PREAMPLIFIERS [Preamplifier and Filter Circuits]

Operational amplifier techniques seem to dominate modern preamplifier design. Even where preamplifiers are not built around true operational amplifier integrated circuits, they are usually based on specialist audio chips that are fundamentally just operational amplifiers plus some built-in biasing and feedback components. Preamplifiers based on discrete circuitry are something of a rarity these days, but even these tend to be based on what are basically operational amplifier style circuits, complete with differential inputs.

This is really not all that surprising. Operational amplifier techniques make it very easy to set any desired input impedance and voltage gain figures. Unlike many other types of audio amplifier, these figures can be set quite accurately.

Using 1% resistors, which are now available quite cheaply, the error in the input impedance would be no more than 1%. The maximum error in the voltage gain would be no more than 2%. This degree of precision is not available with most other types of audio circuit, where the wide variations in the current gains of the transistors tend to give rather "hit and miss" results.

Feedback

Operational amplifiers were originally designed as d.c. amplifiers in analogue computers. They were used to perform mathematical operations, and it is from this that their name was derived. It is unlikely that many of the operational amplifiers currently produced end up in analogue computers.

Computers of this type are mainly in the "museum piece" category these days. The precision they can provide though, makes them well suited to many applications, including audio, oscillator, and control applications.

These devices were designed to operate with dual balanced supplies so that they could provide positive or negative results.

In some applications they are still used in this manner, but in audio applications they are mainly used with a more

Fig. 1.1 The basic inverting amplifier configuration conventional single supply plus a bias circuit. There are two basic modes of operation, which are the inverting and non-inverting modes. As their names imply, one provides an inversion of the signal while the other has the input and output signals in-phase. The basic inverting and non-inverting amplifiers are shown in Figures 1.1 and 1.2 respectively.

These are the normal audio style amplifiers, having a single supply rail and bias circuits.

The basic action of an operational amplifier is very simple indeed. It is a very high gain differential amplifier. In other words, it amplifies the voltage difference across its two inputs, but due to the very high voltage gain only a minute voltage difference is needed in order to send the output fully positive or negative. The output goes positive if the non-inverting (+) input is at a higher voltage than the inverting (-) input, or negative ...

Fig. 1.2 The basic non-inverting amplifier configuration.

... if the relative states of the two inputs are reversed. In theory an operational amplifier has infinite voltage gain, but this is obviously not achieved in practice. However, at d.c. the voltage gain is typically around 200,000 times. Note that with practical operational amplifiers the voltage gain is rolled-off at frequencies of (typically) more than about 10 KHz or so.

When using these devices in audio amplifiers it has to be borne in mind that their voltage gains at high audio frequencies are often quite modest.

Such high and frequency-dependent voltage gain would seem to render operational amplifiers of little use for audio applications. The secret of success is to tame the innate (open loop) voltage gain of the amplifier by applying negative feedback. This gives a much lower gain to the overall circuit (the closed loop gain), and a frequency response that is essentially flat up to the point where the gain of the device is inadequate to provide the required gain figure. It also reduces the noise and distortion to reasonable levels. In fact the noise and distortion levels are very low indeed for some specialist audio types such as the NE5534A.

If we consider operation of the inverting amplifier first (Fig. 1.1), R2 and R3 bias the non-inverting input to approximately half the supply voltage. C2 is not strictly necessary, but is normally included. One function it provides is to remove any stray pick-up at the non-inverting input. In most cases this would not be high enough to produce a significant amount of "hum" (or whatever) on the output, but with high gain circuits it could lead to instability. Another function of this capacitor is to decouple any "hum" on the supply rails, and to decouple any feedback due to loading of the supply.

Operational amplifiers have quite high supply ripple rejection when used with dual balanced supplies, but lend to be more vulnerable when powered from a single supply rail.

C1 and C3 simply provide d.c. blocking at the input and output of the circuit. Their polarities are the ones normally used, but one or both might need to be altered to suit some sources and loads. If in doubt you can always use a test meter to check the polarity of the d.c. voltage across a capacitor to see if it matches the component's polarity. With a high input impedance or load impedance a non-electrolytic type can be utilized. As a "rule of thumb", a value of about 2u2 is suitable for a coupling capacitor that is feeding into an impedance of 10k. For higher impedances the values of the capacitor should be reduced proportionately - for lower impedances it is increased proportionately. As a couple of examples, a value of 4u7 would be suitable for a 4k7 load, while a value of 470n (0u47) would be satisfactory for a 47k load.

R1 and R4 are the negative feedback resistors. These set the closed loop voltage gain of the amplifier (i.e. its actual voltage gain). The basic action of the circuit is to stabilize the inverting input at the same voltage as the non-inverting input. If the output should drift higher in voltage, the coupling through R4 unbalances the inputs, and sends the output more negative. Any negative drift unbalances the input voltages in the opposite direction, and sends the output more positive.

An input voltage also results in the inputs becoming unbalanced, and the feedback action again corrects this. If we assume that R1 and R4 have the same value, taking the input one volt positive results in the output going one volt negative.

A simple potential divider action across these two resistors then gives the required voltage at the inverting input. If R4 is made higher in value, then the output voltage must be greater in order to maintain the balance at the inputs. In other words, making R4 higher in value boosts the voltage gain of the circuit. In fact the gain of the circuit is merely equal to the value of R4 divided by that of R1.

What is termed a "virtual earth" is formed at the inverting input. In the d.c. coupled version of this configuration the non-inverting input is biased to the central 0 volt earth rail, and the feedback therefore maintains the inverting input at the earth voltage. As far as the current flow through R1 is concerned, it is much the same as if the end that connects to the inverting input really was connected to earth. Hence the term "virtual earth". With the single supply configuration the inverting input is stabilized at the half supply voltage bias level.

However, with an a.c. audio input signal (such as an audio signal), the current flow through R1 is again much the same as if it was actually connected to earth.

Calculations

The practical importance of this is that the input impedance of the circuit is equal to whatever value is given to R1. This makes it very easy indeed to set the required input impedance and voltage gain figures for the circuit. The value of R1 is simply made equal to the required input impedance. The correct value for R4 is obtained by multiplying the value of R1 by the required voltage gain. For example, assume that the amplifier must have an input impedance of 50k and a voltage gain of twenty times. The correct value for R1 is 50k. This is not a preferred value, so it must either be made up from two resistors in series or parallel, or the nearest preferred value could be used. In most cases the exact input impedance of a circuit is not critical, and a value of 51k or 47k would almost certainly be perfectly satisfactory.

The correct value for R4 is 50k multiplied by twenty, which is obviously 1 megohm (1000k). There is no problem here as this is a preferred value. In most audio applications the precise voltage gain of a circuit is not critical, and it is then quite acceptable to choose the nearest value above the calculated figure where the latter is not a preferred value.

