Vacuum Tube FAQ
The Vacuum Tube FAQ
by Henry Pasternack
Originally posted on rec.audio.tubes
Edited/formatted by Ken Gilbert
—
Q: Briefly explain vacuum tube functional elements.
A: The vacuum diode is a two-terminal device having two active elements, the cathode and the anode (or plate). The cathode is designed so that when it is heated, it freely emits electrons. The application of a high positive potential from plate to cathode causes a conventional current to flow through the tube. The magnitude of the current that flows is proportional to the three-halves power of the applied voltage. No current flows when a negative potential is applied. The diode is commonly used as a rectifier in power supply circuits.
The triode is similar to the diode, but contains a third element, located between the cathode and plate, called the control grid. A negative bias applied to the grid relative to the cathode shields the cathode from the electrostatic field due to the plate. Consequently, the flow of current from the plate to the cathode is reduced by the effect of grid. If the negative bias is sufficiently large, no current will flow in the anode circuit and the triode is said to be cut off. Current flow in the triode is thus dependent on both grid and plate voltage.
A second grid, called the screen grid, may be added to the triode. A tube with two grids is called a tetrode. The screen grid shields the control grid from the plate, greatly reducing the dependency of plate current flow on plate voltage. Consequently, the tetrode has a much higher output resistance and behaves more nearly as a current source than the triode. The tetrode is capable of greater voltage amplification than the triode because of its higer output resistance.
A third grid is often added to the vacuum tube, turning it into a pentode. The supressor grid is normally connected to the cathode. Its effect is to reduce the secondary emission of electrons from the plate, allowing the tube to operate over larger excursions of plate voltage. The pentode is thus capable of large output voltages with minimal distortion. Because of this fact, the pentode is a more efficient power amplifier than the triode.
Q: Describe in greater detail the operating characteristics of power vacuum tubes.
A: The vacuum tube is a voltage-controlled device, often compared to the depletion-mode n-channel MOSFET. High plate current will flow when zero control-grid bias is applied; a negative bias is required to control (limit) tube conduction. Compared to a MOSFET, the power tube has very low transconductance, on the order of thousands of micromhos. Maximum continuous cathode current is also low, in the range of several hundred milliamperes. Popular output tubes have plate dissipation ratings on the order of twenty to forty watts.
In order to extract reasonable amounts of power from a low current device, a high load impedance is required. Although it is possible to direct-couple a vacuum tube output stage to a low-impedance loadspeaker load, the vast majority of tube power amplifiers are output transformer-coupled. The output transformer permits the amplifier to operate with reasonable efficiency into eight-Ohm loads.
Popular power tubes are typically operated with DC plate voltages ranging from 350 to 600 volts, screen voltages on the order of 250 to 360 volts, control grid bias voltages from -60 to -30 volts, and idle plate currents of 25 to 75 mA, depending on the application and the tube type.
The output resistance of a triode is moderate, typically no more than a few thousand ohms. Its characteristic curves suggest a voltage-controlled voltage source with a series plate resistance. The tetrode and pentode have output resistances an order of magnitude higher than that of the triode. Because of the greatly reduced dependency of plate current on plate voltage, the tetrode/pentode may be thought of a voltage-controlled current sources. Indeed, the characteristic curves of the pentode are quite similar to those of the MOSFET.
The screen grid significantly reduces the capacitance between the plate and the control grid, reducing Miller feedback and increasing the bandwidth of the tetrode/pentode. The increased output resistance is revealed by a flattening of the characteristic curves. Similar changes occur when two triodes are connected in cascode.
