waveform distortion, when it’s present, can usually be seen on a cathode-ray oscilloscope display, provided that the change in waveform shape is big enough. However, even when the distortion can be seen, it may still be very difficult to define the effect in numerical terms.
With the rather more restricted category of sinusoidal (single tone) signals, a variety of techniques are available which can define, and quantify, the wave form distortions introduced by the system. These types of test may not be so informative about system defects or malfunctions, but are widely used simply because they do give numerical values by which the linearity of the system can be specified. These methods fall, generally, in two categories, 'notch filtering' and 'spectrum analysis'.
THD measurement by notch filtering:
This technique for measuring total harmonic distortion (THD) relies on the fact that a 'pure' sinusoidal signal will contain only the fundamental frequency,/0, free from any harmonics at 2/0,3/0, and so on. If the signal under test is fed to a measuring system which has a notch in its frequency response, as shown in FIG. 25, and if this notch is sufficiently sharp that it will remove the signal at/0 completely, but give negligible attenuation at 2/0 and higher frequencies, then a measure of total harmonic distortion can be obtained by comparing the magnitude of the incoming signal, measured on a wide-bandwidth AC milli-voltmeter, in the absence of the notch, with that which remains after the fundamental frequency has been 'notched out'. There are a variety of circuit arrangements which can be used to provide a notch in the frequency response of an AC measuring instrument, such as the parallel tuned LC tuned circuit shown in FIG. 26a, the Wien nulling network shown in FIG. 26b, or the parallel-T notch network shown in FIG. 26c. Of these layouts, the tuned circuit would only be useful, in practice, at frequencies above some 50kHz. Below this frequency the size of the inductor would be inconveniently large. There could also be a problem of hum pick-up due to the inductor itself. Both the Wien and the parallel-T arrangements are practicable layouts for use in the 3Hz-100kHz range, though they will normally need to be somewhat rearranged to allow the use of negative feedback around the loop to sharpen up the frequency response of the system.
FIG. 25 Frequency response notch
FIG. 26 Possible notch circuit arrangements
The principal advantage of the parallel-T notch net work is that it can be connected directly between the incoming signal and the measurement amplifier (e.g. an AC millivoltmeter) so that no additional noise or waveform distortion is introduced by any input buffering or amplifying stages. Two practical notch type distortion measuring systems of this kind, in which the sharpness of the notch is improved by the use of negative feedback around the notch-filter/amplifier loop, are shown in FIG. 27 and 28. These are capable, respectively, of measuring harmonic distortion residues as low as 0.008% and 0.0001% of the magnitude of the fundamental frequency, but this measurement of signal residues will also include any hum and noise voltages present on the input signal -- since these will still be left after the test signal, fo, has been removed by the notch circuit- and this snag often limits the sensitivity of the system.
The Wien network notch type of distortion meter is easier to use, on the bench, once the display sensitivity has been set, because it only requires two-knob adjustment (frequency tuning and bridge balance adjustment), whereas the parallel-T network generally requires three independent adjustments to achieve a precise null setting. If the signal being measured is at, say, 400Hz and above, reductions in the meter reading error, due to the presence of 50/60Hz and 100/120Hz mains hum, can be made by introducing a steep-cut high-pass filter, with a turnover frequency at, say, 400Hz, giving, perhaps, a -50dB (316x) attenuation at frequencies of 120Hz or lower.
Similarly, the error in the true THD reading due to the presence of the wide bandwidth white noise volt age present in the output can be reduced by restricting the measurement bandwidth by including a sharp cut off low-pass filter in the measurement circuit -- prefer ably with a switched choice of turnover frequencies, at, say, 10kHz, 20kHz, 50kHz and 100kHz.
Intermodulation (IM) distortion measurements:
One of the principle problems caused by nonlinearities in an amplifier transfer characteristic is that the magnitude of the output signal from such a circuit will vary according to whereabouts it sits on the transfer curve, as shown in an exaggerated form in FIG. 29. This effect is always present to some extent in any amplifier which has not got a ruler-straight input/output characteristic. In practice. This means that when two signals are present simultaneously -- let us assume that one is large and one is small -- the output voltage swing produced by the larger one will sweep the smaller up and down the transfer curve, modulating the amplitude of the smaller signal by so doing. This will lead to the generation of sum and difference output frequencies, so that if the two signals are f1 and f2 respectively, there will be present at the output, in addition f1 and f2, signals at f1+f2 and f1-f2. Putting some numbers to this effect, if two signals, at 60Hz and 5kHz, were introduced to a distorting amplifier, the resulting 60Hz modulated 5kHz output voltage will contain spurious frequency components at 4940Hz and 5060Hz. An IM distortion test based on signals at these frequencies, at a 4:1 amplitude ratio, has been recommended by the SMPTE (the USA Society of Motion Picture and Television Engineers).
