Polski (PL) 
English (United Kingdom) 
R.M. Malinin, 1st edition
Wydawnictwa Komunikacyjne (Communication Publishing House), Warsaw 1957
A brief commentary: This excerpt from the booklet contains interesting tips for designing low-frequency amplifiers. Most of these tips are used today by manufacturers of high-quality tube amplifiers (grounding, supply voltage filtering, location and alignment of the mains transformer relative to the tubes, etc.). The description is somewhat confusing at times. It's worth keeping in mind that the book is almost 70 years old and, moreover, translated from Russian, so the translation itself contains various linguistic "complexities."
In Chapter 16, the author explains a number of solutions, both long-established and now readily available, that manufacturers often incorporate into amplifier circuits without even considering their purpose. Examples include resistors inserted in series into control grid circuits, etc. In any case, it's worth reading at your leisure.
16. 16. MAINS HUM AND LOSS OF AMPLIFIER STABILITY
Permissible supply voltage ripple. In low-frequency amplifiers powered by AC mains, a number of measures are used to reduce mains hum, which degrades the quality of broadcasts. To minimize the hum generated by the voltage ripple generated by the rectifier powering the amplifier, it is essential that the rectifier filter provides adequate ripple suppression. In practice, the anode voltage ripple amplitude should not exceed 0.5 ÷ 1% of the useful signal present in the anode circuit of a given stage, meaning that the ripple voltage level should be 40 ÷ 50 dB lower than the maximum signal level in the worst-case scenario.
Assuming that the humming voltage arising in one or another voltage amplification stage is amplified by the subsequent stages, it can be stated that the ripple coefficient of the anode voltage of the first stages of a low-frequency amplifier (the ripple coefficient of the rectified voltage is the ratio of the amplitude of its variable component to the constant component) cannot exceed a value of the order of hundredths or thousandths of a percent, while for the output stage operating in a standard single circuit with a triode, the permissible ripple coefficient is 0.1 ÷ 0.5%.
The same ripple factor can also be assumed for the anode voltage of the output stage with a beam tetrode or pentode, provided, however, that the voltage supplying the screen grid of the same stage tube has a reduced ripple factor. This requirement is necessary because the variable component of the screen grid voltage is amplified by the tube, which increases the hum voltage at the amplifier output.
However, if the same voltage values are applied to the screen grid and the anode of the output stage tube (the screen grid is connected directly to the end of the transformer's primary winding, the opposite end to the end connected to the tube's anode), then the screen grid voltage ripple should not exceed the anode voltage ripple of the output stage.
It can be assumed that when using beam pentodes and tetrodes in low-frequency amplifiers, in order to avoid the occurrence of alternating current hum at the outputs of these amplifiers, voltages with a ripple should be applied to the screen grids of the tubes that are approximately ten times smaller than the permissible ripple of the anode voltages of the same tubes.
The higher the amplifier's gain, the more carefully the voltages feeding its first stages must be "filtered." Even the very low hum voltage generated in the first stages of a low-frequency, high-gain amplifier may prove too high at the output after amplification, while even the many times greater ripple voltage present in the anode circuit of the output stage may be completely imperceptible because it is not amplified.
Push-pull amplifier stages can meet lower requirements for filtering ripple voltages from their supply voltages. As is well known, hum compensation from power supplies occurs in these cases. Therefore, push-pull stages can be powered by anode voltage sources with a ripple factor of 1 ÷ 3%.
If negative feedback from the secondary winding of the output transformer is used in the amplifier, a voltage with a higher ripple than is permissible in amplifiers without feedback can be used to power the amplifier's feedback stages, because, as is known, the effect of negative feedback significantly reduces the mains hum.
Application of multi-stage filters. The output stages, having voltages with higher ripple factors during supply than in other stages, simultaneously consume a larger part of the rectifier current.
This justifies the use of multi-stage filters to "smooth" the current (Fig. 45). The amplifier's output stage is supplied with voltage from capacitor C1 (the output of the first filter stage), thus providing sufficient filtration for the output stage. The preceding stages are supplied with a voltage filtered by one or two stages, reducing the ripple to acceptable values for those stages. If the output stage operates in a push-pull configuration, the voltage at the anodes of its tubes is sometimes supplied from the filter's input capacitor (C0), which, in a given case, should have sufficient capacitance to keep the voltage ripple within the acceptable limits for powering that stage.
