The following artical was first published in Practical Television magazine in December 1962.
Right from the beginning it has been the aim of electronic engineers to create equipment for the amplification of very weak radio signals which itself does not add spurious noise signals to the radio signals. The maximum usable sensitivity of any radio or television amplifier is governed directly by the amount of noise signal that the process of initial amplification produces.
Very high gain amplifiers are relatively easy to make, but these are of very little use if the signal to be amplified is so weak that the noise generated masks the amplified signal. It has been intimated in past articles dealing with noise that a "snow" free television picture is only possible when the noise contributed by the initial amplification is at least 200 times (46dB) down on the required signal. This makes it clear that the weaker the signal to be amplified, the better must be the noise performance of the amplifier itself to maintain that 200-to-1 ratio.
Even the very best of amplifiers produce some noise signal, but modern techniques have greatly reduced the noise content. For example, the first amplifier on the Telstar receiver has a very low noise factor indeed, this being essential because of the incredibly weak signals picked up on the aerials from the satellite relay station.
Noise is generated not only in the amplifier but also in the aerial, and then, of course, there are those noises attributable to general static and space signals from stars and so on. Excluding the first amplifiers in the receiver, therefore, some noise will always be present resulting from the mode of radio propaga- tion as we know it at present.
Nevertheless, under practical conditions, the majority of the noise is produced in the first amplifier, for after that the signal is usually strong enough to outweigh the noise contribution of subsequent stages.
Much work has been undertaken to enhance the noise performance of low-level amplifiers, and in one sphere the transistor is being found to be of considerable help, because here there is virtually no " thermal" noise resulting from the emission of electrons from a hot cathode to a positively charged anode.
There is already available a v.h.f. transistor preamplifier whose noise performance of some 3dB better than that of a valve counterpart, and this is designed for the radio and television v.h.f. bands.
Competition to the transistor in this respect has been offered by the fairly recent Nuvistor. This is either a triode or tetrode valve of special design, which in the past was of Americah origin, under the designation 6CW4. Of recent months, however, a Mullard counterpart has been evolved under three main versions. The 7586 which is a medium-p triode, the 7895 which is a high-mu triode and the 7587 which is a sharp cut-off tetrode.
Although Nuvistors are really designed for professional andindustrial applications, they may, nevertheless, prove of interest to the experimenter. Indeed, one excellent design for a Nuvistor Band III Amplifier has already been published in these pages (A Nuvistor Band III Pre-Amp, PRACTICAL TELEVISION, June 1961, p.156). This employs the American 6CW4, but there is no reason at all why the Mullard high-mu version could not be used instead.
The type of electrode structure employed in the Mullard series of Nuvistor valves is based on a concentric arrangement of cylindrical electrodes. These are supported by three pins which proiect through a ceramic base plate (see Figure 1). The valve is finally encased in a metal shell which needs to be adequately bonded to the chassis of the ampli- fier for optimum stability.
Although pins are available far earthing the metal shell it is rather important that something better in the way of earthing is produced by means of the earthing lugs on the, metal shell; During the course of experimenting with the American version, it was found on several occasions that instability tendencies resulted from poor r.f. earthing of the shell, even though the earthing pins were in excel- lent d.c. contact with the chassis.
The characteristics of the Type 7895 Nuvistor are given in Table 1. This shows the very low internal capacitances and the high slope (9.4mA/V) which is attainable under the operating conditions specified. The characteristic range values for equipment design are given in Table 2. The valves can be used with either grid-leak (e.g., grid current) biasing or conventional cathode biasing. The low internal capacitances coupled with the very small lead inductances permit the valves to be used in the earthed-cathode mode, with the input signal applied to the grid. Under this condition, however, neutralisation is necessary to secure optimum stability and noise factor.
In Figure 2 is shown a basic amplifier circuit using grid-leak biasing and capacitive neutralisation. Here it will be seen that the cathode is strapped direct to chassis and that a resistor is used in the grid circuit. Owing to electrochemical activity between the cathode and grid, a small potential develops across the grid resistor, and it is this which is used to bias the valve. This type of biasing usually permits a slightly greater gain to be obtained from an r.f. stage as compared with cathode biasing. The noise figure is also slightly better with grid-leak biasing in most cases, depending upon how well the unit is designed, neutralised and mechanically constructed. Neutralisation is effected in Figure 2 by the pre-set trimmer C1, and if the amplifier is to tune over a band of frequencies it is best to adjust for optimim neutralisation at the low-frequency end of the band. This does not apply, of course, to v.h.f. radio and television preamplifiers, for then it is the usual practice to adjust at the vision carrier frequency.
