Testing Audio Amplifiers [Troubleshooting With the Oscilloscope]

Home | Articles | Forum | Glossary | Books





A wide variety of useful tests can be made in audio equipment with a scope. Such equipment ranges from the simple audio amplifiers used in table-model radios, through commercial-sound installations, to high-fidelity amplifiers. The vertical and horizontal amplifiers in service-type scopes are seldom capable of hi-fi response. It might therefore be supposed that accurate checks of distortion could not be made.

It is a general rule that test equipment must have performance characteristics equal to or better than the device under test. There are, however, certain exceptions which are made possible by suitable test techniques.


Fig. 12-1. Check of scope linearity.

LINEARITY CHECKS

Amplitude nonlinearity is a basic cause of distortion in audio amplifiers. In order to make a linearity test with a scope, first determine the linearity of the scope itself. This provides a reference pattern for use in evaluating the linearity of an audio amplifier. Connect the output from an audio oscillator to both the vertical- and horizontal input terminals of the scope, as shown in Fig. 12-1. (The waveform of the audio oscillator is of no concern here.) Now set the audio oscillator frequency to approximately 40 Hz. A diagonal-line display appears on the scope screen.

If the scope amplifiers are linear, a perfectly straight line will be displayed. If the amplifiers are not linear, the line may have some curvature, as in Fig. 12-2. For an accurate evaluation, place a straightedge along the pattern. This is the reference pattern used in the following test. Connect the equipment as shown in Fig. 12-3. Load resistor R must have an adequate wattage rating, and its resistance should equal the recommended load impedance for the amplifier. The amplifier should be driven to its maximum rated power output. Power output is determined by measuring the ac voltage across R. The voltage is measured in rms units with an ordinary vom. The power in watts is equal to E^2/R.


Fig. 12-2. Reference linearity pattern.

Now, observe the pattern on the scope screen. If it is exactly the same as the reference pattern, the amplifier under test is linear. On the other hand, more or less nonlinearity is indicated by more or less departure from the reference pattern. If the amplifier under test has good performance characteristics, there may be some doubt whether or not the scope pattern really shows any departure from reference. In fact, very small amounts of nonlinearity are difficult to evaluate with certainty.


Fig. 12-3. Amplifier linearity check.

An amplifier which has substantial nonlinearity at high power output usually shows less nonlinearity when the power output is reduced. Any amplifier develops increasing nonlinearity as the power output is increased. Objectionable nonlinearity at rated power output can be caused by incorrect grid bias, low plate or screen supply voltages defective transformers, off-value resistors, or open bypass capacitors. Leaky coupling capacitors change the normal grid bias on a tube. Leaky or shorted cathode bypass capacitors change the normal cathode bias. If a coupling capacitor is low in value, the preceding stage must be overdriven to obtain rated power output, with resulting nonlinearity. An open capacitor in a feedback network causes amplitude nonlinearity.

PHASE SHIFT

Unless the amplifier is defective, it is highly unlikely that you will observe any phase shift in the pattern at 400 Hz. Phase shift in the amplifier under test causes the line pattern to open into an ellipse.

The proportions of the ellipse indicate the amount of phase shift.

Some key patterns are illustrated in Fig. 12-4. Amplifier defects resulting in phase shift include low-value coupling, decoupling, and bypass capacitors; defective transformers; or a defect in the feedback circuit.


Fig. 12-4. Typical phase shift patterns.

Any amplifier, including the scope amplifiers, will exhibit phase shift at some limiting upper frequency. Here, stray circuit capacitances begin to become significant. The stray capacitances have a partial bypassing effect around plate-load resistors in particular, causing the load to become noticeably reactive at the high test frequency.

Phase shift is always the result of reactance. Unless amplifiers are dc coupled, they also exhibit phase shift at some limiting low frequency.

This occurs because the values of coupling, decoupling, and bypass capacitors are insufficient to maintain negligible reactance at the low test frequency.

In case of simultaneous amplitude nonlinearity and phase shift, a distorted ellipse is displayed. The ellipse appears more or less flattened, skewed, or egg-shaped, with one end more "open" than the other. In hi-fi amplifiers, nonlinearity is more objectionable than phase shift, because listeners detect nonlinear distortion more readily than phase shift in the audible output. The better hi-fi amplifiers are designed, however, to minimize phase shift.

LINEAR TIME-BASE DISPLAYS

The cyclogram test depicted in Fig. 12-3 is preferred to a display on a linear time base (sawtooth deflection) because small amounts of distortion are much more difficult to observe on a linear time base.

If substantial distortion is present, as in Fig. 12-5, it is immediately evident, but on the other hand, it is practically impossible to observe small amounts of distortion. If a linear time base is used, adopt the same precautions in establishing a reference pattern, as previously described. Connect the audio-oscillator output directly to the scope's vertical-input terminals, and observe this reference pattern. It shows the combined effect of observable distortion in the generator waveform, plus possible additional distortion from the scope's vertical amplifier.