FIG. 27 Practical distortion meter based on Wien bridge notch arrangement.
FIG. 28 Practical distortion meter circuit based on parallel T notch layout.
A practical measurement instrument for the SPMTE type test is shown, in outline, in FIG. 30. In this, the composite signals, after transmission through the system under test, are passed through a high-pass filter which removes the 60Hz component. The modulated 5kHz signal is then demodulated, which will result in the re-emergence of a 60Hz signal, and passed through a steep-cut low-pass filter to remove the 5kHz signal.
The magnitude of the residual 60Hz signal, as a pro portion of the 5kHz one, is then defined as the IM distortion figure.
An alternative IM distortion measuring technique has been proposed by the CCIF (the International Telephone Consultative Committee). In this procedure, the input test signal consists of two sine-waves, of equal magnitude, at relatively closely spaced frequencies, such as 9.5kHz and 10kHz, or 19kHz and 20kHz. If these are passed through a nonlinear system, sum and difference frequency IM products will, once again, be generated, which will result in spurious signals at 500Hz and 19.5kHz, or 1kHz and 39kHz, respectively, for the test signal frequencies proposed above. However, in this case, the measurement technique for determining the magnitude of the IM difference-frequency signal is quite simple, in that all that is needed is a steep cut low-pass filter which will remove the high frequency input test tones, and allow measurement of the low frequency spurious signal voltages. A suitable circuit layout is shown in FIG. 31.
In audio engineering practice, the value of IM testing, as, for example, in the case illustrated above, is that it allows a measurement of system nonlinearity at frequencies near the upper limit of the system pass band, where any harmonic distortion products would require the presence of frequency components beyond the pass-band of the system. Because these harmonics would be beyond the system pass-band they would not be reproduced, nor would they be seen by a simple THD meter, although the presence of nonlinearities would still result in the generation of spurious (and signal muddling) IM difference tones which were within the pass-band.
FIG. 29 Intermodulation distortion due to nonlinear amplification characteristics
Mixer; Low-pass; Demodulator filter; Amplifier; High-pass under test filter; AC millivoltmeter.
FIG. 30 The SMPTE intermodulation test arrangement
FIG. 31 The CCIF intermodulation distortion measurement procedure
The second common method of measuring harmonic and other waveform distortions is by sampling the output signal from the system under test by a metering system having a tunable frequency response, so that the measured output voltage will be proportional to the signal component over a range of individual output frequencies. The advantage of this technique is that if the system under test generates a series of harmonics of the fundamental frequency, but there is also an inconveniently large amount of mains hum and circuit noise, which would reduce the accuracy of reading of a simple notch-type THD meter, the relative amplitudes of the individual harmonics -- as well as hum voltages -- can be individually isolated and measured.
In use, the gain of the analyzer amplifier is adjusted to give a 100% (0dB) reading for the fundamental (input test signal) frequency, and the relative magnitudes of all the other output signal components are then read of f directly from the display. This type of analysis can also be used to reveal the presence of IM distortion products resulting from two or more input test tones.
The background noise level will depend on the narrowness of the frequency response of the frequency selective circuitry in the analyzer, and this will generally depend on the frequency sweep speed chosen, with a lower resolution for a fast traverse-speed oscilloscope display than for an output on a more slowly moving paper chart recorder.
Spectrum analyzers of this type are used at radio frequencies as 'panoramic analyzers', to show the presence and magnitudes of incoming radio signals in proximity to the chosen signal frequency, as well as for assessing the freedom from hum and spurious signal harmonics in an oscillator or amplifier design.
A typical circuit layout for a spectrum analyzer using the superhet system is shown, schematically, in FIG. 32. In this, the signal to be analyzed, and the output from a swept frequency local oscillator signal are separately fed to a low distortion mixer, typically using a diode ring modulator layout, and the composite sum and difference frequency signals are then amplified by an intermediate frequency gain stage, and converted into a unidirectional voltage by a logarithmic response detector circuit. If the frequency of the local oscillator is coupled to the sweep voltage which determines the position of the 'spot' on the horizontal trace of the oscilloscope, or the position of the pen on a paper chart recorder, there will then be a precise and reproducible relationship between measurement frequency and spot or pen position, so that the oscilloscope graticule or the recorder chart paper can then be calibrated in terms of signal magnitude and signal frequency.
