Mullard introduce a modified 904V with small input damping.
The new Mullard 904V detector valve with low inter-electrode capacity.
It has long been known that any calculation of stage amplification which does not take into account the important effect of valve capacities is often so erroneous as to be worthless. In a low-frequency resistance-coupled stage, for instance, we may use a valve of high amplification factor together with a coupling resistance of large value and glibly work out, according to the well-known simple formula, that the signal voltage will emerge magnified, say, some thirty times. Actually, at the higher audible frequencies there may be no amplification at all because the working valve capacity will shunt away the energy and our calculation, at any rate so far as high notes are concerned, gives a figure which is thirty times too high.
The influence on amplification of valve capacity, generally known as the 'Miller effect', is even more marked in a high-frequency amplifier. Here the energy fed back through the anode-grid capacity may be large enough and in the right phase to cause uncontrollable self-oscillation, and respectable HF stage gain becomes possible only by decreasing the Miller effect, either by neutralising or by the use of the screen-grid valve.
Under working conditions the anode-grid capacity is reflected back to the input circuit as a much larger capacity depending upon the effective amplification of the valve. There is, in addition, a resistance component thrown back which in the case of a detector is out of phase with the signal, and can be looked upon as having a damping or anti-reaction effect. The two effects - capacity and resistance are generally lumped together and called input impedance, and it is with this that we are primarily concerned in this review.
Valve capacity in the detector stage has hitherto not had its due share of attention, for, although harmful effects are known to exist, they are often masked by the reaction control which is left to make up for sundry losses. With two stages of modern HF amplification, deliberate reaction is not usually employed, and the damping of the tuned circuit immediately preceding the detector is only too obvious when it comes to balance the various members of a ganged tuning condenser. Assuming that we have a receiver with four tuned circuits, of which two are used as a pre-selector, the trimming of the aerial circuit will be flat, that of the next two circuits really critical, whilst that of the fourth next to the detector will be quite flat; in fact, one always feels that this circuit is contributing little to the overall selectivity.
The damping or load imposed by the detector on its input grid circuit is, in the case of a power-grid detector, very large, and any new valve design which aims at reducing this is to be warmly welcomed. Measurement shows that a typical indirectly heated three-electrode detector throws a shunt resistance across the preceding circuit of about 50 kΩ to 60 kΩ, partly due to grid current and partly to Miller effect, so that, however 'good' the tuning coil, its dynamic resistance at resonance will always be below this value.
It is with a keen appreciation of these limitations that the Mullard Company have developed a new detector valve styled the 904V having characteristics which are considerably modified when compared with those of an earlier valve bearing the same type number. By careful design the anode-to-grid capacity (when measured in a cold valve) has been reduced to the low figure of 5 pF, decreasing the total input damping to half the normal figure, namely, 100 kΩ. The selectivity of the preceding tuned circuit is considerably improved and the undistorted output, allowing only 5% second harmonic, is greater, and this in spite of a reduction of mutual conductance from. 6.5 to 2.2.
From the foregoing it would appear that current views on the very steep slope detector may have to be modified. The effect of increasing the mutual conductance is to increase the grid-cathode conductance which may be interpreted as an increase in damping. Add to this the increase in Miller effect or input impedance, which is nearly proportional to the mutual conductance in a detector stage, and we find that the total damping increases approximately as gm2. There appears to be a certain value of gm above which the disadvantages of damping and reduced input outweigh any advantages of greater amplification.
Fig. 1. - Characteristics of the new Mullard 904V detector valve. The anode-grid capacity has been reduced to 5 pF with the result that the input impedance is considerably decreased.
In Fig. 1 are given various operating data for the new 904V. About 170 Volts should be applied actually to the anode, giving a current of about 6 mA., which should not be depressed below 4.9 mA by a signal to keep the output within the 5% distortion limits. The grid capacitor should have a value of 0.0001 μF, the leak 0.5 or 1.0 megohm, and the grid return lead must be taken directly to cathode. It should be noted that the working AC resistance with a high anode voltage and approximately zero grid is about 25,000 Ω, this being a suitable value for coupling by a transformer of fairly high primary inductance or by an RCC stage with 60,000 Ω anode resistance.
Fig. 2. - Curves comparing the LF Volts output of the new 904V with a steep-slope detector (gm = 6.5) under fixed input conditions. The input damping with the valve of low mutual conductance is so much smaller that it gives greater detector efficiency. Depth of modulation - 80%
The relative LF volts developed across a 3½:1 transformer for the new 904V and a high slope detector with gm equal to nearly 7 are shown in Fig. 2. Although the input excitation from the oscillator source was kept constant, the output from the valve with low mutual conductance and intentionally reduced valve capacity was greater.
The Mullard Company should be congratulated on tackling detector valve design from a new angle and translating theory into practical achievement.