FFT spectrum analyzer for audio amplifier design

Please note this application note is a few years old and makes use of the now discontinued ADC-216 oscilloscope running an old version of PicoScope.  To repeat this experiment today we would recommend using the PicoScope 4262 16-bit oscilloscope, which has a low-distortion signal generator built in so avoiding the need for the Blackstar unit mentioned below.

Continuing our audio experiments with the ADC-216 spectrum analyzer, we decided to show how the ADC-216 can be used to aid the amplifier design process.

For all of the following tests, we used a high quality signal generator from BlackStar. The PicoScope trace below shows a pure 1 kHz tone from the signal generator (Figure 1).

The diagram below shows a basic power amplifier circuit. The output stage shown is the simple and widely used emitter follower topology. The following tests show how the quality of the signal from output stage can be monitored. Modifications can then be made to the circuit and any improvements recorded. This output stage topology has a stage gain of just less than 1 so it can be easily moved outside the feedback loop as in the circuit below (Figure 2).

Initially the circuit was constructed as above. The ADC-216 was connected to point ‘A’ in the circuit and the signal generator was connected to the input of the circuit. The PicoScope screenshot below shows the signal at point ‘A’. It is clear that the op-amp is doing a reasonable job as could be expected with such a large amount of negative feedback applied (Figure 3).

If we now look at the output at point ‘B’, we can see excessive crossover distortion clearly visible on the scope trace. The harmonic information also indicates the problem with the third harmonic component being the largest of the harmonics. The load resistor connected to point ‘B’ was 2k2. It is obvious that the output stage is suffering from severe crossover distortion effects (Figure 4).

Sig gen

Figure 1

FFT spectrum analyzer for audio amplifier design

Figure 2

FFT spectrum analyzer for audio amplifier design

Figure 3

FFT spectrum analyzer for audio amplifier design

Figure 4

If we continue to monitor point B but move the output stage inside the feedback loop by connecting the inverting input of the opamp to point ‘B’, we notice a massive reduction in the output stage distortion. Indeed, if you were looking at just the scope trace on a conventional scope, you would not see any problems. This is where the power of the Pico Spectrum Analysis combined with the sensitivity of the ADC-216 are crucial. The problem is still evident in the spectrum view (Figure 5).

If we connect the ADC-216 back to point ‘A’ but keep the output stage in the feedback loop, we can see what kind of corrections the opamp is having to apply in order to remove the error generated by the output stage. The opamp follows the sinusoidal signal for the positive and negative peaks of the waveform but around the crossover points, it has to work very hard to cover up the output stage error. It has to quickly traverse the point where the output stage does not conduct. It is clear then that the opamp needs to have a much higher slew rate, than that which text books may initially suggest, in order to compensate for the poorly designed output stage. This makes the design of earlier gain stages much more difficult and expensive than they need to be (Figure 6).

Next the output circuit was improved by applying a bias voltage between the bases of the output devices using the simple diode drop technique. The view below shows the output at point ‘B’ again. It can bee seen that the addition of a small amount of biasing has improved the THD reading by almost 10 dB (Figure 7).

If we look again at point ‘A’ with this added bias voltage, we can see that the op-amp does not require such a high slew rate since it does not have to work as hard to cover up the output stage imperfections (Figure 8).

This tech note has described a simple application whereby the performance of an amplifier output stage design can be evaluated using the powerful FFT spectrum plot in PicoScope. With some experience, many problems can be identified from their characteristic spectrum plots which might otherwise go unnoticed but lead to a poor or coloured sound reproduction.

FFT spectrum analyzer for audio amplifier design

Figure 5

FFT spectrum analyzer for audio amplifier design

Figure 6

FFT spectrum analyzer for audio amplifier design

Figure 7

FFT spectrum analyzer for audio amplifier design

Figure 8