Oscilloscope Linear, display 20kHz mixer 100 kHz Demodulator; 20kHz mixer; Input signal; Swept frequency oscillator; Time-base sweep voltage
FIG. 32 analyzer
Schematic layout of spectrum:
A typical spectrum analyzer display for a high quality audio amplifier, showing mains hum components, as well as the test signal and its harmonics, is shown in FIG. 33.
FIG. 33 Typical spectrum analyzer display
Transient distortion effects:
While the discussion above has been confined to the types of waveform distortions which will affect steady-state (i.e. sinusoidal) waveforms, for many circuit applications it’s important to know of the waveform changes which have occurred in 'step function' or rectangular wave signals, as illustrated in FIG. 34. Since these types of abrupt change in a pre-existing voltage level will imply, from Fourier analysis, an infinite series of harmonics (mainly odd order- such as 3rd, 5th, 7th and so on -- in square-wave or rectangular-wave signals), any signal handling system with less than an infinitely wide pass-band will result in a distortion of the input waveform. This may be just a rounding-of f of the leading edge of the signal waveform -- due to inadequate HF bandwidth, or 'overshoot' or 'ringing', as illustrated in FIG. 34, due to the introduction of spurious high frequency components, or just due to errors in the relative phase of the composite signal harmonics.
Input waveform; Time
FIG. 34 Typical transient distortion effects
It’s difficult to derive numerical values, from simple instrumentation, which relate to the nature of any of these rectangular waveform signals, except 'rise time', overshoot magnitude, 'settling time' -- to some specified accuracy level -- or slewing rate (the maximum available rise-time, where this is limited by some aspect of the internal system design), characteristics also shown in FIG. 35. These effects are most easily seen on an oscilloscope display, and if it has calibrated X and ‘Y' axis graticules, their magnitudes in time or voltage, may be approximately established.
Voltage; Settling time; Rise time
FIG. 35 Quantifiable transient response characteristics.
The oscilloscope is one of the most useful and versatile analytical and test instruments available to the electronics engineer. Instruments of this type are available in a wide range of specifications, and at a similarly wide range of purchase prices.
In its most basic form, this type of instrument consists of a cathode ray tube in which the electrons emitted by the heated cathode are focused onto the screen to form a small point of light, and the position of this 'spot' can be moved, in the vertical, 'Y-axis', and horizontal, 'X-axis', directions, by deflection volt ages applied to a group of four 'deflection plates' mounted within the body of the cathode ray tube. In normal use, the plates controlling the horizontal movement of the spot are fed with a voltage derived from a linear sawtooth waveform generator, giving an output voltage, as a function of time, which is of the kind shown in FIG. 36. The effect of this is to allow signals displayed in the vertical, Y axis, to be seen in a time sequence. A range of sawtooth waveform repetition frequencies is normally provided, giving a range of deflection speeds, which will provide cm/s, cm/ms, or cm/ps, calibrations.
FIG. 36 Output sweep voltage applied to oscilloscope X-deflection plates --- Time:
The X deflection (time-base) scan speeds are con trolled by a range switch, allowing a switched choice in a 1:3:10:30:100:300, type of sequence, but it’s also necessary for there to be a variable time-base scan speed control to allow the choice of intermediate values between these set choices of time-base frequency.
An important feature of any time-base sweep generator system is that it should allow the X scan to be synchronized with the periodicity of the signal to be displayed, where this is repetitive, so that the wave form pattern will appear to be stationary along the X axis. The more versatile and effective this synchronization facility is, the easier it will be to use the scope.
A typical time-base sweep voltage generator circuit is shown, schematically in FIG. 37. It’s normal practice in better quality instruments to provide both a time-base sawtooth waveform output, for the control of other instruments, such as spectrum analyzers or frequency modulated oscillators (wobbulators), and also a time-base switch position -- sometimes designated 'XY' -- which will allow the injection of an external signal into the X amplifier circuit. This facility will allow the oscilloscope to be used for measurements such as the relative phase of two synchronous waveforms, and also for determining the repetition frequency of an unknown signal, by means of 'Lissajous figures', of which some simple patterns are shown in FIG. 38.
FIG. 37 Schematic layout of time-base generator circuit ---[ Signal from ? T* amplifier / Trigger circuit /sawtooth A ramp voltage; Time base voltage g output generator Vertical L- deflection 1 voltage ]
FIG. 38 Typical Lissajous figures --- [ Y;n in phase Y;n in antiphase f in with]
A separate signal input to a low distortion wide bandwidth voltage gain stage provides the vertical,' Y' axis, deflection of the oscilloscope 'trace'. In most modern instruments this will be a DC amplifier design, but with a switched capacitor in series with the input circuit to allow the DC component of an input compo site signal to be blocked. This would be essential in cases where the presence of a DC offset voltage would otherwise prevent the full range of amplifier gain being used, such as when it was desired to examine a relatively low magnitude AC signal present at the same time.