Fig. 45. Multi-stage smoothing filter systems
With this type of supply voltage filtering, the choke in the first stage should be designed to handle the entire current drawn by the anode and screen grid circuits of all amplifier tubes. Significantly smaller currents flow through the subsequent filter stages, powering only the voltage amplification stages. For this reason, and because lower voltages can be applied to the anode of the voltage amplification stage tubes than to the output stage tube, the second and subsequent filter stages are usually resistive-capacitive, with relatively high resistances.
The voltage drop across the screen grids is usually solved by adding fixed resistances to their circuits; in this case, a fixed capacitor is connected between the screen grid and the amplifier chassis.
With the appropriate selection of the resistance and capacitance values of the capacitor, the created system simultaneously ensures a reduction in the ripple of the voltage supplied to the screen grid in relation to the ripple of the voltage supplying the anode of the same tube.
Hum generation by induction. AC hum is often caused by the influence of internal alternating magnetic fields on the circuits and components of the first stages of an amplifier and its amplifier tubes. In audio recording and playback devices, AC hum can also occur due to the influence of magnetic fields stray from electric motors.
Input transformers (e.g., microphone transformers) and turntable heads (especially electromagnetic ones) are very sensitive to AC induction. Therefore, the microphone transformer must be carefully shielded, and the headshell shield, if metal or metallized, must be grounded. Connecting the headshell and microphone to the amplifier input must be done using a two-conductor shielded cable, and its shield should be connected to the chassis ground at a single point. The other end of the shield should be connected to the chassis of the low-frequency input voltage source.
It is absolutely unacceptable to connect the input voltage source to the amplifier input using an unshielded cable or a single-core cable using the shield as a second conductor. The problem is that the stray field powering the amplifier can induce a variable electromotive force (EMF) in the cable's shield. This EMF acts between the amplifier input and the input voltage source, meaning it is connected in series with the source's EMF. Reaching the amplifier input, this parasitic EMF increases the hum level at the input. Furthermore, hum voltage can arise at the ends of the cable shield if there is leakage or capacitance between the shield and the AC circuits located within the amplifier itself. A significant length of single-core cable connected to the input can result in a significant hum at the input. Using a two-core, shielded cable allows for separation of the shield from the input circuits, thus largely eliminating extraneous influences at the amplifier input.
External magnetic stray fields from a power transformer, filter choke, or motor in audio recording and playback equipment can also cause AC hum by affecting electron tubes. It is known that magnetic fields have the ability to control the electron flow. Under the influence of a magnetic field, the number of electrons reaching the anode of an amplifier tube can be reduced.
If a tube is subjected to a changing magnetic field, its anode current will fluctuate in response to the field change. This phenomenon is particularly dangerous for the first stage tubes of an amplifier, which produce high gain. When tubes with metal housings are used in these circuits, the hum is obviously less than when using glass tubes, but the tubes' metal envelopes do not completely prevent the influence of magnetic fields on the electron flow between the electrodes. For this reason, first stage tubes should not be placed near (in the stray field of) a mains transformer or a turntable motor.
If the output stage tubes are placed too close to the mains transformer (usually glass tubes), AC hum can also occur at the output due to the deflection of the electron flow due to the transformer's stray field. If the flux tetrode is placed close to the supply transformer, the mains hum is strongest when the stray magnetic field lines intersect the electron flow in the direction shown by line A in Fig. 46. If the magnetic lines are oriented along the electron flow, i.e., in the direction shown by line B in the same figure, the hum will be weakest. Therefore, if it is necessary to mount the output stage tubes relatively close to the mains transformer, before finally securing their sockets, try rotating the socket around its axis and listening for any changes in the hum, and find the socket position where the hum is least noticeable.
Fig. 46. If the lines of force of the stray magnetic field intersect the electron streams of the flux tetrode in the direction shown by line A, the strongest hum will be produced, and if these lines are in the direction shown by line B, the hum will be the weakest
When mounting a power supply with a mains transformer on a shared chassis with the amplifier, remember that the mains transformer's stray magnetic field is at its highest intensity in the direction of its coil axis. Therefore, horizontal mounting should be avoided. When mounting the transformer vertically, the coil axis will be perpendicular to the chassis plane, and the transformer's stray field will have a weaker effect on the amplifier components, meaning the likelihood of humming will be much lower.
The distance between the mains transformer and the output stage tubes must be no less than 50 mm. Voltage amplification tubes should be placed even further from the mains transformer. This distance should also be maintained when mounting electric motors in audio recording and playback equipment.
The humming caused by the filament circuits. Some radio amateurs believe that using indirectly heated tubes in amplifiers completely eliminates the possibility of humming noise from the filament circuits, but this isn't always true. If there's leakage between the heater and the cathode (or other electrodes), a variable potential can appear at the cathode, causing hum. This phenomenon is particularly pronounced in tubes operating in the first stages of an amplifier and producing high gain. Such leakage can exist both inside the tube and in its base, and even during assembly. Therefore, it's recommended to use bases and sockets made of the best insulating materials for first-stage tubes.