With a theoretical operational amplifier you can set any desired input impedance and voltage gain. In practice there are definite limitations though. As already pointed out, a real operational amplifier has a massive voltage gain, but only at d.c. and very low frequencies. In audio applications it is generally not possible to obtain a high voltage gain from a single operational amplifier, since its open loop voltage gain will simply not be high enough at the upper end of the audio range.

The important parameter in this context is the unity gain bandwidth of the operational amplifier. This is 1MHz for the industry standard uA74IC and its many direct equivalents.

Some more recent devices have higher figures, but apart from a few "specials" there are few devices which do much better than about 3MHz to 4MHz. The significance of this parameter is that you can obtain the open loop voltage gain of the device at audio frequencies by dividing the unity gain bandwidth figure by the frequencies in question. Therefore, the uA74IC has a voltage gain of only fifty times at the top audio frequency of 20kHz (1MHz [1000kHz] divided by 20kHz equals 50).

More recent types, particularly the popular f.e.t input types, achieve a more useful gain at 20kHz of around 150 to 225 times. Even so, in many audio applications two stages of amplification are needed in order to obtain the required voltage gain. Bear in mind that you are unlikely to get the best from an operational amplifier if you use it at its limits.

A two-stage circuit with the amplifiers used well within their maximum gain figures is likely to give much better results.

It should perhaps be pointed out that we are talking here about so-called internally compensated operational amplifiers.

These are devices which have an internal capacitor which rolls-off their high frequency response. This capacitor ensures that the amplifier is stable at any gain level, right down to unity gain. Instability can be caused by stray capacitance, etc., causing phase shifts in the feedback network. These can result in the negative feedback being inverted into the positive variety, possibly causing oscillation. The compensation capacitor ensures that gain at the frequency where this inversion occurs is too low to produce oscillation.

It is important to realize that a compensation capacitor does no more than provide a circuit that is fundamentally stable. It will not necessarily ensure good stability with a circuit that has a poorly designed component layout that encourages stray feedback.

Some operational amplifiers do not have a compensation capacitor, and require a discrete capacitor or even several components in order to prevent them from becoming unstable in use. The salient point here is that the higher the closed loop voltage gain of the circuit, the lower the amount of roll-off that is required. This enables higher gain at higher frequencies to be obtained using one of these externally compensated devices.

Although the difference in performance can be quite high, externally compensated operational amplifiers do not seem to be particularly popular. The uA748C is the externally compensated version of the uA74IC, and at high voltage gains this can achieve something like ten times the gain-bandwidth product of the uA741C. This enables quite high voltage gains to be obtained from a single device while retaining the full audio bandwidth, but for many purposes the noise and distortion performance would not be good enough. As pointed out previously, in audio applications two low gain stages generally offer better performance than one stage operating flat-out.

Non-Inverting

The non-inverting circuit of Figure 1.2 is similar in operation to the inverting circuit. As before, R1 and R2 bias the non-inverting input to about half the supply voltage. In this case though, no decoupling capacitor can be used because the input signal is coupled to the non-inverting input. This changes the mathematics somewhat, since the input impedance of the circuit is not governed by a resistor in the feedback network.

Instead, it is equal to the parallel impedance of R1 and R2.

As these resistors will have the same value in most cases, this means that the input impedance of the circuit is equal to half this value. In other words, simply make R1 and R2 equal to double the required input impedance.

This obviously assumes that the operational amplifier has an infinite input impedance, and that it will not reduce the input impedance of the circuit. In practice the input impedance of an operational amplifier will be finite, but too high to be of significance. At least, it will at low frequencies. It is worth keeping in mind that the small amount of input capacitance can drastically reduce the input impedance of a circuit at high audio frequencies. This is a factor that is common to practically all audio amplifiers, and one which can easily be ignored in practice. It tends to be of more significance in test gear and instrumentation applications.

The voltage gain of the circuit is again controlled by a two resistor negative feedback circuit (R3 and R4). At d.c. there is 100% negative feedback through R4, because C2 has an infinite resistance, and effectively cuts R3 out of the circuit.

This gives unity voltage gain, and results in the output of the circuit being biased to the required level of half the supply voltage. This permits high output levels to be accommodated without the signal becoming clipped. At audio frequencies C2 has a low impedance in comparison to R3, and this reduces the amount of feedback applied to the circuit. Accordingly, the closed loop voltage gain of the amplifier is increased.

The method of calculating the voltage gain is slightly different to the method used for the inverting configuration. First add the values of R3 and R4, and then divide this figure by the value of R3. For example, if the values of R3 and R4 are respectively 10k and 100k, the voltage gain is eleven times (100k plus 10k equals 110k, 110k divided by 10k equals 11). Suitable values for R3 and R4 can be determined by first setting a likely value for R3. Something in the region of 10k is usually satisfactory, but a lower value may be necessary where high voltage gain is required. The correct value for R4 is then equal to the value of R3 multiplied by one less than the desired voltage gain. For example, for a voltage gain of twenty-five times and with R3 at a value of 10k, R4 would need to have a value of 240k (10k multiplied by (25-1) equals 240k).

When designing circuits using operational amplifiers you need to be careful when a combination of high gain and high input impedance are required. The first problem is that you may find that impractical resistor values are required. It is quite easy to end up with a circuit that requires a feedback resistor of around 100 megohms in value! Where possible it is better if feedback resistors of even a few megohms can be avoided. These leave the circuit's frequency response open to fluctuations caused by quite small stray capacitances.

Even where suitable resistors for the required gain and input impedance are a practical proposition, high gain and input impedance in a single stage are best avoided. There can be extreme difficulties with instability due to stray feedback, especially with non-inverting amplifiers (which is the mode that would probably have to be used). It is generally better to adopt a two-stage approach. The first stage provides the high input impedance together with a modest voltage gain, while the second stage provides the "lions share" of the voltage gain.

It is often necessary to incorporate a "hum" filter in the bias circuit when using the non-inverting mode. Figure 1.3 shows a non-inverting circuit which includes this facility.

R5 and C4 are the "hum" filter, and these simply form a basic lowpass filter in the supply to the bias circuit. In other respects the circuit is exactly as before. Ideally the time constant of these two components should be quite long so that they provide good attenuation of the noise. This is especially important if there is a lot of low frequency noise, such as mains "hum". In general, results will be satisfactory if multiplying the value of R5 (in kilohms) by the value of C4 (in microfarads) gives an answer of around 200 to 500. For example, values of 3k3 and 100u should be suitable (3.3 x 100 = 330). With this type of thing though, you often have to take a "suck it and see" approach in order to find out what will give satisfactory results.

Fig. 1.3 Adding a "hum" filter to a non-inverting amplifier

Remember that any noise which reaches the input of the amplifier will then be subjected to the full amplification of the circuit. With a high gain circuit any noise which reaches the input must be kept to an extremely low level indeed if a low output noise level is to be achieved. This generally requires both a low ripple level and general noise content on the supply rail, plus a "hum" filter.