Typical amplifier specifications offer gain/frequency characteristics which are substantially flat from DC to perhaps 10, 20,50, or 100MHz, with some specialist scopes giving upper -3dB turn-over frequencies as high as the GHz range. In the lower bandwidth designs, preset switched input sensitivity values ranging from 1mV/cm to, perhaps, 300V/cm are offered.
Multiple trace displays:
The enormous usefulness of being able to display two or more signals, simultaneously, so that one could, for example, with a dual trace instrument, look at the input and output waveforms from a signal handling stage, has made multiple beam scopes much more common. The early twin-beam oscilloscopes used a cathode ray tube which had two entirely separate electron gun, focus and' Y deflection assemblies, so that the incoming X axis signals could be displayed entirely independently. However, this type of tube is both expensive and bulky, so that, in modern instruments, a single gun tube is used, and the time base trace is split, electronically, into two or more separate scans, by applying a rectangular waveform voltage signal to the input of the final stage of the deflection amplifier. The same signal is also applied to an electronic switch between the outputs of the amplifiers and the output display driver stage. A schematic layout for this type of multiple trace display system is shown in FIG. 39.
What happens, in practice, at higher scan frequencies, is that, for a time interval corresponding to the duration of a single 'Y' axis sweep, a vertical deflection voltage derived from, say, the G front panel shift control, combined with the output from the Y amplifier itself, are switched to the Y axis output voltage amplifier and deflection plates. During the next X axis sweep, a vertical deflection voltage from the front panel 'Y2' vertical shift control, and the output from the amplifier, are switched to the display system, but the outputs from the amplifier and shift controls are disconnected. Similarly, if there are input circuits, these will also be used to display their outputs on successive time-base sweeps.
Because of the persistence of vision, at higher time base speeds, the successive appearances of signals from the various Y inputs will be merged in the perception of the viewer, and will seem to be present simultaneously. Moreover, since the vertical shift volt ages are also switched, it’s possible to move the sequential Y displays, independently, on the scope screen, exactly as if there were a number of completely independent gun and deflection assemblies within the tube.
At lower sweep frequencies, where the limited persistence of vision would no longer allow a flicker free display, it’s normal practice to switch the amplifier outputs and shift voltages at a some high multiple of the scan frequency, so that, if the resolution of each trace were high enough for it to be seen, each trace would consist of a sequence of closely spaced dots.
The normal 'real-time' oscilloscope is only able to display the character of a signal, in a given volt age/time sequence, at the moment it happens, and only then if it’s accurately repetitive -- so that each successive X direction scan repeats and reinforces its predecessors. However, there are instances where the signal which is of interest is sporadic or randomly occurring, and it’s wished to examine in detail this particular event, recorded, briefly perhaps, on an externally triggered trace.
In early 'scopes, the only available technique was to use a 'long-persistence' screen phosphor, so that the image of the trace would only fade away fairly slowly.
An improvement on this system is to use a type of 'scope design in which one layer of a two layer screen is coated with a phosphor which has a fluorescence threshold somewhat above the energy level provided by the 'writing' beam. The image of the last scan pattern can then be recovered by flooding the screen with electrons from a so-called 'flood' electron gun, and this image will persist, at high brilliance, until the flood gun is switched off , and normal writing continues.
However a whole new family of 'scopes has arisen, to make use of the ability of fast analogue to digital (A-D) converters and high storage capacity digital memory techniques to store and recover digitally en coded signals. Using this technique, an incoming ?' axis voltage signal is sampled, at a repetition frequency high enough to give, say, 500 or more voltage sampling points across the X axis scan. This series of discrete signal voltage levels can be stored in random access memory (RAM), cells, from which the digital information can be recovered in its proper time sequence, and decoded, once more, by a digital to analogue. (D-A), converter, to generate an accurate replica of the signal present on the scope at the time when the signal was stored.
The great advantage which this technique offers, by comparison with long-persistence phosphors, or flood gun systems, is that, depending on the resolution of the stored signal -- which depends on the number of volt age sampling points, in the Y axis provided by the A-D and D-A converters, and on the number of time sampling points in the X axis -- the recovered signal can be examined in greater detail by expanding various points of the vertical and horizontal image scale. Although the shape of the signal will be preserved, the time axis is now an arbitrary one, reconstructed from pulses derived from the system clock.