Idealized Plate Characteristics
Triode Pentode
------ -------
Vg2 = Constant
| Vg = 0 |
| / -10 | ___________________ Vg1 = 0
| / / -20 ||___________________ -10
| / / / -30 ||___________________ -20
Ip| / / / / -40 Ip ||___________________ -30
| / / / / / -50 ||___________________ -40
| / / / / / / -60 ||___________________ -50
| / / / / / / / ||___________________ -60
|/ / / / / / / |/
---------------------- ----------------------
Vp Vp
The tetrode/pentode is typically operated with a fixed screen grid potential. Varying the static screen voltage scales the characteristic curves up and down correspondingly. If the screen grid is connected directly to the plate, the tube is said to be operating in “triode” mode and has characteristic curves like those of a conventional triode.
Plate current continues to increase as the control grid potential swings positive above zero volts. At the same time, grid current begins to flow. Many tubes will operate linearly while drawing grid current on signal peaks. Because maximum grid dissipation is low, it is necessary to limit the magnitude and duration of grid current flow. Most amplifiers with high-impedance driver circuits will distort heavily at the onset of grid current.
Q: What is “ultralinear” operation?
A: In ultralinear operation, the screen grid is connected to a tap on the output transformer primary such that the screen voltage varies in proportion to the plate signal voltage. The constant of proportion- ality typically ranges from 30-50%, although other ratios will also work. The resulting tube characteristic has properties intermediate between those of the triode and the pentode. In essence, the ultra- linear connection forms a local negative feedback loop around the output stage. This may be advantageous depending on circuit topology and gain distribution. Advocates of ultralinear operation claim this connection combines the best features of triode and pentode mode, while detractors claim it lack the virtues of either.
Q: Compare the general properties of triode, pentode, and ultralinear power amplifiers.
A: It is probably misleading to characterize the various types of output stage connections in terms of sound quality. Many factors contribute to the sound of an amplifier, and a good designer will blend these elements in order to achieve a particular goal. On the other hand, certain sonic qualities are associated with each type of output stage frequently enough that they deserve repeating here. In addition, there are objective differences that are worth mentioning.
The triode amplifier is characterized by low efficiency and low power output. This is because a smaller voltage swing is available from the triode for a given DC plate voltage. Consequently, the triode amplifier burns up more power at idle relative to its peak output. The sound of the triode amplifier is often described as “rich” or “sweet”, conveying in a natural and realistic way the harmonic structure of musical instruments and voices.
The pentode amplifier is often described as having a more analytical sound than comparable triode units. Others may accuse it of sounding harsh. Objectively, the pentode output stage tends to produce more high-order distortion products than a comparable triode. In addition, the pentode is more sensitive to load impedance variations and may clip more sharply than the triode.
The ultralinear amplifier combines the benefits (or flaws, depending on your point of view) of the triode and pentode connections. The ultralinear characteristic curves resemble those of the triode in some ways, those of the pentode in others, and have unique characteristics as well (regretably, they are hard to render in ASCII). While the general concensus favors triode mode above all, there seems to be no strong trend supporting ultralinear over pentode mode, or vice-versa. Perusing the high-end magazines, one can find examples of well-regarded amplifiers ueing either type of output connection.
The one certainty is that the ultralinear connection is the cheapest way to get good performance and high power out of a pentode. Whereas a quality pentode design requires a stiffly regulated screen grid supply, all that is needed to implement an ultralinear output stage is a pair of transformer primary taps. Perhaps the economic argument leads the sound quality argument in this case.
Q: Can I convert my amplifier back and forth between pentode, triode and ultralinear modes in order to hear the difference for myself?
A: In general, the answer to this question is “No.” Under some circumstances, it may be possible to perform such experiments, but subject to limitations.
If the amplifier is an ultralinear design, it is possible to convert it to pentode operation by connecting the screens to a fixed voltage source. The correct screen voltage depends on the type of output tube, the B+ supply voltage, and the output transformer primary impedance. For audiophile performance, a regulated screen supply may be required. This makes the pentode conversion a major modification.