When a heater (filament) is connected to the negative terminal of an anode voltage source, humming noise can also occur due to the heater's emission of electrons. These electrons, along with those emitted by the cathode, tend to the anode and other tube electrodes, which have a positive potential relative to the heater. Because the heater's thermal inertia is relatively small, the emitted flux fluctuates with the heater voltage (at a frequency twice that of the mains), resulting in a ripple current in the anode circuit. This phenomenon is also most dangerous in the first stages of the amplifier.
It should be noted that the hum that appears in stages with individual automatic grid voltages tends to be more pronounced the higher the voltage, i.e., the higher the voltage between the cathode and the heater. The cathode, being very close to the heater and having a positive potential relative to it, strongly attracts the electrons it radiates. Obviously, the higher the cathode potential and the greater the electron flow emitted by the heater, the more pronounced the hum.
The hum caused by the heaters can be effectively reduced by applying a positive potential to the heaters relative to the cathodes (or a negative potential to the cathodes relative to the heaters). To achieve this, disconnect the tube filament winding from the negative terminal of the rectifier (amplifier chassis) and connect it to a voltage divider R1R2 (Fig. 47), set to the full rectifier voltage. The values of resistors R1 and R2 forming the divider are chosen so that a voltage of around 10 V can be obtained across resistor R2. This voltage exceeds the voltage values typically used in voltage amplification circuits, i.e., it guarantees a negative potential at the tube cathodes relative to the heaters. The effect of negative voltages on the control grids is not affected in this case.
Reducing electron emission from the heaters of the voltage amplifier tubes, and consequently attenuating the output mains hum, can also be achieved by lowering the tube filament voltage to 5.5 ÷ 5.6 V (instead of the nominal 6.3 V). This does not significantly reduce the gain, especially if these stages have higher resistances in the anode circuits.
Fig. 47. A system for obtaining a negative potential on the cathodes of electron tubes in relation to the heaters.
Grounding the center of the tube filament winding in the power transformer also helps reduce hum. If directly heated tubes are used in the output stages, the filament winding of these tubes should be grounded at the center.
We'll also discuss a number of ways to reduce mains hum. Experience shows that using tubes with a grid lead located on the bulb in the first stages produces less hum than tubes with a grid lead in the base (for example, a 6Ż7 tube is better than a 6Ż8). Furthermore, the grid lead on top of the tube bulb should be shielded with a steel cap, and the wire to the grid should be shielded. The cap is connected to the shield, which in turn is connected to the amplifier chassis. Using a triode in the first stage can keep the hum level several times (5 ÷ 7) lower than when operating the stage with a pentode.
To reduce mains hum in amplifiers with high gain factors, the amplifier chassis should not be used as a conductor. All "negative" circuit components should be mounted using insulated wire and connected to the chassis at a single point. Care should be taken to make this connection at various points, ultimately connecting it where the hum is least significant.
In multi-stage amplifiers providing high gain, in order to reduce hum, the filament power supply of the first stage tubes is sometimes provided by selenium rectifiers.
Loss of amplifier stability. As previously mentioned, negative feedback amplifiers sometimes become excited (parasitic oscillations occur) because different phase shifts occur in the amplifier at different frequencies; for some frequencies this feedback may turn out to be positive.
The phenomenon of excitation (loss of stability) can occur in amplifiers (especially high-gain ones) that lack a dedicated feedback circuit, and even in amplifiers with negative feedback, without the use of a feedback circuit. An amplifier can become excited because of unwanted (parasitic) positive feedback between the anode circuit of a tube in any stage and the grid circuit of any tube operating in the preceding stages (rarely the same tube). This causes the amplifier to operate abnormally, producing a constant-frequency sound in the loudspeaker (usually a whistling sound), and playback is accompanied by severe distortion. If parasitic vibrations arise in the superacoustic frequency range, even though we don't hear them, they are the cause of nonlinear distortion.
In many cases, the causes of such excitation are very similar to those of mains hum. For example, positive feedback, leading to excitation, can arise from the interaction of the external magnetic stray field of a mains transformer or the magnetic or electric field of the amplifier's output circuits with the input circuits. Shielding the input circuits to help combat mains hum (especially the microphone transformer), followed by rational component layout and proper amplifier assembly, can often prevent excitation. For example, in a high-gain amplifier, the output stage should not be mounted close to the first stages, as their mutual proximity creates favorable conditions for parasitic feedback. Connecting wires, especially in the first stages, should be as short as possible. Shielding the output transformer also helps in this regard.
