This power amplifier has been designed to support and develop the audio reproduction concept based on the use of my audio inverting amplifier. I don't think that power amplification is more crucial for the sound we create than, say, preamplification. On the contrary, preamplification seems to be more mysterious, particularly if the initial audio signals don't exceed 10mV, require amplitude-frequency correction or other transformation. It's why we note a greater variety of sounding various preamplifiers than that of sounding various well designed power amplifiers.
     A power amplifier's task is to properly deliver the already created and formed signal to the loudspeakers and there are some straightforward rules how to do that best. Most full investigation of the problem is described in D. Self's series of articles entitled "Distortion in power amplifiers" which was published in Electronics World in 1993-1994 and now can be found here. I agree with the conclusion that we shouldn't spend time discussing at what level distortion in power amplifiers becomes audible (1%, 0,1%, 0,01% or 0,001%) and what distortion harmonics are most pleasant. A radical solution is to push them all below a 0,001% level at a maximum power and at all audio frequencies. Only after that it will become possible the right understanding of such effects as the influence of cables and interface between the power amplifier and connected loudspeakers.
     The VK-5 power amplifier is built according to a classic generalized scheme: the input differential stage supplied by a current source and loaded by a current mirror, an intermediate voltage-amplifying stage and at last the output current-amplifying stage consisting of two complementary transistor triples. Each of the stages has its own features to achieve outstanding characteristics of the whole amplifier.
     The first stage is optimized in every detail, it's also made balancable for reaching minimum common-mode distortion. The voltage-amplifying stage simply is the inserted inverting amplifier configuration which contains a biasing network for the output stage, the voltage across this network being dynamically varied by the automatically controlled current source Q4 (see Fig.1). The emitter follower Q3 provides additional buffering which allows the amplifier to work with a heavy 4-Ohm load without rise of distortion.

Fig.1. Voltage-amplifying stage.

     The transistor triple chosen for the output stage (see Fig.2) produces the lowest distortion among other triple configurations. If properly used, the circuit can give a notable reduction of distortion in any power amplifier.

Fig.2. Lower half of the output stage.

     A voltage sample across R3 serves as input for current limiting and bias controlling systems, and it's convenient that its lower point -VCC is fixed. Another convenience concerns the output transistors of both complementary halves of the output stage - these transistors have electrically connected collectors and therefore can be mounted on a common heatsink.
Temperature of these and other associated with them transistors doesn't affect the output stage quiescent current which is controlled only by a novel, very effective and simple servo system. The system statically and dynamically maintains the quiescent current at a level ensuring minimum of distortion, it therefore solves, once and for all, the problem of quiescent stability being the weakest point of any class-B power amplifier.
     A good indication of the correctly made amplifier is the ease of understanding its circuitry by a simulation program and close matching of the obtained simulation characteristics with those measured with the help of real instruments. In this respect the VK-5 power amplifier is exemplary and further I would like to adduce a lot of data being the result of running the Multisim 10 simulation software.
     To illustrate how perfectly the amplifying stages work and interact with each other, two characteristics of the amplifier without overall negative feedback are shown. Open-loop gain (Fig.3) is high enough in the whole audio frequency range (about 3000 or +70dB), the second stage providing most of the gain, its correction capacitor C=18pF is responsible for the gain's roll-off to maintain the amplifier HF stability after applying the overall feedback.
     Open-loop distortion (Fig.4,5) is measured on a 4-Ohm load, the applied to it voltage being 14,1V that corresponds to a 50W output power. In these conditions, the second harmonic of distortion lies below -70dB level (0,032%) in the whole audio range - an excellent contribution of the output stage and the quiescent current dynamic control. The third harmonic of distortion is even lower at mid frequencies, but has a tendency to rise at higher ones, reaching -62dB (0,08%) at 20kHz.

Fig.3. Power amplifier open-loop gain.

Fig.4,5. Power amplifier distortion.

     Increasing the output power up to 100W produces slightly higher open-loop distortion (-62dB or 0,08%), the maximum power measured on the verge of output clipping with +35V, -35V supply voltages is 130W on the same 4-Ohm load.

Fig.6. VK-5 power amplifier full circuit diagram (click to view in detail).

     The next characteristics are obtained when analyzing the amplifier final circuit (see Fig.6) with the overall negative feedback applied. Its closed-loop gain is 34 (+31dB) - just what is needed for the chosen maximum 0,7V input signal to produce full output power. Fig.7 represents this closed-loop gain with its high frequency roll-off which is caused by an input low-pass RC-filter deliberately limiting the amplifier bandwidth to prevent external RF-signals from interfering in the amplifier operation.

Fig.7. Power amplifier closed-loop gain.