Also bear in mind that the resistor in the filter circuit alters the biasing of the circuit. Consequently, the value of this resistor must be low in comparison to that of R1 in order to minimize the error. Alternatively, the value of R1 can be reduced to compensate for the extra resistance provided by R5.

As a simple design example for a non-inverting circuit, suppose that an input impedance of about 10k plus a voltage...

Fig. 1.4 The design example non inverting amplifier circuit. This version uses E12 series values.

...gain of about 26dB (twenty times) is required. We will also assume that "hum" filtering is required. Figure 1.4 shows an amplifier circuit which meets these requirements.

With an input impedance of 10k, the bias resistors must have a value of double this figure, or some 20k for each one.

This is a preferred value in the E24 series, but 22k resistors from the E12 series will give an input impedance of 11k which should be close enough for most purposes. R1 and C2 act as the "hum" filter, and have a long enough time constant to give good low frequency attenuation. I suppose that if you require things optimized as far as possible, R2 could be reduced to 18k. This would largely remove the slight offsetting of the biasing caused by R1, and would reduce the input impedance from Ilk to a slightly more accurate 9k9.

Whether or not this would make a noticeable improvement in results is open to doubt though.

The voltage gain of the circuit is set by R4 and R5 at 23 times. With a value of 10k for R4, the calculated value for R5 with a voltage gain of twenty times is 190k. A value of 200k in the E24 series is the nearest preferred value above the calculated figure, or a 220k component from the ordinary El2 series can be used. Values of 2u2 for C1 and C3 are appropriate for the input impedance of 10k, and the same value for R4. The impedance into which the circuit will feed is frequently an unknown quantity, as preamplifiers are often designed for use with many different power amplifiers (or whatever) having a wide variety of input impedances. In such cases I generally use a value of 10u for the coupling capacitor as this suits input impedances down to a couple of kilohms, and few amplifiers, etc., have an input impedance lower than this.

The operational amplifier used in this circuit is the standard uA74IC, which is adequate for an audio amplifier having a voltage gain of only about 20 times or so. However, if low noise and (or) distortion is important, a more modern device such as the LF351N would be a better choice.

Where the voltage gain is high enough to dictate the use of a two-stage circuit, the two stages can be coupled together via a capacitor. In most cases though, there is no problem in simply coupling the output of the first stage to the input of the second stage. Apart from avoiding the need for a coupling capacitor, this usually enables the bias components for the second stage to be omitted as well. A few examples of d.c. coupled two-stage circuits are to be found further on in this section.

Guitar Preamplifier

Anyone who designs electronic gadgets for use with electric guitars soon learns that there is a major problem. This is simply the difference in the output levels from various guitar pick-ups. The home-made guitar pick-ups that were popular some years ago, plus possibly a few older commercially produced types, are low impedance devices which have quite low output levels. We are talking here in terms of an output signal that is only a few millivolts r.m.s. after the initial peak has subsided. Cheaper ready-made pick-ups have similar characteristics, but generally have slightly higher impedances and output levels. The more expensive units have medium output impedances and much higher output levels. In fact some guitar pick-ups apparently have output levels of around 2 volts r.m.s. after the initial peak. This is around one hundred to one thousand times higher than that of the lowest output pick-ups.

Many pieces of electronic equipment that are designed for operation with electric guitars get around this problem by having a control (possibly an internal preset) that can be set to suit practically any guitar pick-up. Others are simply designed to be able to accommodate a wide range of input levels. Despite this, users of low level guitar pick-ups can find themselves with effects units and (or) amplification equipment which will not work properly with their guitar. The opposite problem is also possible, but seems to be rare in practice. Units that are designed for only one type of guitar pick-up invariably seem to be aimed at the higher output types. If overloading should occur, a simple attenuator will solve the problem. A series resistor added in the guitar lead, possibly mounted in one of the plugs, will often suffice. Where an inadequate drive level is the problem, a simple preamplifier is required.

A suitable preamplifier circuit for this application is shown in Figure 1.5. The amount of extra gain required is not normally massive, and a single stage amplifier is therefore sufficient. This is a simple non-inverting mode circuit having an input impedance of 50k. This should suit most guitar pick-ups, but the values of C1, R1, and R2 can be altered to give a different input impedance if necessary.

The voltage gain can be varied from about four times to approximately 51 times by means of VR1. In practice this preset resistor is adjusted for the lowest gain that gives satisfactory results. The LF351N specified for the IC1 position has a gain bandwidth product of 4MHz, which is more than adequate for the full audio bandwidth (20kHz) at maximum gain. Similar devices such as the TL07ICP and TL08ICP will also work well in this circuit, and in the other circuits in this guide where an LF351N is specified.

Fig. 1.5 The guitar preamplifier circuit. VR1 is a preset gain control

Although the supply voltage for this circuit, and most of the others in this guide, is given as 12 volts, this should be regarded as a minimum figure for good results. A 9 volt battery supply will probably give adequate results with all the circuits, but gives less "headroom". Output levels of around 3 to 4 volts r.m.s. can be handled using a 12 volt supply, but with a 9 volt supply the maximum figure is likely to be under 2 volts rms. The importance of this clearly depends on the sort of output levels that will occur, and a 9 volt supply may be adequate in many cases.

Where battery operation and more headroom are required, using two batteries in series to give an 18-volt supply will probably be the simplest solution. This enables output levels of well over 5 volts r.m.s. to be achieved without clipping or serious distortion occurring. The absolute maximum operating voltage for the LF351N (and most other operational amplifiers) is 36 volts, although the maximum recommended supply voltage is lower at 30 volts. The current consumption of the guitar amplifier is only about 2 milliamps incidentally, which means that the circuit can be powered economically from one or two small 9-volt (PP3 size) batteries. No "hum" filter is included in this circuit, or any of the others in this guide.

Obviously such a filter should be added if the supply is not a "clean" type, such as a battery or highly smoothed mains derived supply.

Construction

Construction of this circuit should not present any major problems as it is so simple. A small piece of stripboard will accommodate everything, and there is no real need to resort to a custom printed circuit design. However, bear in mind that the circuit has a fairly high input impedance, moderately high voltage gain with VR1 at maximum resistance, and that the input and output are in-phase. Stray feedback from the output to the input can easily cause instability, and the component layout needs to be carefully designed. As explained previously, VR1 is adjusted for the lowest gain (i.e. set at the lowest resistance) that gives satisfactory results, and this is just a matter of using a bit of trial and error.

With a very low output pick-up the voltage gain of the circuit might be inadequate. If only a small boost in gain is required it is just a matter of making R3 higher in value. The output levels from a small minority of pick-ups are so low that this will not cure the problem, and for these it is necessary to use a two-stage amplifier. Using an input designed for a high impedance microphone usually gives good results, or the high impedance microphone amplifier described later in this section can be used between the guitar and a high level input.

In the components list for this project, and the others in this guide, the resistors are specified as 5% carbon film types.