Feedback through power sources. Occasionally, even when all necessary precautions have been taken to avoid parasitic coupling in the design of a low-frequency amplifier, it can still arise from the common power supplies used by the amplifier stages. This phenomenon can be explained as follows.
During amplifier operation, the anode currents of all stages vary with the frequency of the amplified oscillations. Because the anode voltage sources have a certain internal resistance, the voltages drawn from them fluctuate. In other words, variable voltage components arise at their terminals. Because the currents from the source to the output stages vary within the widest range, the values of the variable components at the voltage source terminals depend primarily on these stages. These variable components enter the anode tube circuits of the preceding stages, and from there, they reach the grid circuits of the stages coupled to them. In this way, feedback occurs between the stages. If the internal resistance of the power supplies is sufficiently high, the variable components arising at them and reaching the preceding stages as feedback voltages can be so large that conditions for natural oscillations to occur in the circuits arise. Excitation, of course, occurs when the feedback is positive.
It should be noted that under these conditions, amplifiers typically excite even in the absence of a varying voltage at their input. Any random changes in the currents in the anode circuits lead to voltage changes at the anode source terminals and, subsequently, to voltage changes in the upstream stages, which in turn trigger natural oscillations in the circuit.
Removing feedback through power sources. Positive feedback from an anode voltage source can sometimes be removed by bypassing it with a sufficiently large capacitor. Including such a capacitor, which has a relatively low resistance for low-frequency currents, reduces the AC voltage component between the power supply terminals, thus eliminating feedback between stages. This method often eliminates excitation in low-power amplifiers powered by dry batteries, whose internal resistance can be quite high.
The excitation of a low-frequency amplifier powered by a single-stage filter rectifier can sometimes be removed by increasing the capacitance of the filter output capacitor.
Using rectifiers with multi-stage filters to power low-frequency amplifiers typically ensures more stable operation of the amplifiers and makes them less prone to excitation. With this power supply method, the anode circuits of different stages are separated by chokes or filter resistors, which prevent the transfer of low-frequency AC components from the anode circuits of one stage to the other. For example, in the circuits shown in Figures 45a and b, a low-frequency AC voltage component is created on capacitor C1 due to changes in the output stage's anode current, i.e., a certain supply voltage ripple occurs at the frequency of the amplified currents. These ripples reach the upstream stages through subsequent filter elements, which reduce the ripple amplitude in the same way they reduced mains hum. We can see that in this respect the circuit in Figure 45c gives better results than the circuits in Figure 45a and b, because in this circuit, in the path of the pulsating current from the anode circuit of the output stage to the anode circuits of the preceding stages, there is both a choke Dł and a resistor R1.
To prevent low-frequency amplifier excitation, the anode circuit of each stage is powered by an individual filter, consisting of a resistor Rfa and a capacitor Cfa (Fig. 48). These filters impede the penetration of low-frequency AC voltage components from the anode circuits of some tubes into the anode circuits of others through the power supply, thereby weakening possible feedback loops and improving the amplifier's operational stability. Such filters are called decoupling filters. The capacitors in these filters typically have a capacitance of tenths of a microfarad (sometimes several microfarads) and a resistance of a thousand or several thousand ohms. In some cases, when the anode current of the stage is quite high and a significant DC voltage drop across the decoupling filter is unacceptable, a choke is used instead of a resistor.
Fig. 48. Decoupling filter system
It should be noted that the decoupling filters in amplifier stages powered by rectifiers also serve as filters that reduce mains hum.
In high-gain multistage amplifiers, where the negative voltage on the tube grids of different stages is obtained from a common source, to prevent feedback from this source, decoupling filters are also incorporated into the tube driver grid circuits, as shown in Fig. 48 (Rfs Cfs).
In conclusion, we should add that parasitic vibrations sometimes arise in amplifier stages (especially in high-power output stages) at very high frequencies, corresponding to the ultrashort wavelength range. These vibrations can occur when the inter-electrode capacitances of the tubes, together with the assembly capacitances and inductances of the wires, create vibration circuits tuned to such frequencies, and conditions favorable for the generation of vibrations in these circuits exist. Predicting the possibility of such vibrations in advance is practically impossible. An effective measure to combat vibrations is to increase the damping of the parasitic vibration circuits. This can be achieved by series-connecting resistors with resistances of several dozen or several hundred ohms, or small high-frequency chokes, to the grid or anode circuits of the stages where such vibrations arise.