     Subtracting the closed-loop gain from the previously measured open-loop gain gives a 39dB amount of the applied overall feedback (36dB at 20kHz) which guarantees an absolute high-frequency stability of the amplifier and brings down the closed-loop distortion below 0,001% level.
     The amplifier was designed and built at the time when any simulation program hadn't still appeared, so all its original characteristics were obtained with the help of real measurements carried out by real instruments. The amplifier input was fed from my VK-1 audio oscillator with THD of less than 0,0001% within 20Hz-20kHz and the amplifier output was analyzed by my VK-2 distortion meter whose rejection filter provides -135dB suppression of the fundamental frequency. The amplified residuals, distortion harmonics and noise, were then measured by a RMS-millivoltmeter and observed on the oscilloscope screen. Data was obtained for several audio frequencies and the corresponding curves were drawn (Fig.8).

Fig.8. Power amplifier closed-loop distortion.

     The upper curve 1 in Fig.8 represents total harmonic distortion plus noise of the whole system VK-1 oscillator + VK-5 power amplifier + VK-2 distortion meter. Here the amplifier output voltage is 14,1V RMS, power delivered to a 4-Ohm load is 50W, the distortion meter measurement bandwidth is 100kHz. The lower curve 2 represents practically the system pure noise because all measurements were performed here with the pressed noise reference button of the VK-1 oscillator, producing a 4 times lower signal level everywhere within the system, its noise remaining intact. In these conditions, distortion becomes vanishingly small everywhere and can be neglected at all.
     However, this assumption can be justified only if distortion in the system with the previous nominal signal levels was lower or comparable with its noise. Visually, the VK-2 distortion meter output in the first case was displayed on the oscilloscope screen as practically straight noise band which acquired slight undulation only at the highest audio frequencies, thus indicating some rise of distortion. In the second case, with the noise reference button pressed, an absolutely straight noise band with completely "dissolved" in it distortion was seen at all frequencies within 20Hz-20kHz. The root-square difference between the two readings obtained by the above method at each test frequency defines total harmonic distortion of the whole system at this frequency. The VK-1,2 instruments are included in the measurement system but their joint distortion contribution is so insignificant (<0,0001%), that the result of calculation can be considered as only the VK-5 amplifier distortion. From the curves of Fig.8 we have at 1kHz:

     As shown earlier in the VK-1 oscillator description, the noise generated within the VK-1,2 instruments is 0,0004% relative to 1V level in the set 100kHz bandwidth. Proper subtracting this figure from the measured total system noise of 0,001% yields the VK-5 amplifier noise in the same bandwidth:

In the 20kHz bandwidth the amplifier noise is (100/20)0,5=2,2 times lower and equals 0,00042%= -107,5dB.
     The adduced above results of traditional measurements well correspond to the amplifier circuit direct simulation and Fourier analysis of the amplifier output carried out at several most important frequencies (see Fig.9).

Fig.9. Power amplifier Fourier analysis.

     The amplifier circuit used for simulation in Multisim 10 is depicted in Fig.10. The readings of virtual instruments specify the main parameters of the simulation process: 14,1V output RMS voltage corresponding to a 50W power on a 4-Ohm load, 20kHz signal frequency, 1,57A DC current through the output transistors and at last <-100dB total harmonic distortion of the output voltage. Other details of the amplifier operation are provided by the measurement probes placed at some points of the circuit.

Fig.10. Power amplifier - simulation 50W, 20kHz.

     Maximum linearity of the amplifier, particularly when driving low-impedance loads, was the main target in the course of its development, not a maximum power delivered to these loads. Nevertheless, the simulation shows that distortion remains extremely low also at a 100W output power - <-100dB at frequencies below 10kHz and still -93,4dB (0,00214%) at 20kHz (see Fig.11,12).

Fig.11,12. Power amplifier - simulation 100W, 10kHz and 20kHz.

     All that is achieved with +35V,-35V supply voltages, so the amplifier is running here in the vicinity of output clipping. Fig.13 represents the results of testing the amplifier circuit in quiescent operating conditions which can be created simply be reducing the input signal to practical zero (0,62uV). The quiescent current of output transistors is chosen about 215mA, the used servo system sets this value and then maintains it unchanged over months and years of operation, this current doesn't depend on temperature of any transistor. In active conditions with the output voltage applied to a heavy 4-Ohm load, the quiescent current is varied dynamically to remain always optimal for reducing just switchoff distortion being particularly annoying at highest audio frequencies.

Fig.13. Power amplifier - simulation, quiescent conditions.