This is a minimum requirement, and superior resistors such as 1% metal film types are also suitable. In fact there is some advantage in using metal film resistors in that they have lower noise levels. This could give a noticeable improvement in performance, particularly with resistors that appear at the input of high gain circuits. There is also some slight advantage in using close tolerance resistors, as they give more accurate bias levels, gain levels, etc. However, in most cases using something better than ordinary 5% tolerance carbon film resistors will not give any obvious improvement in performance. For good noise performance VR1 should be a good quality preset resistor.

Components for Figure 1.5

Resistors (all 0.25 watt 5% carbon film)

100k

100k R1 R2 R3 15k R4 4k7

Potentiometer VR1 220k sub-min preset

Capacitors 470n polyester

Apl 63V elect 10u 50V elect C1 C2 C3

Semiconductor LF351N or similar IC1

Miscellaneous

JK1 standard jack socket standard jack socket 8 pin d.i.l. holder, circuit board, case, etc.

JK2

Microphone Preamplifier

Figure 1.6

One characteristic that is common to practically all microphones is that they have an extremely low output level. Off hand, the only form of microphone I can think of which has a reasonably high output level is the carbon type, as used in old style (non-electronic) telephone handsets. This type of microphone is little used these days though, and we will not consider them further here.

Probably the most popular types of microphone currently are the high impedance dynamic type, and electret types which have a built-in step-up transformer. These have similar output characteristics, which are an output impedance of around 10k to 50k, and a typical output level of around 2mV to 5mV r.m.s. Figure 1.6 shows the circuit diagram for a preamplifier for use with microphones of either type.

This is basically just an inverting amplifier (IC1) driving a non-inverting amplifier (IC2) via a volume control style variable gain control (VR1). R1 sets the input impedance of the circuit at 22k, which is a good match for most high impedance microphones. The voltage gain of IC1 is ten times, and that of IC2 is a little in excess of 22 times. This gives a total gain of just over 220 times. This is adequate to permit most high impedance microphones to drive most high level inputs. If the gain of the amplifier proves to be inadequate though, increasing R4 to around 470k in value should cure the problem.

Normally VR1 is left at maximum gain, but it might be necessary to back it off slightly if the microphone is subjected to high sound levels. Someone singing loudly into a micro

phone at short range for example, can produce a much higher than normal output level. This could result in overloading of IC2 or the unit fed from the preamplifier unless VR1 is backed-off slightly.

The NE5534A specified for IC1 is a special very low noise and distortion device. It is internally compensated, but only for closed loop voltage gains of ten or more. Obviously in this case the internal compensation will suffice, and external compensation components are not required. It is not essential to use such a high quality operational amplifier for IC1, and the circuit will work using an LF351N or even a pA74IC. As this would give a noise level as much as ten times higher than that provided by the NE5534A, 1 would not recommend using a cheaper device unless the unit is to be used in a fairly low- fidelity setup.

Although both operational amplifiers have internal compensation, this does not guarantee that the circuit as a whole will be stable. To aid good stability, C7 is used to provide extra feedback over IC2 at high frequencies above the upper limit of the audio range. This provides additional high frequency roll-off. Also, the circuit has been designed so that the input and output of the circuit are out-of-phase. This still does not guarantee freedom from instability as there are still parts of the circuit that are in-phase with other parts of the circuit.

You should also bear in mind that phase shifts can occur through the stray capacitances and inductances that are responsible for stray feedback. This can turn out-of-phase feed

back into in-phase feedback that can cause high frequency oscillation. The circuit can be built satisfactorily on strip- board, but reasonable care has to be taken with the layout.

Having a copper strip carrying an output signal running alongside a strip carrying an input signal is not a good idea! Another problem to keep in mind when dealing with any high gain audio circuit is that of stray pick-up of mains "hum", etc. The microphone should have a good quality screened lead that will keep stray pick-up to an insignificant level. The lead that connects JK.1 to the main circuit board should also be a screened type. Alternatively, if the project is housed in a case of all-metal construction, provided this is earthed to the negative supply rail it will provide overall screening of the unit. If the lead to JK1 is kept quite short, it is then unnecessary to use a screened lead. With any sensitive audio circuit I would recommend using a metal case earthed to the negative supply rail so that the circuit board, etc., are screened. Diecast aluminum boxes are ideal for this sort of application, but any case of all-metal construction will probably give the desired effect.

Components for Figure 1.6

Resistors (all 0.25 watt 5% carbon film)

R1 22k R2 10k R3 10k R4 220k R5 47k

R6 47k R7 4k7 R8 100k

Potentiometer

VR1 22k log carbon

Capacitors 100m 25V elect Ip 63V elect 4u7 63V elect 2p2 63V elect Ip 63V elect Apl 63V elect 27p ceramic plate 10m 50V elect

C1 C2 C3 C4 C5 C6 C7 C8

Semiconductors

IC1 NE5534A LF351N IC2

Miscellaneous

JK1 standard jack socket standard jack socket

Two 8 pin d j.l. i.c. holders, circuit board, case, JK2 etc.

Low Z Microphone Preamplifier

Some high impedance dynamic microphones genuinely have high impedance coils, but many of them seem to be low impedance types with a built-in step-up transformer. Similarly, many electret microphones achieve a higher output level and higher output impedance by using an integral step-up transformer. Some dynamic and electret microphones actually offer two output impedances and output levels. These are usually the direct output and the output via the step-up transformer, although in some cases the "direct" output is actually obtained from a low impedance tapping on the transformer.

In theory there is no point in using a step-up transformer as it simply introduces signal losses. In practice it is also likely to produce some distortion and loss of frequency response.

The advantage of using a step-up transformer is that it provides an output signal that is a better match for semiconductor amplifiers. Normal transistors are not at their best with low source impedances and very low signal levels. Better performance is normally obtained with higher signal levels and source impedances.

On the other hand, if a circuit that operates well with the direct signal can be produced, it does offer a potentially higher overall performance. The preamplifier circuit of Figure 1.7 is for a low impedance dynamic microphone, or one having similar output characteristics. This basically means electret microphones without a built-in step-up transformer, plus one or two unusual types which you are unlikely to encounter (certain ribbon type microphones for example).

This circuit is basically just a slight reworking of the previous circuit. The only difference is in the feedback circuit for IC1 where the values have been selected to give a higher voltage gain of 40dB (100 times) and a lower input impedance of 680 ohms. The output level from a low impedance microphone is only about one-tenth of that from a high impedance type, making it necessary to boost the gain of the input stage by a factor of ten in order to compensate for this.

The lower source impedance makes it acceptable to use a lower input impedance, which gives a lower noise level from IC1.

Fig. 1.7

Traditional transistor theory dictates that the noise level drops as the source impedance is reduced, until a certain ideal source impedance is reached. Decreasing the source impedance further then produces a slight increase in the noise level.