The most common conversion is to modify an ultralinear or pentode mode amplifier for triode operation. In many cases, this modification can be made successfuly and with little effort, but some caveats apply. One would like to be sure that the maximum triode-connected plate potential is not exceeded. For many EL-34/6L6/KT-66/5881 amplifiers running B+ supplies on the order of 400V, there is no problem converting to triode operation. On the other hand, a 6550 amplifier with 550V on the plates is probably not a candidate for triode conversion without a reduction in B+ voltage. When the conversion is made, a 100 Ohm non-inductive resistor is usually specified, connected directly between the screen and plate pins on the tube socket, to suppress RF instability.
Changing the output stage connection from pentode to triode mode typically lowers the open-loop gain of the amplifier. As a result, the closed-loop global feedback factor also goes down. The output impedance of the amplifier, its sensitivity, the total harmonic distortion and the distortion spectrum will all change. Overload behavior and stability will likely be improved. Typical comments are that the triode-connected amplifier sounds “more relaxed”, “warmer”, and “sweeter” after the conversion. Whether this is due to an inherent quality of triode-strapped pentodes, or is a consequence of modifying a topology that was not designed with triode output in mind, is open for debate.
Q: What about “pure-triode” amplifiers?
A: The vintage triode power tubes, such as the 845, 2A3, and 300B, are classic devices from the earlier days of vacuum tube technology. They are still available in limited supply and at high cost (although there are now Chinese copies on the market that offer a reasonable, lower- price alternative). A significant structural difference between these tubes and more modern units is the use of a directly-heated cathode. In this design, the cathode heater also serves as the emissive element. In contrast, newer tubes employ a separate heater that is electrically and mechanically isolated from the cathode.
These tubes are “pure triodes”, meaning that there is no screen grid to be strapped to the plate in order to achieve triode operation. The classic triodes have very low plate resistance and low voltage gain. Many require significantly higher plate supply voltages than ordinary pentodes. In exchange for these limitations, these tubes offer very linear characteristic curves, making possible the design of low- distortion amplifiers that use little or no local or global feedback. The sound of a pure-triode amplifier is reputed to be exceedingly musical, with a natural harmonic structure, very low grain or noise, and a realistic, inviting nature. Triode adherents claim that the pure-triode output stage is sonically superior to one constructed with strapped screen grid pentodes. Other listeners will find the pure-triode amplifier to be colored, restricted in bandwidth, inefficient, and overpriced.
Single-ended triode amplifiers have been very popular in Japan for some time, and are making a limited comeback in North America.
Q: What is the difference between a single-ended and push-pull amplifier?
A: A push-pull output stage uses one or more pairs of output devices connected in a symmetrical arrangement such that output current flows to the load first through one half of the circuit and then through the other half. The advantages of the push-pull topology are higher efficiency, higher power output, much lower even-order distortions, immunity from power supply ripple, and zero DC current in the output transformer primary.
In contrast, the single-ended output stage employs only one set of output devices which conduct continuously throughout the output current cycle. This forces the stage to be operated in class A mode, limiting the available power output and greatly lowering efficiency. Total harmonic distortion is higher because there is no cancellation of even-order harmonics. Power supply ripple is not rejected by the single-ended output.
The most significant difficulty of the single-ended output stage is that the output transformer is required to carry a large DC current in its primary. Due to magnetic saturation and nonlinearity effects, a very special output transformer design is required. Such a transformer is large, heavy, expensive, and has a low power rating. The resulting amplifier is restricted in bandwidth at both extremes of the audio spectrum and produces a great deal of distortion. To minimize distortion (and to add to the single-ended mystique), it has become fashionable to design single-ended amplifiers with pure-triode output stages.
While no one claims the pure-triode, single-ended amplifier is “neutral” or “accurate”, devotees of the genre describe in almost mystical terms the sonic attributes of these amplifiers. The word “magic” is often used. Listeners will have to judge for themselves.
Q: What are the meanings of Class A, B, and C?
A: Because virtually all active devices pass current in only one direction, it is necessary to go to some trouble in order to amplify audio signals, which are alternating currents. There are basically two strategies for making one-way components amplify two-way signals. The first is to use a pair (or pairs) of devices arranged so that one half of the circuit conducts current exclusively during positive swings of the signal, and the other half conducts exclusively during negative swings of the signal. This arrangement is called “push-pull” operation.