Fig. 12-5. Distorted sine wave.

Where the technician is concerned only with the signal amplitude, as in gain measurements, a linear time base serves satisfactorily. It is also appropriate for checking bypass capacitors, as seen in Fig. 12-6.


Fig. 12-6. Checking a screen bypass capacitor.

If the bypass capacitor is satisfactory at the test frequency, little or no ac voltage is present. An open capacitor, however, causes a large deflection on the scope screen. A linear time base is also used when a scope supplements a harmonic-distortion meter, as shown in Fig. 12-7. The harmonic-distortion meter filters out the fundamental in the test frequency, and passes the harmonics. The meter indicates only the percentage of harmonic distortion, but the scope will show whether second, third, or higher harmonics are present.


Fig. 12-7. Scope identifies the nature of distortion.

When the positive peaks of a sine wave are clipped or compressed (Fig. 12-8), even harmonics are generated. The waveform is unsymmetrical. If both positive and negative peaks are clipped equally, the resulting waveform is symmetrical, and odd harmonics are generated.

Again, if positive and negative peaks are clipped unequally, both odd and even harmonics are developed. Any change in the shape of a sine wave, no matter how gradual, and regardless of the portion of the wave affected, generates harmonics.


Fig. 12-8. Severe even-harmonic distortion of a sine wave.

Parasitic oscillation is identified easily in scope tests. It causes a "bulge" on the waveform, usually at the peak. (See Fig. 12-9.) The bulging or ballooning interval consists of a high-frequency oscillation, generally occurring on the peak of drive to a tube which is being driven into grid current. When the grid is being driven positive, the grid input resistance falls to a comparatively low value. Stray reactances in leads and transformer windings can then "see" a high Q which permits a brief interval of high-frequency oscillation. Parasitic oscillation is commonly controlled by connecting small resistors in series with the grid and plate leads at the socket terminals.

Notch distortion, if appreciable, can also be seen in a scope pattern. This difficulty occurs principally in push-pull amplifiers which are incorrectly biased. This distortion is exhibited as irregularities in the shape of the sine wave in passing through the zero axis. Notch distortion is aggravated by high-level drive. Any push-pull amplifier develops this type of distortion when driven too hard. If the distortion occurs at rated power output, check the bias voltages at the push-pull tubes. If the bias is correct, . check for low plate or screen voltages.

SQUARE-WAVE TESTS


Fig. 12-9. Appearance of parasitic oscillation on a sine wave.

High-fidelity amplifiers are often rated for a square-wave response. A different class of information is provided by square-wave tests, which supplements the data from steady-state tests with an audio oscillator. The leading and trailing edges of a square wave are very steep, and therefore the rise and fall times of an amplifier become apparent. This is sometimes ref erred to as the attack time of the amplifier. It is a transient response, as contrasted to a steady-state response. In theory, it is possible to deduce the transient characteristics from a study of the frequency and phase response over the passband of the amplifier. Practically, however, this becomes almost prohibitively difficult, particularly when several audio stages are cascaded.


Fig. 12-10. Checking the transient response of a scope.


Fig. 12-11. Tilt in top of square wave is caused by low-frequency attenuation and phase shift.

Fig. 12-12. Low-frequency over-compensation causes top of square wave to slope uphill.

Most oscilloscopes have vertical amplifiers which exceed the capabilities of a hi-fi amplifier in square-wave tests, but this is not true at all. It is advisable, therefore, first to check the transient response of the scope, as illustrated in Fig. 12- 10. Any distortion over the contemplated range of square-wave test frequencies must be taken into account, so that it is not improperly charged to deficiencies in the audio amplifier. It is common to find tilt in the top of a 60-Hz square wave, as seen in Fig. 12-11. In an ac scope, the coupling, decoupling, and bypass capacitors in the vertical amplifier may be too small in value to reproduce a 60-Hz square-wave without tilt. Or the square-wave generator itself may not be free from tilt at low frequencies.

When a 60-Hz square wave is passed through the audio amplifier, any tilt contributed by the amplifier will be added to the reference waveform. It is much less common to find uphill tilt, as shown in Fig. 12-12. This response is caused by overcompensation of low frequencies. In audio amplifiers, the cause is usually traced to a defect in the feedback network. An open capacitor in some feedback circuits can result in more negative feedback at low than at high frequencies.

Theoretically, a 60-Hz square-wave test gives all the information about transient response which can be obtained. There is no reason why tests are required at higher square-wave frequencies. A practical difficulty arises, however, in evaluating the extremely high harmonic responses from a 60-Hz test. High harmonics have less amplitude, and their effect on the reproduced waveform tends to be masked by low frequency harmonics. Moreover, fine detail of corner reproduction is so highly compressed at low test frequencies that the display is not readily evaluated. Finally, attack time becomes plainly visible only at high test frequencies, when ordinary sawtooth deflection is used.