My experience, and that of others, would suggest that this is not the case, and that the lower the source impedance, the lower the noise level obtained. At least, this is the case using bipolar transistors, or an integrated circuit such as the NE5534A used for IC1 which has a bipolar input stage. With f.e.t. input circuits the source impedance has little effect on the noise level.

The practical result of all this is that the low source impedance helps to give a low noise level from IC1, and to some extent compensates for the higher noise level caused by the boost in its voltage gain. Together with the fact that IC1 is a type which has a very low noise level anyway, this produces a very acceptable noise performance from the circuit. In fact a signal-to-noise ratio of around 80dB might be achieved, but the precise figure depends on the output level of the microphone, the noise performance of the particular device used for IC1, etc.

Many low impedance dynamic microphones are very much in the "cheap and cheerful" category, and do not provide a very high level of performance. If you are using one of these in an application which is something less than hi-fi, you might prefer to use a cheaper operational amplifier for IC1. Due to the low input impedance, a f.e.t. input type such as the LF351N is not likely to provide very good results. In this instance an ordinary uA74IC will provide superior noise performance, but with somewhat compromised high frequency performance. Probably the best low cost option is a uM748C (the non-internally compensated uA74IC) with an external compensation capacitor of about 4p7 connected between pins 1 and 8. The noise level will probably be about 20dB worse (i.e. ten times higher) than that obtained from an NE5534A though.

Like the previous circuit, this one has high gain and is therefore vulnerable to stray pick-up of mains "hum", etc.

Accordingly, it should be housed in a metal case to provide overall screening, and due care should be exercised when designing the component layout.

Components for Figure 1. 7

Resistors (all 0.25 watt 5% carbon film)

680R R1 R2 10k R3 10k R4 68k R5 47k R6 47k

R7 4k 7 R8 100k

Potentiometer VR1 22k log carbon

Capacitors

C1 100m 25V elect 22m 50V elect 4m7 63V elect 2p2 63V elect 1M 63V elect 4m7 63V elect 27p ceramic plate 10m 50V elect

C2 C3 C4 C5 C6 C7 C8

Semiconductors

IC1 NE5534A LF351N IC2

Miscellaneous JK1 standard jack socket standard jack socket

Two 8 pin d.i.l. i.c. holders, circuit board, case, JK2 etc.

Crystal Microphone Preamplifier

Although they were all the rage about twenty years or so ago, crystal microphones seem to have steadily fallen from favor over the intervening years. They have been ousted by dynamic and electret types. Cheap dynamic types now tend to be used where inexpensive crystal microphones might once have been used, while electret microphones are now used in place of the higher quality crystal types. Probably the main cause of this decline is the fact that crystal microphones have output characteristics that are fine for use with valve circuits, but which are less well suited to bipolar transistor circuits. As valves were used less and less in audio equipment, crystal microphones became less and less popular.

Crystal microphones are still available today, mainly in the form of inexpensive lapel microphones. Also, with large numbers of them having been sold over many years, there must still be large numbers of them in circulation. Where a rugged and moderately good quality microphone is required they still represent a good choice. Although they are not ideally suited to use with old style bipolar transistor circuits, they actually operate well with modern semiconductor components. In particular, they work well with f.e.t.s and f.e.t. input operational amplifiers.

The output level from crystal microphones is quite high (by microphone standards anyway). An output of around 10 milli-volts r.m.s. is generally produced by crystal lapel microphones, with the higher quality "stick" types usually producing a slightly lower output level. This is not quite as good as it might at first appear, because the output signal is at a very high impedance. 1 suppose that this is not strictly true, and it would be more accurate to say that a crystal microphone must feed into a very high load impedance.

A microphone of this type effectively consists of a low impedance signal source in series with a capacitor (Fig.1.8). The value of this capacitor varies substantially from one microphone to another, but it is never very large. It would typically be around 10 nanofarads. This gives the microphone an output impedance that is almost totally dependent on frequency, and which halves with every doubling of frequency.

The output impedance is quite low at the highest audio frequencies, but is one megohm or more at the low end of the audio range. The built-in capacitance and the input impedance of the preamplifier effectively form a single stage highpass filter. This makes it essential to have a high load impedance, as the microphone otherwise has a frequency response that rises at 6dB per octave over a large part of the audio range! An input impedance of about 1 megohm is usually sufficient to give good results, with a reasonably well extended low frequency response.

Figure 1.9 shows the circuit diagram for a crystal microphone preamplifier. This consists of a unity gain buffer stage at the input, a volume control style variable attenuator, and an inverting amplifier to provide the voltage gain. The specified values for R1 and R2 give an input impedance of 1.1 megohms, which should be satisfactory with all crystal microphones.

However, these values can be raised if necessary, so as to provide a higher input impedance. For example, 3.9 megohm resistors could be used for a microphone that has a recommended load impedance of 2 megohms.

IC2 provides a voltage gain of 40dB (100 times), which should match practically any crystal microphone to any normal high level input. If higher gain should be needed it would be possible to boost the gain slightly by making R6 higher in value. As IC2 is operating close to its maximum practical closed loop gain it is not possible to obtain a large increase in gain from this stage. A negative feedback network could be introduced to IC1 in order to boost its gain, but I would warn against trying to produce a large increase in the gain of this stage. As it has a very high input impedance and its input and output are in-phase, more than a modest amount of voltage gain could easily cause instability. The output levels of crystal microphones are such that nothing more than a modest boost in gain is likely to be needed anyway.

Fig. 1.8 A crystal microphone is effectively a signal source in series with a capacitor

Fig. 1.9

Once again, construction of this unit is basically straightforward, but the high gain makes it mandatory to be careful about the component layout. As with the other microphone preamplifiers, good screening is essential in order to avoid problems with pick up of mains "hum". Although the gain of this preamplifier is somewhat lower than that of the previous two circuits, it is probably more prone to stray pick-up in the input circuitry. This is due to its much higher input impedance, which renders fully effective screening essential.

Components for Figure 1.9

Resistors (all 0.25 watt 5% carbon film)

2M2

2M2 R1 R2 10k R3 47k R4 R5 47k R6 1M

Potentiometer

VR1 22k log carbon

Capacitors C1 1 00u 25V elect 47n polyester 4^7 63V elect 2p2 63V elect 10p 50V elect C2 C3 C4 C5

Semiconductors

IC1 LF351N

LF351N IC2

Miscellaneous

JK1 standard jack socket standard jack socket

Two 8 pin d.i.l. i.c. holders, case, circuit board, JK2 etc.

R.I.A.A. Preamplifier

Fig. 1.10.

Despite predictions by many that it would be a number of years before sales of compact disks overtook those of L.P. gramophone records, it did not really happen that way. In fact L.P.s were soon in third place behind compact cassettes.

Despite their fall in relative popularity, gramophone records are still popular in absolute terms. With large numbers of people having substantial record collections, and much of the program material never likely to appear on compact disks, this medium will remain in use for many years to come.