The other strategy is to superimpose a direct current on the AC signal of such a magnitude that the combined current remains net positive at the negative signal peaks. The result is that the amplifying device is never required to reverse the direction of its current flow. The superimposed direct current, which is known as a bias current, may be filtered out of the amplified signal using a transformer or blocking capacitor. In its simplest form, this type of circuit is known as “single-ended”.
If we consider purely sinusoidal signals, it is clear that the output device in the single-ended circuit conducts current during 100% of the audio signal cycle. In contrast, each device in the push pull circuit conducts during exactly 50% of the signal cycle. It is also possible to construct amplifiers in which the output devices conduct for less than 50% of the signal cycle, although these circuits are generally not employed in audio amplifiers. The Class A, B, and C designations refer, respectively, to these three modes of operation.
Returning to the push-pull, Class B amplifier, it can be seen that during zero crossings of the output signal, the positive half of the circuit switches off just as the negative half begins to conduct, and vice versa. At the precise instant that the output current is zero, no current flows in either half of the circuit, i.e., there is no bias current. In practice, amplifying devices are quite nonlinear near their low-current cutoff points. A type of nonlinearity called “crossover distortion” can be eliminated if the Class B circuit is modified so that a modest bias current flows while the amplifier is idling. The small overlap in conduction between the two halves of the circuit smooths over the transition that occurs during zero crossings. Because the bias currents in the positive and negative halves of the circuit are are equal and opposite, they cancel one another automatically. This is why the net DC current in a push-pull output transformer primary is zero, and why solid-state push-pull amplifiers require no output DC blocking capacitors.
As the push-pull bias current increases, the conduction angle of each half of the circuit increases from the minimum value of 50%. If the bias current is set to a value equal to one half of the maximum output signal current, the conduction angle will equal 100% and the amplifier will be operating in Class A mode. The flow of bias current in the absence of signal current dissipates energy in the form of heat. Thus, the efficiency of the amplifier is reduced as the bias current increases from zero to full Class A operation.
An amplifier that is biased part-way between Class B and Class A operation is said to operate in Class AB mode. In vacuum tube amplifiers, an additional distinction between Class AB1 and Class AB2 is made. In Class AB1, the driver stage has a high output impedance and clips at the onset of output tube grid current flow. In Class AB2, the driver stage has a low output impedance, allowing it to drive the output grids linearly into the positive grid current region. This allows greater output power with a given bias current.
Despite the efficiency penalty, many listeners believe that Class A amplifiers drive difficult loads with more authority and a smoother sound than comparable class AB units.
Q: How do I determine plate and screen supply voltages, bias currents, primary impedance and output power for vacuum tube output stages?
A: There are textbook equations for determining these quantities, but the designer’s best friends are the tube characteristic curves and the load line. The triode and pentode characteristic curves are reproduced below with a load line indicated by a series of asterisks.
Triode Pentode
------ -------
Vg2 = Constant
| Vg = 0 |
|* / -10 Im |*___________________ Vg1 = 0
| * / / -20 ||_*_________________ -10
Im |----*/ / / -30 ||___*_______________ -20
| / */ / / -40 ||_____*_____________ -30
| / / */ / / -50 ||_______*___________ -40
| / / / */ / / -60 ||_________*__________ -50
| / / / / */ / / ||___________*________ -60
|/ / / / / */ / |/ *
-----------------*---- -----------------*----
Vb Vb
To construct the load line, it is necessary to know the effective load impedance. For single-ended amplifiers, this is equal to the output transformer primary impedance. For push-pull amplifiers, the load impedance is equal to he plate-to-plate primary impedance divided by four (Rpp / 4). The slope of the load line is equal to the negative inverse of the effective load impedance, (-1 / Rl). The intersection with the x-axis occurs at a point, Vb, equal to the B+ voltage minus the cathode voltage (for cathode biased output stages). The y-axis intercept occurs at the current, (Vb / Rl).