     So far the distortion measurements have been conducted in the interactive mode with the help of Multisim distortion analyzer performing Fourier analysis of the tested devices' outputs. But this method has two serious limitations - distortion below 0.001% can not be measured and manual setup of the distortion analyzer fundamental requires this frequency to be exactly equal to the test frequency. This condition is easily satisfied when testing amplifiers by the signal originated from the Multisim pure sine-wave generator whose frequency is set manually too. But in the case of tested oscillators the slightest mismatch of their own frequency and the Multisim distortion analyzer setting may lead to inaccuracy in the carried out THD measurements.
     Recently, I've developed the unprecedented in its reliability, accuracy and sensitivity method of measuring distortion with the help of my virtual VK-2 distortion meter. Its circuit is entered to the simulation program along with the circuit of device under test whose normalized 1V output is connected to the meter's input active rejection filter. The latter automatically tunes itself to the fundamental frequency, its more than -170dB suppression being reached in less than 3sec within 20Hz-20kHz.
     Distortion residuals, amplified by +80dB, are appearing at the VK-2 distortion meter main output during the whole process of this interactive simulation, they are free of any swamping noise and any added distortion because the Multisim oscillator and the VK-2 virtual meter don't create distortion and noise by definition. It's why the measured by an ordinary RMS millivoltmeter and seen on the virtual oscilloscope screen is the distortion harmonics' sum attributed exclusively to the device under test (amplifier or oscillator), its registered value may lie between 0.01% and 0.0000003% -(80-170dB) that corresponds to the meter's main output between 1V and 3nV.
     After stopping the simulation, all obtained measurement data and screenshots can be saved in a file. By the way, the screenshot of extracted output distortion superimposed to the sine-wave input signal is very helpful in understanding the cause of this distortion. For example, narrow spikes on the distortion curve at the moments of the input sinusoid zero-crossing tell about the underbiased output transistors of a power amplifier, the third harmonic of distortion tells about clipping issues and so on.
     The accuracy of measuring the residual distortion harmonics can be easily verified by applying their calibrated amounts, say -120dB, to the meter's input, along with a 1V signal of the fundamental frequency, and analyzing its output, this accuracy being better than 0.5dB at all audio frequencies.

Fig.14. VK-2 distortion measurement of the VK-5 power amplifier at 16kHz.

     The test scheme of Fig.14 illustrates all the said above when testing my VK-5 power amplifier. Its 16kHz-14V RMS output, corresponding to a 50W power delivered to a 4ohm load, is normalized by a voltage divider R149, R150 and applied to the input of the active rejection filter which performs a -172dB notch reduction of the fundamental frequency 16kHz. The whole process of extracting distortion and its +80dB amplification takes 1.36sec, its obtained RMS value is 32mV/10000=3.2μV or -110dB relative to the 1V-16kHz input.
     However, the computer simulation requires much more time than the real 1.36sec measurement process, with the average PC it usually continues 40-50min at 16kHz and 5-10min at 1kHz, all takes place as in the slowed video. This effect allows to study the measurement process in every detail.

Fig.15. VK-2 distortion measurement of the VK-5 power amplifier at 1kHz.

     The similar procedure was conducted and for a 1kHz test frequency, the obtained in 2.87sec RMS value of extracted distortion is 7.7mV/10000=0.77μV or -122dB relative to the 1V-1kHz input, it appears to be the second 2kHz harmonic (see Fig.15). The described measurement method is transparent and very accurate, the above distortion figures are absolutely reliable, but they are lower than the results of Fourier analysis performed earlier (see Fig.9) because the tested here VK-5 amplifier is an improved version of the earlier built VK-5 amplifier.
     Low distortion at the highest audio frequencies is a good indication of the amplifier excellent transient characteristic and transient analysis of the amplifier output confirms that. During this test a square-wave 20kHz signal with 0,5V amplitude and 1ns rise/fall time is applied to the amplifier input, its circuitry kept unchanged, including an output Zobel network. The obtained transient characteristic is depicted in Fig.16 and it clearly can be seen that slew-rate of the amplifier output is more than 20V/μsec.

Fig.16. Power amplifier - transient analysis.

     This power amplifier also includes two circuits ensuring its safe operation. First of them is an output current limiter which prevents destroying of the power transistors when an occasional short-circuit of the output terminals occurs, the amplifier may run at this moment at a full power. The second circuit indicates the start of output voltage clipping, when THD becomes more than 0,5%. Duration of flashing the indicating LED is always constant (about 0,3s), so the clipping can be noted even if it happens in a single period of the highest audio frequency, at the same time these two circuits absolutely don't affect the amplifier linearity.
     The amplifier indicates and withstands not only its overloading with an excessively loud musical program, it behaves adequately also if sudden huge voltage transients come to its input. They may be caused by improperly made commutations and connections in the preceding audio equipment during the amplifier operation or, for example, when the input signal ground is accidentally disconnected. I don't remember any case of the output transistors' emergency replacement during the last five years of everyday work with the unit. As for the output transistors, their parameter matching isn't necessary at all, they can be also of the following types: MJ15024-MJ15025, 2SC5200-2SA1943, KT818GM-KT819GM and others.
     Constructively, the amplifier is built on a single printed circuit board (size 190×80mm) which contains also its protection and overload indication (see Fig.17). The picture clicking produces more detailed animated images of various color performance. The recommended for the board polyester film capacitors – WIMA MKS2 (pin spacing 5mm, voltage 63V), metal film resistors – 0,25W.

Fig.17. The VK-5 power amplifier PCB (scale 1:1).

     Photograph of the exposed stereo prototype of the VK-5 power amplifier is shown in Fig.18.

Fig.18. Stereo power amplifier photo.
                                                                                                                                                            pdf version here
     HOME                                                                                                                                              circuit diagram