The preamplifier requirements of a magnetic cartridge are slightly unusual in that they involve doctoring of the frequency response. This is not due to any innate problems with the frequency response of the recording and playback system.

This gives what is essentially a flat response, apart from any slight irregularities caused by imperfections in the equipment used. The tailoring of the preamplifier's frequency response is needed in order to counteract the treble boost and bass cut that are purposely applied to the recorded signal.

The high frequency boost is a form of noise reduction, and one that is used in most recording systems, as well as many radio broadcasting systems. The idea is that by applying some treble boost during the recording process, it is possible to apply complementary treble cut during playback without compromising the performance of the system. Tire overall frequency response is flat, but the treble cut applied during playback helps to reduce "hiss" type noise generated within the recording system. The treble boost is called "preemphasis", and the treble cut is called "de-emphasis". The bass cut during the recording process is needed to avoid excessive groove modulation on strong low frequency signals.

Again, by applying complementary response tailoring during playback a flat overall frequency response is obtained.

Figure 1.10 shows the ideal response for a magnetic cartridge which has standard R.I.A.A. equalization. The response is rolled-off at 6dB per octave above 2120 hertz (i.e. the gain halves above this figure for each doubling of the input frequency). The gain is boosted by 6dB per octave at frequencies below 500 hertz. There is also a requirement for the response to be rolled-off below 20 hertz in order to avoid problems with the low frequency noise which tends to be generated by this recording medium. This roll-off would normally be provided by the coupling capacitors in the system anyway, and does not require any additional components.

In practice the ideal response can never be achieved, since the boost and cut can not be suddenly introduced at the full rate. With complex filters it would be possible to achieve something very close indeed to the required response, but practical R.I.A.A. equalization circuits generally use simple filtering techniques. These introduce some slight errors in the response, but these should not be large enough to noticeably degrade performance.

An R.I.A.A. preamplifier circuit based on operational amplifiers is shown in Figure 1.11. As we shall see later, there is an alternative approach in the form of audio integrated circuits designed specifically for this application.

IC1 is a high quality audio operational amplifier which has suitably low noise and distortion figures for a hi-fi application such as this. It acts as an input stage which provides some of the circuit's voltage gain (just over 20dB of it). It also sets the input impedance at about 47k, which is the optimum figure for most magnetic cartridges. Note though, that some have a recommended load impedance of 100k, and would require R1 and R2 to be raised to 200k each for optimum performance.

Although the specified values for R1 and R2 would seem to give something less than optimum biasing, this is not really the case. Most operational amplifiers have non-symmetrical output stages which can provide output levels much closer to one supply rail than the other. In most cases this means that the output voltage can go within about one volt of the positive supply rail, but can not get within about two volts of the negative supply potential. Thus, a slightly higher bias level than the normal half supply level actually gives slightly improved results with these devices.

IC2 provides further voltage amplification, and also gives the required tailoring of the frequency response. Incidentally, the treble cut applied by IC2 helps to combat the noise produced by IC1. It is therefore better to have the equalization at the output of the circuit rather than at the input. The equalization relies on the fact that the impedance of a capacitor halves with each doubling of frequency. Therefore, if a capacitor is used in the negative feedback of an operational amplifier it can be made to give the required 6dB per octave roll-off rate.

Fig. 11.

In this case we do not require a straightforward 6dB per octave roll-off across the entire audio range. Instead, the filtering must be applied over two bands of frequencies. This is achieved by having two feedback capacitors, with each one connected in parallel with a resistor. These components are R6, R7, C5, and C6. At low frequencies the impedances of C5 and C6 are too high to have any effect on the gain of the circuit, and it is therefore set at roughly 800 times by the three resistors in the feedback circuit.

At frequencies of about 50 hertz or more the impedance of C6 becomes relatively low in comparison to that of R7, and it provides the required rolling off of the frequency response.

However, at frequencies of around 500 hertz its value becomes low relative to that of R6, and it then provides little further rolling-off of the frequency response. At frequencies above 2 kilohertz or so C5 has an impedance which is low in relation to that of R6, and it therefore provides the required high frequency roll-off.

High frequency stability should not be a major problem with this preamplifier, since its high frequency gain is not very high, and the input and output of the circuit are out-of-phase.

Low frequency stability is another matter. The gain of the amplifier is very high at low frequencies, being several thousand times at frequencies of around 20 to 50 hertz. This makes the circuit very vulnerable to pick-up of mains "hum" and other low frequency noise. The circuit in general, but the input wiring in particular, therefore needs to be scrupulously screened.

Another problem with this high gain at low frequencies is that of "hum" and feedback loops. These are caused by the small voltages that are generated across the earth wires due to their minute resistances. These voltages are insignificant if everything is earthed properly, but can (and probably will) cause problems if the earthing is not arranged sensibly. This is a potential problem with any high gain circuit, which includes the microphone preamplifiers described previously.

There is no real problem when the preamplifiers are used in isolation; it is when they are used with a power amplifier that there are potential problems.

There are two basic approaches to successful earthing.

These are called the "spider" and "bus-bar" earthing systems.

In theory the spider earth is the simpler, but it can be difficult to implement properly in practice. It simply has everything earthed to a common point, as in the diagrammatic representation of Figure 1.12. The "bus-bar" earthing system is usually easier to implement in real-life, and this has a wire (or p.c.b. track) going to the earth points in their natural sequence. In other words, first the supply and the loudspeaker are earthed, then the output stage, followed by the driver stage, the pre

amplifier, and the input socket. This arrangement is shown in Figure 1.13, and it really just has things earthed in the same order that the earth connections normally appear on a circuit diagram. What must be avoided at all costs is having the input earthed somewhere between the loudspeaker/output stage and the supply.

-------

Preamplifier Power Supply Output Stage Driver Stage Loudspeaker Input Common Earth Point

Fig. 1.12 The "spider" earth system has one earth point

--------

Preamplifier Output Stage Power Supply Input Driver Stage Loudspeaker Earth Rail

Fig. 1.13 The "bus-bar" method of earthing.

----------

This circuit should work well with most magnetic cartridges.

If overloading should occur at high volume levels, make R5 a little higher in value. A value of 2k2 should be satisfactory.

Conversely, if the output seems to be inadequate, make R5 lower in value. A 680R component should be satisfactory.

It is perhaps worth mentioning that many magnetic cartridges are at their best when loaded with a capacitance of a few hundred picofarads. This is substantially higher than the innate input capacitance of most R.I.A.A. preamplifiers, even when the capacitance of the screened cable is taken into account. Therefore, results might be improved somewhat by adding a capacitor of around 100p to 270p in value across the input of the preamplifier. Of course, the exact effect this will have depends on the particular cartridge used, but it might be worthwhile trying a capacitor of about 150p in value across the input.