For Class AB1 operation, the maximum output current, Im, can be found by locating the intersection of the load line with the Vg = 0 curve on the graph. It can be seen from the diagrams that the pentode connection provides a higher peak output current at the clipping point than the triode connection. The maximum available RMS ouput power is determined from the standard equation, Pmax = 0.5 * (Im ^ 2 * Rl).
For example, assume a fixed-bias, push-pull, EL-34 output stage with B+ = 400V, Rpp = 4000 Ohms, and Im = 260 mA. This gives:
Pmax = 0.5 * [0.260 ^ 2 * (4000 / 4)] = 33.8 W
In the case of the triode output stage, optimum efficiency is achieved when the effective load impedance is equal to the output tube plate resistance. This value can be determined by approximating the triode characteristic curves to straight lines and computing their slope from the graph. Because the triode is relatively insensitive to changes in load impedance, it is permissable to increase the primary impedance by a factor of two from the optimum value. The result will be a moderate reduction in output power, a lowering of distortion, and a margin of safety for driving low-impedance loads.
The optimum pentode load line cannot be determined directly from the characteristic curves. The primary constraint is to choose a load line that maximizes the undistorted signal output. For a given power output near clipping, increasing the load impedance will cause an increase in third-order distortion, while lowering the load impedance will result in increased second-order distortion. Lowest distortion is achieved when the load line passes through the knee of the Vg1 = 0 curve at the upper left-hand corner of the graph.
A certain amount of flexibility can be had by adjusting the screen grid potential to accommodate a given primary impedance. As the screen grid voltage is raised or lowered, the characteristic curves expand or contract in the vertical direction. The optimum screen voltage is the one that places the knee of the Vg1 = 0 curve on the load line. The value of the B+ supply voltage also influences the relationship between the load line and the characteristic curves and must be taken into account when selecting pentode operating parameters. In reality, the selection of these parameters may not be as critical as the textbooks suggest. Particularly since loudspeaker impedance is highly variable, one may assume that the pentode output stage will tolerate modest variations in its operating parameters.
Q: How do I design with paralleled output tubes?
A: The effective plate resistance of ‘n’ tubes in parallel is equal to the individual plate resistance divided by ‘n’. If the optimum load impedance for a single tube (or push-pull pair) is Rl, the proper load impedance when ‘n’ tubes are paralleled is thus (Rl / n). The available output power becomes:
Pmax(n) = 0.5 * (n * Im) ^ 2 * (Rl / n) = 0.5 * n * Im ^2 * Rl = n * Pmax(1)
If, on the other hand, the output tubes are paralleled without changing the output transformer primary impedance, each tube will see an effective load impedance equal to ‘n’ times Rl. It may be tempting to double up on output tubes and then drop the effective primary impedance by placing the load across the next higher-impedance secondary winding. Unfortunately, the frequency response, power handling, and insertion loss characteristics of an output transformer are dependent on its operation between the specified source and load impedances. Mismatching the transformer in an attempt to achieve a particular load line not advisable. This topic is discussed in greater detail in (you guessed it) “Sound Practices” magazine, volume 1, issue 2.
It is necessary to insure that the driver stage is capable of driving the capacitive load imposed by the parallel output tube grid circuits. The driver must have sufficient bias current to charge the grid-cathode capacitance and the Miller feedback capacitance of the output stage. If insufficient current is available, a slew rate problem will result. When many tubes are driven in parallel, it may be worth considering the use of a low-impedance, cathode follower driver stage.
One must address the problem of tube matching when parallel outputs are used. Mismatched tubes will result in “current hogging”, reduced performance and shortened tube life. The use of modest cathode resistor will provide degenerative feedback and help to minimize the impact of tube-to-tube transconductance variations.
Back to Tube Amp Theory.