The meaning of attack time is seen in Fig. 12-13. It is the time required for the square wave to rise from 10 percent to 90 percent of its final amplitude. In order to measure attack time, advance the square-wave test frequency until the attack interval occupies a usable horizontal interval. Compare the attack interval with the total interval for one complete square-wave cycle. Knowing the frequency of the square-wave signal, its period (time of a complete cycle) is given by the reciprocal of the frequency. The attack time is given, in turn, by the fraction of the total interval occupied by the attack interval.


Fig. 12-13. Slow rise and fall time.

OVERSHOOT

Overshoot is a characteristic often associated with attack time (Fig. 12-14). An amplifier which has a very short attack time may, in turn, display objectionable overshoot. High-fidelity amplifiers are occasionally rated for overshoot at a specified square-wave frequency. This rating is given as a percentage. To measure percentage overshoot, compare the amplitude of the overshoot pulse with the total amplitude of the square wave between its flat-topped portions.

Causes of overshoot are a rising high-frequency response in the amplifier circuits (sometimes due to a defective feedback network), or to uncoupled inductance in transformers. Make certain that the amplifier under test is working into the rated load. Some amplifiers are more load sensitive than others. Overshoot may be accompanied by ringing, as shown by the severe situation in Fig. 12-15. When ringing is encountered, first check the feedback network. Defective transformers can also be responsible for this symptom.

Most audio transformers ring and otherwise distort a square wave if the test frequency is too high. Hence, a meaningful test is obtained only within the square-wave limits specified by the manufacturer. An improperly loaded transformer may also ring within its rated range.

Normal loading damps the windings and suppresses the response of uncoupled inductance and distributed capacitance. Therefore, in event of difficulty, check for circuit defects which may reduce normal loading, for even a high-resistance connection can be responsible.


Fig. 12-14. Overshoot occurring in a square-wave pattern.

Fig. 12-15. Severe ringing occurring in square-wave test.

Parasitic oscillation occasionally occurs in square-wave tests (Fig. 12-16) just as in sine-wave tests. If the output from the square-wave generator is reduced, the spurious oscillation will usually disappear.

It is eliminated in most cases by connecting 50-ohm resistors at the plate and grid terminals of the offending tube. If, due to a defective feedback network, suppression resistors will not help, the defective component must be located and replaced.


Fig. 12-16. Parasitic oscillation occurring in square-wave test.

SQUARE-WAVE TEST OF STEREO-MULTIPLEX ADAPTER

Because a multiplex adapter for fm stereo normally responds to an L + R and an L - R signal it cannot be directly energized by a square-wave signal for meaningful tests. Therefore, we must first use an fm multiplex generator to process the square-wave signal as shown in Fig. 12-17. The output from the square-wave generator is fed to the external-modulation terminals of the multiplex generator. In turn, the multiplex generator supplies a composite audio signal for testing the square-wave response and separation of the multiplex adapter.


Fig. 12-17. Test setup for square-wave check of multiplex adapter.

The composite audio signal appears as illustrated in Fig. 12-18; we see this waveform if the scope is connected directly to the output of the multiplex generator. Note that the waveform appears superficially the same whether the generator is switched to L-channel or R-channel output. The superficial similarity results from the fact that only a phase difference distinguishes one signal from the other.


Fig. 12-18. Modulation of composite audio signal with 1-kHz square wave.

When the composite audio signal is passed through the multiplex adapter, as illustrated in Fig. 12-17, the signal is processed by the adapter to develop the L-channel and R-channel signals. When the separation control on the adapter is adjusted correctly, maximum square-wave output is normally observed from the R channel, and practically zero signal from the L channel. Fig. 12-19 illustrates the ideal outputs from the L and R channels. In practice. we expect to see a small and distorted square-wave output from the left channel, instead of a straight horizontal trace. The separation control is adjusted for best separation.


Fig. 12-19. Ideal response of multiplex adapter to square-wave test. (A) Right-channel response. (B) Left-channel response.

It is customary to make a 2-kHz square-wave test. Of course, other square-wave repetition rates can also be used. At 60 Hz, we usually see a noticeable tilt in the square-wave reproduction. At high frequencies, we expect to observe some integration of the reproduced square wave. A 2-kHz square wave is normally reproduced with little noticeable distortion. Excessive distortion and/or poor separation points to a defective component in the multiplex adapter. Check the tubes first. If they are not defective, a faulty capacitor is the most likely cause.

Semiconductor diodes may not be well matched, which could cause poor separation. Resistors seldom become defective, but this possibility should not be overlooked. In transistorized adapters, a defective transistor could be causing trouble.


Also see our audio electronics website: "gammaelectronics.xyz"

Top of Page

  PREV.     Guide Index HOME