Components for Figure 1.11

Resistors (all 0.25 watt 5% carbon film)

91k 100k R1 R2 R3 10k R4 120k R5 1k R6 51k R7 750k

Capacitors

C1 220u 25V elect Ip polyester C2

C3 2u2 63V elect 22u 25V elect ln5 polyester (5% or better)

6n8 polyester (5% or better)

50V elect C4 C5 C6 C7

Semiconductors IC1 NE5534A LF351N IC2

Miscellaneous

JK1 JK2 standard jack socket standard jack socket

Two 8 pin d.i.l. i.c. holders, circuit board, case, etc.

Note that for stereo operation two preamplifiers will be needed, one for each stereo channel. Therefore, two of each component will be required, apart from the case which must be large enough to accommodate two preamplifier boards.

Alternative R.I.A.A. Preamplifier

There are numerous audio preamplifier integrated circuits currently being manufactured, including some that are primarily designed for use in R.I.A.A. preamplifiers. The HA12017 is an example of such an integrated circuit. It has impressive performance figures with a typical total harmonic distortion figure of 0.002% (0.01% maximum), and a typical unweighted output noise voltage of 53 microvolts (90 microvolts maximum). This equates to a signal-to-noise ratio that would normally be between 80 and 100dB.

Figure 1.14 shows the internal arrangement used for this device, and it also shows the pin numbers. The encapsulation is an unusual one incidentally. It is basically an 8-pin s.i.l. (single in line) type, but with pin 2 omitted. If you do not wish to solder it direct to the circuit board, and you can not obtain an 8-pin s.i.l. socket, it will plug into one side of a 16-pin d.d.l. holder. It will also plug into a row of eight "Soldercoh" pins.

Fig. 1.14 The internal arrangement of the HA 12017

Fig. 1.15

There is a differential amplifier at the input, and the device is basically a high gain operational amplifier optimized for audio applications. The differential input stage is followed by driver and output stages. The circuit includes a bias generator stage, but this does not include any input biasing components.

It provides bias currents and voltages for later parts of the amplifier chain. This means that in a practical circuit the device is used much like an operational amplifier.

The circuit diagram for an R.I.A.A. preamplifier based on the HA12017 is shown in Figure 1.15. Although this device could probably be used successfully with a single supply rail, the manufacturers' data only seems to recommend dual supply operation. Therefore, unlike the other circuits in this guide, this one operates from dual balanced 12-volt supplies. The recommended supply voltage is plus and minus 24 volts, but unless high output voltage swings are required it will work perfectly well from dual 12-volt supplies. Note that the absolute maximum supply voltage is plus and minus 26.5 volts, and that the minimum requirement is plus and minus 6 volts.

The supply current is typically 4 milliamps, but can be up to 6 milliamps.

R2 biases the non-inverting input to the central 0 volt supply rail while C1 provides d.c. blocking at the input. The d.c. flow through the cartridge without C1 would be very small, but it is best to avoid any d.c. flow through the coil of the cartridge. C1 also provides protection if there should be a fault. R1 shunts the input impedance of the circuit to 50k.

For an input impedance of 47k use a 91k component, or omit R1 altogether if an input impedance of 100k is required. C2 and C3 aid stability, and also provide the circuit with a more suitable input capacitance.

The negative feedback network has C5, R5, C6, and R6 as the R.I.A.A. equalization network, and R3 as the other arm of the network. C4 provides d.c. blocking in the feedback network, and in theory this component is not required. However, with practical differential amplifiers there tend to be unwanted offset voltages. These are quite small, but are effectively multiplied by the d.c. voltage gain of the circuit. This can produce very large offsets at the output, and could prevent the circuit from operating at all. This is avoided by the inclusion of a d.c. blocking capacitor which gives the circuit a d.c. voltage gain of just one. This capacitor also helps to provide the required roll-off below 20 hertz. C7, C8, and R7 aid high frequency stability, and are what might be regarded as external compensation components.

Some audio preamplifier integrated circuits tend to be fussy about component layouts. However, with the HA 12017 there is no major stability problem, and the component layout is not particularly critical. Bear in mind though, the potential problems of "hum" pick-up and loops that were mentioned previously. The gain of the circuit should be suitable for most magnetic cartridges, but the gain can be boosted somewhat, if necessary, by using a slightly lower value for R3.

Components for Figure 1.15

Resistors (all 0.25 watt 5% carbon film)

100k

100k 220R R1 R2 R3 R4 47k R5 7k5 R6 100k R7 1k8

Capacitors

C1 Ip 63V elect 100p polystyrene 100p polystyrene 220p 25V elect 10n polyester (5% or better) 33n polyester (5% or better) 100p ceramic plate 390p ceramic plate 10m 50V elect 220p 25V elect 220u 25V elect

C2 C3 C4 C5 C6 C7 C8 C9 C10 C11

Semiconductor

IC1 HA12017

Miscellaneous

JK1 standard jack socket standard jack socket 8 pin s.i.l. i.c. holder (see text), circuit board, case, etc.

JK2

Note that for stereo operation two preamplifiers will be needed, one for each stereo channel. Therefore, two of each component will be required, apart from the case which must be large enough to accommodate two preamplifier boards.

The HA 12017 is available from RS outlets, including "Electromail".

Ceramic Cartridge Preamplifier

Crystal and ceramic cartridges were once popular for use in low cost record playing equipment, but in recent years they seem to have been replaced to some extent by low cost magnetic cartridges. They are still far from obsolete though, and there must be very large numbers still in use. They are certainly still used in some new equipment. Incidentally, the only difference between crystal and ceramic cartridges is that one is based on a natural crystal, whereas a piece of man-made ceramic material forms the basis for the other.

They operate on the same principle and have similar output characteristics. They rely on the Piezo effect. This is where a piece of suitable crystal/ceramic material has two electrodes on opposite surfaces. If the unit is twisted or bent, a charge is produced across the two electrodes. The greater the twisting or bending force, the higher the charge voltage. Movement in one direction produces a positive charge opposite direction generates a negative charge. Obviously this effect can be used to convert groove modulations of a record into equivalent electrical signals.

Crystal and ceramic cartridges operate on the same principle as crystal microphones, and have similar output characteristics. This means that they must feed into a high load impedance if they are to achieve a good bass response. The output level is generally much higher than that from a crystal microphone, and is usually somewhere between 100 millivolts r.m.s. and 1 volt r.m.s. at high modulation levels. This means that in order to drive most high level inputs some high impedance buffering is needed, but little voltage amplification is required.

A single stage amplifier such as the one shown in the circuit diagram of Figure 1.16 should be satisfactory. This is basically just an operational amplifier used in the non-inverting mode.

R1 and R2 set the input impedance at about 1.1 megohms.

Their value can obviously be changed in order to give a different input impedance if your cartridge has a recommended load impedance that greatly differs from this figure of 1.1 megohms.