Possibly the best way that the experimenter ran adjust for optimum neutralisation is first to discon- nect the l.t. feed to the valve, then apply a very strong signal at the required frequency and finally adjust the neutralising trimmer for minimum output.
This is fairly easy to undertake when the stage takes the form of a preamplifier in front of a television receiver. The applied signal can either be on the sound or vision carrier frequency. If the former, the receiver's volume control, should be set at maximum and sufficient input signal applied at the aerial terminal of the preamplifier to be heard adequately in the speaker (the signal modulated, of course).
The neutralising trimmer is then adjusted for minimum output at the speaker. On the vision carrier, a modulated signal will give horizontal bar patterns on the raster, and the neutralising adjustment should be carefully set to minimise these.
Alternatively, the adjustment can be made for the best noise performance, but as this requires a noise generator it can rarefy be undertaken by the experimenter and, in any case, there is very little difference in the overall performance whichever method is used.
The circuit of Figure 2 shows a common earthing point for the various components, and this should be maintained so far as any additional decoupling of r.f. by-pass components are concerned. The base connections and dimensions of the valve are given in Figure 3.
The circuit shown would be suitable for a television preamplifier on any channel in Bands I and III, and it is hoped later to experiment with the valve on the u.h.f. bands, when further details will be published. As a television preamplifier, low impedance coupling windings would be used to apply the aerial signals and extract the amplified signal - L1 and L4 respectively. If high impedance output is required, however, L4 would not be used and the signal would be extracted direct from the anode, via C2.
The number of turns and mode of construction for the coils L2 and L3 will depend on the channel which it is required to amplify. Normal coil winding techniques should be followed, and the dust-iron cores should be used to provide a range of inductance con- trol for tuning. The low impedance coupling coils should be positioned towards the C3 end of L2 and the C1 end of L3.
A slightly modified circuit is shown in Figure 4, where inductive neutralisation is used instead of a capacitor and cathode bias is used instead of leaky- grid bias. R1 provides the bias due to the volts drop across it, while r.f. across the resistor is by-passed by C3. The grid of the valve is returned direct to chassis through L2, and neutralisation is effected by L5. C1 here acts solely as a d.c. blocking capacitor to prevent h.t. from reaching the grid circuit and being shorted through L2.
This kind of circuit has a slightly less effective mutual conductance due to the effect of the cathode bias, but due to this is probably less critical from the stability point of view.
Note that inductive neutralisation may be used with leaky-grid bias and capacitive neutralisation with cathode bias. The inductive neutralisaton and cathode bias combination is shown in Figure 4 simply to compare with the opposite combination in Figure 2.
Inductive neutralisation has several design problems, one being that the very low grid-to-anode capacitance of the valve demands rather a lot of neutralising inductance. Another is that it is often difficult to avoid coupling between the neutralising inductor and the tuning coils, especially in a small compact chassis with closely positioned components.
If there is coupling between the two circuits the grid-to-anode feedback may be increased rather than decreased (neutralised), and great problems of instability will result.
From the experiment's point of view capacitive neutralisation is invariably the best bet. The neutralising capacitor should be of the air-dielectric variety, and the concentric type of trimmer is ideal for this purpose.
Figure 5 shows how two Nuvistors may be connected in cascade. The first is arranged in the earthed-cathode mode with leaky-grid biasing, as in Figure 2, while the second is wired in an earthed-grid circuit with the amplified signal tapped off L2 at a suitable point to match the cathode circuit of the second valve. Ordinary cathode bias is used on the second stage by R1, and the neutralisation is applied capacitively from the end of L4 to the grid circuit of the first stage, via C2. Cascaded circuits are rarely required, however, for a single-stage Nuvistor circuit operating in front of a television tuner will almost certainly make a 2-3dB improvement in the noise figure while also providing a gain up to 40 or 50 times, depending upon the bandwidth and how well the amplifier is constructed and neutralised.
|Input capacitance (grid-to-cathode)||4.5pF|
|Output capacitance (anode-to-cathode)||1.7pF|
|Anode-to-cathode capacitance (excluding shell capacitance)||0.22pF|
|gm (mutual conductance)||9.4mA/V|
|mu (amplification factor)||64|
|ra (anode resistance)||6.8KOhm|
|Grid voltage for 10uA of anode current||-4V|
|Anode current with 110 volts of anode voltage and -1.1 volt grid bias||5.5mA||7.0mA||8.8mA|
|Amplification factor with 110 volts of anode voltage and -1.1 volts grid bias||54||64||74|
|Mutual conductance with 110 volts of anode voltage and -1.1 volts grid bias||7.9mA/V||9.4mA/V||10.9mA/V|
|J.Evans 2001||CLICK HERE TO RETURN TO
THE VALVE TECHIE PAGE
17th September 2001