Markers indicating the voltage and time characteristics of the original waveform may be superimposed on the recovered signal to simulate the characteristics of a real-time signal.
A schematic layout of a digital storage oscilloscope system is shown in FIG. 40.
FIG. 40 Schematic layout of digital storage oscilloscope
Although an approximate indication of signal frequency can be obtained from a Lissajous figure, on an oscilloscope display, this is an awkward measurement to make, and is usually time consuming. For high frequencies, the standby of the 'amateur' transmitter enthusiast was the 'grid dip' meter, or its transistorized variants, such as that shown in FIG. 41. In this, the current drawn from the power supply by a low power oscillator is reduced when the oscillator frequency is adjusted to coincide with that of the unknown signal, when this signal is inductively are capacitatively coupled to the dip-meter oscillator coil.
This, also, is only as accurate as the calibration of the dip-meter tuned circuit.
FIG. 41 Grid dip frequency meter.
A much better system, and one which is now almost universally used, is the digital display 'frequency counter'. This operates in the way shown, schematically, in FIG. 42. In this, the incoming signal, after 'conditioning' to convert it into a clean rectangular pattern pulse train, of adequate amplitude, is passed through an electronic switch (gating) circuit, which feeds those pulses received during a precise time interval on to a digital pulse counter system. For convenience in use, the final count is switched to the display through a storage register, to allow the counter to be reset between each measurement interval. The accuracy obtainable is determined by the precision of the setting of the frequency of the crystal oscillator which controls the timing cycle, and by the number of pulses which are allowed through the gating switch, which, in turn, depends on the input frequency. At 1MHz, a gate duration of 0.01 seconds would allow an accuracy of one part in ten thousand, but at 1000Hz, a gating interval, and display refresh periodicity, of once every ten seconds would be needed to achieve the same accuracy.
For high precision frequency measurements where very protracted counting periods would otherwise be necessary, 'frequency difference' meters are usually employed. In these instruments, the difference between the incoming signal frequency, and that of a heterodyne signal derived from an accurate reference frequency is determined. This technique allows accurate frequency measurements to be made in greatly reduced time intervals.
Input 1; Signal conditioner; Xtal Clock Gate Time interval Control logic; Counter MM Store Display
FIG. 42 Schematic layout of frequency counter
The majority of the tests carried out on AC amplifying or signal manipulation circuits require the use of some standardized input signal, in order to determine the waveform shape or amplitude changes brought about by the circuit under test. These input signals are derived from laboratory 'signal generators', and these are available in a range of forms. In low frequency sinewave generators, the major requirements are usually stability of frequency, purity of waveform, and freedom from output amplitude 'bounce', on changes of frequency setting, with the accuracy of frequency or output voltage being relatively less important.
In HF and RF signal generators, precision and stability in the output frequency are essential, as is usually the accuracy of calibration of the output volt age -- particularly where the instrument is to be used to determine receiver sensitivity or signal to noise ratio. The purity of the output waveform is usually less important provided that the magnitudes of the output signal harmonics, or other spurious signals, are sufficiently small.
RF signal generators usually make provision for amplitude modulation of the signal, either by an internal oscillator or by an external signal source. This facility is useful to determine the performance and linearity of the demodulator circuit of RF receivers, and may also be helpful in aligning RF tuned circuits.
Other test signals, of square-wave or triangular waveform, are also valuable for test purposes, and these are usually synthesized by specialized waveform generator ICs, and available as wide frequency range 'function generators'. These offer square and triangular output waveforms, as well as a sinewave signal of moderate purity, usually in the range 0.1-0.5% THD. They do have the advantage that they allow a rapid frequency sweep to be made, to check the uniformity of frequency response of the circuit under test.
These instruments derive a relatively high purity, wide frequency range, output signal by heterodyning two internally generated RF signals, as shown in FIG. 43, and then filtering the output from the mixer stage so that only the difference frequency remains.
The major drawback of this technique is that any relative instability in the frequency of either of the RF signals, from which the heterodyne output signal is derived, will be proportionately greater in the resulting difference frequency signal.
FIG. 43 Layout of high frequency BFO (see FIG11)
Frequency modulated oscillators (wobbulators)
These types of signal generators are almost exclusively used in RF applications, such as in determining the transmission characteristics of the tuned circuits or filter systems which define the selectivity of the receiver, and in aligning these receivers. If the linearity of the modulation characteristics of the oscillator are high enough, it will also allow the distortion characteristics of an FM receiver or demodulator to be measured.
Prev. ------- Next
Home | Glossary | Books | Links/Resources
|EMC Testing | Environmental
Updated: Wednesday, 2016-08-10 21:17 PST