R3 and R4 set the voltage gain at between five and six times, which should be adequate for all but the lowest output pickups. If necessary, the gain can be boosted by making R3 a movement in the

Fig. 1.16 A preamplifier circuit for use with crystal and ceramic pick-ups

little higher in value. It is more likely that a high output pick

up will overload the circuit, in which case R3 should be made lower in value.

This circuit is very simple, and does not have very high voltage gain. This makes the component layout something less than critical. On the other hand, bear in mind that the circuit has a high input impedance, and that the input and output are in-phase. Reasonable care needs to be taken with the component layout in order to avoid problems with mild instability. The high input impedance of the circuit also makes good screening important, as the circuit is vulnerable to stray pick-up of mains "hum", etc.

Components for Figure 1.16

Resistors (all 0.25 watt 5% carbon film)

2M2 R1 R2 2M2 R3 22k R4 4k7

Capacitors

C1 47n polyester

4u7 63V elect 10u 50V elect

C2 C3

Semiconductor IC1 LF351N

Miscellaneous

JK1 standard jack socket standard jack socket 8 pin d.i.l. i.c. holder, circuit board, case, etc. JK2

Note that for stereo operation two preamplifiers will be needed, one for each stereo channel. Therefore, two of each components will be required, apart from the case which must be large enough to accommodate two preamplifier boards.

Figure 1.17

Tape Preamplifier

Do-it-yourself tape recorders are a difficult prospect due to the difficulty of getting everything set up properly. Quite minor errors in a.c. bias levels and other factors can seriously degrade performance. These problems are not insurmountable, but make such a project unsuitable for anyone who does not possess a reasonable amount of audio test equipment, and a good knowledge of how to use it. The same is not true of a tape replay system. This avoids the complications associated with bias levels, etc., of a recording system, and requires only some fairly basic circuitry.

Cassette mechanisms complete with stereo heads are available from time to time on the surplus market at quite low prices. One of these plus a simple tape preamplifier can make a cheap but good cassette player. You might even be able to salvage a suitable mechanism and heads from a defunct cassette recorder or deck. Most cassette mechanisms have two heads, which are the erase head and a combined record-play type. In this application two heads are not better than one.

The erase head is not needed, and can be removed or just ignored. If you are in doubt as to which head is which, things are normally arranged so that the tape passes the erase head first (so that anything already recorded on the tape is erased before new material is placed on the tape). You can always try the two heads to determine which one gives the best results. The output quality of an erase head is such that you will be in no doubt as to which head is which after this "acid" test! Figure 1.17 shows the circuit diagram for a simple cassette tape preamplifier. Like an R.I.A.A. preamplifier, a tape type must provide both a large amount of voltage amplification and some equalization. The output from a tape head is usually well under one millivolt r.m.s., and the source impedance is quite low. In fact the output characteristics of a tape head are broadly similar to those of a low impedance dynamic microphone.

The equalization is required due to the fact that the output from a tape head rises at 6dB per octave. Therefore, in order to produce a fiat frequency response the playback preamplifier must provide a 6dB per octave roll-off. In practice matters are not quite as simple as this. Treble boost (pre-emphasis) is applied during the recording process as part of a simple noise reducing process. On the face of it, this makes it necessary to apply more treble cut in order to produce a flat overall frequency response.

In reality the amount of treble cut required is far less than one would expect. This is due to the imperfect performance of the tape heads which give lower than optimum output levels at high frequencies. In fact they give far lower output levels at treble frequencies than would be expected. Thus the roll-off only needs to be applied over the low and middle frequency ranges, and must flatten out at high frequencies.

The input stage is an inverting mode circuit having a voltage of a little over 26dB (twenty times), and an input impedance of about 4k7. A low noise operational amplifier is used for IC1 in order to produce a good signal-to-noise ratio despite the high sensitivity of the circuit. In fact the noise is likely to be predominantly tape "hiss" rather than noise from the pre-amplifier itself. If you are not intent on having the ultimate in noise performance it is quite acceptable to use a uA74IC, LF351N, etc., for IC1.

IC2 is a non-inverting amplifier, and it is this stage that provides the equalization. C6 provides the 6dB per octave roll-off while either R7 or R8 limits the frequency range over which the roll-off is applied. R8 is switched into circuit in order to give the standard tape preamplifier equalization characteristic.

With R7 switched into circuit the attenuation is applied over a wider frequency range, and at high frequencies there is slightly more attenuation. This setting is used when playing cassettes that have been encoded using the Dolby B system.

This does not constitute proper Dolby B decoding, which requires dynamic filtering. In other words, the degree of filtering must be continuously altered to suit the input level (the higher the input level the more severe the top-cut filtering). This extra filtering helps to give a flatter overall frequency response, and a small amount of noise reduction (about 5dB).

At one time some form of noise reduction was essential in order to obtain reasonable results from cassette tape equipment. Without noise reduction the signal-to-noise ratio was generally only about 40dB to 50dB, with Dolby B noise reduction increasing the noise performance by about 10dB. With modern circuits, tape heads, and tapes, it is possible to obtain quite usable results without noise reduction circuits (as many "Walkman" units demonstrate). In fact modern tape equipment which does not use any noise reduction can easily out-perform older equipment that uses the Dolby B system.

Construction of this amplifier is not critical since the input and output are out-of-phase, and the input impedance is not very high. On the other hand, the high gain at low frequencies does mean that the circuit is very vulnerable to stray pick-up of mains "hum" and other low frequency noise. This includes noise from the motor in the cassette deck. It is also vulnerable to earth and "hum" loops. Therefore, take due care with screening, the earthing, and the decoupling of the supply if a common supply is used for the preamplifier and the cassette deck.

Components for Figure F1.

17 Resistors (all 0.25 watt 5% carbon film)

R1 4k7 47k R2 R3 47k R4 100k R5 680R 680k R6 10k R7 R8 18k

Capacitors 470m 25V elect 4^7 63V elect 47m 25V elect 22m 25V elect 10m 50V elect 4n7 polyester (5% or better)

C1 C2 C3 C4 C5 C6

Semiconductors

IC1 NE5534A LF351N IC2

Miscellaneous

JK1 standard jack socket standard jack socket

Two 8 pin d.i.l. i.c. holders, circuit board, case, JK2 etc.

Note that for stereo operation two preamplifiers will be needed, one for each stereo channel. Therefore, two of each component will be required, apart from the case which must be large enough to accommodate two preamplifier boards.

Finally

This covers preamplifiers for the popular signal sources. For any unusual requirements it would probably be possible to adapt one of the circuits described here to suit your requirements. It should perhaps be pointed out that for high level signal sources such as compact disc players, cassette units, and radio tuners it is not normally necessary to use a pre-amplifier. These have low output impedances and quite high signal levels of about one volt r.m.s., which enables them to drive most power amplifiers, etc., without any assistance.

Where there is a problem with a slightly inadequate output level and (or) a high output impedance, a simple low voltage gain stage will usually suffice. The preamplifier circuit for ceramic and crystal pick-ups is quite good for this type of thing.