Darlington Push-Pull Follower Circuit with Negative Feedback

Negative feedback is an important idea within transistor amplifier circuits. By feeding the output signal back into the input, the amplifier can continuously correct itself, improving stability, reducing distortion, and allowing the output to more closely follow the input signal.

Previously, we reduced crossover distortion in the push-pull follower stage using a transistor bias network. While the circuit already performed reasonably well, it was still missing a proper feedback system and sufficient output drive.

In this article, we’ll improve the push-pull follower by adding a differential amplifier feedback stage and a push-pull Darlington transistor output stage, resulting in a more stable and practical amplifier circuit.

Reviewing the Previous Push-Pull Follower

In the previous circuit, we built a relatively simple transistor push-pull follower circuit, as shown in Figure 1. It consists of three transistors and a few common components.

Of course, it also uses a dual-rail power supply commonly used in push-pull circuits. In this case, it’s a split ±12V power supply.

That circuit already has the following features:

• A complementary push-pull output stage with no DC-biased output.
• A bias network for reducing crossover distortion.
• A pre-driver stage for offsetting the output VBE drop.
• An input voltage divider for controlling signal amplitude.

Based on these features, it already performs well as a voltage follower circuit. In addition, we’ve tested it, and it works quite well in practice too. You can read more about this circuit in our previous article.

However, there are some remaining limitations we’ve yet to address, such as:

• No negative feedback and fixed pre-driver voltage gain.
• Slight signal distortion and non-ideal gain stability.

While these limitations are relatively minor, they still affect the overall performance and stability of the circuit. We can remedy these limitations by adding a negative feedback system in the form of a differential amplifier stage.

Adding Negative Feedback Stage (Differential Amplifier)

In this iteration of the circuit, we’re replacing the previously used input-amplitude divider with a differential amplifier stage. There are also a few other small additions, as shown in the full schematic in Figure 2, but other core components remain the same.

Q1, Q2, R1, and R2 form a differential amplifier stage operating in a differential mode. We won’t go too deep into the theory behind differential amplifiers, but in simple terms, this stage introduces negative feedback that constantly corrects the output signal.

Q1 receives the input signal, while Q2 receives the feedback from the circuit’s output. The differential pair then creates a control signal based on the difference between these two signals. This resulting difference then drives the pre-driver stage (Q3). As a result, any deviation or difference between the input and output is constantly reduced.

In other words, the circuit continuously adjusts its output so that the feedback signal matches the input signal.

This allows the circuit to stabilize itself and produce an output that closely follows the input signal without requiring manual adjustment of the input amplitude, like in the last circuit (Figure 1).

Other additions include the C3 and C4 decoupling capacitors. They help reduce supply noise and improve overall stability. R4 is a feedback resistor, drawing output voltage to the differential pair’s feedback input.

R6 and R7, placed between the push-pull transistors’ (Q4 and Q5) emitters and the output node, help reduce the temperature sensitivity of the two transistors. As the transistors conduct more and more current, a larger voltage forms across R6 and R7, which in turn reduces the base-emitter voltage.

Lastly, we added a loading resistor R8 connected between the differential pair’s input and ground. It provides a defined DC reference for the input node, preventing it from floating and improving stability. Together with C1, they also form a high-pass filter.

Additionally, R8 sets the input impedance of the differential stage. As a result, its resistance depends on the input signal and the previous stage. If the value is too low, it can heavily load the previous stage and distort the waveform. In our case, we found that 75kΩ works well.

Other components were kept the same as the previous version (Figure 1), so the effect of the differential amplifier stage could be observed more clearly. 

Testing The Negative Feedback Stage

Now, let’s test the circuit shown in Figure 2 on a breadboard to see how it performs. First, we assembled the circuit onto the same breadboard, as shown in Figure 3.1.

Afterward, we apply the 1.1kHz, 10Vp-p test signal to the input. The oscilloscope reading of the output shows a clean sine wave signal with a peak-to-peak voltage of 10.22Vp-p. The output being slightly higher than the input is likely caused by a small voltage gain in the circuit (gain > 1) or oscilloscope measurement tolerances.

So far, this circuit works very well. Without any manual adjustment, the output already closely follows the input. This characteristic is a result of the constant signal correction provided by the differential stage.

Next, we add a 560Ω resistor load to the output. Even with a load in place, the output shows no noticeable change. It remains at approximately 10.22Vp-p, nearly identical to the unloaded condition, as shown in Figure 3.2.

Then, we try replacing the 560Ω load resistor with a 100Ω one. With the load resistance decreased, the output waveform drops to about 7.95Vp-p. The waveform also exhibits clipping at the positive peaks, as shown in Figure 3.3.

While BD140 and BD139 are perfectly capable of driving low-resistance loads, such as a 100Ω resistor, this configuration does not provide sufficient base drive to fully drive the output stage under such a load.

As a result, the output became limited, leading to waveform clipping as we see in Figure 3.3.

Anyhow, this push-pull follower circuit is still a big step forward from the previous iteration. It manages to solve the signal instability and greatly improves ease of use, bringing this design that much closer to being a practical amplifier stage.

Adding Darlington Pair

A simple way to increase the output load drive is to use the Darlington transistor. This will introduce two additional transistors to the output push-pull stage—one for the PNP and one for the NPN halves.

How it works is quite interesting, but put simply, a Darlington transistor (or Darlington pair) is a configuration whereby two transistors act as a single transistor with significantly increased current gain (β).

We essentially combine two transistors so they behave like a single transistor with much higher current gain, and thus more driving capability. In theory, the total gain (βTotal) is approximately the product of the two transistor gains.

If we apply this Darlington pair to the push-pull follower stage, we’ll get the following circuit shown in Figure 5 with an NPN Darlington pair on top and a PNP Darlington pair on the bottom.

Notice that the two transistors in the Darlington pair are different. The first transistor is a smaller BC549 and BC557, while the second transistor is a bigger BD139 and BD140 used previously. Why is that the case?

This is because the first transistor does not need to drive as much current as the second one. It only needs enough gain to fully drive the bigger power transistor, which then drives the load. As a result, the first transistor can be much smaller.

Considering BC549 and BC557 have about 90 (hFE) gain and BD139 and BD140 have around 70 (hFE), the theoretical total effective gain is more than sufficient for this usage (90 × 70 = 6,300).

New Bias Network

We also adjust the bias network slightly. This is because the bias network (D1 and D2) was designed for a normal transistor setup, providing +0.65V to the upper transistor (NPN) base and –0.65V to the lower transistor (PNP) base. 

This works for a regular push-pull follower setup, but not for a Darlington pair, which requires roughly twice the base-emitter voltage to begin conducting (~1.3V).

As a result, we need a new bias network that can set the base-emitter voltage of each Darlington pair to about 1.3V. A simple solution is to add more switching diodes (1N4148), two more to be more specific.

If one 1N4148 diode has about a 0.65V drop across it, two diodes will bring it up to 1.3V. So, two 1N4148 diodes set the base of the NPN pair to about +1.3V, while another two 1N4148 set the base of the PNP pair to –1.3V.

However, instead of four diodes, we decided to replace one of them with a potentiometer. The potentiometer allows for finer external control of the bias current and the resulting voltage drop across the bias network.

The full circuit with the additions of D3 and VR3 to the push-pull stage bias network is shown in Figure 5.

Caution: We also found during our testing that if the resistance (VR3) is too high, the base voltage will also be too high, causing the transistors to conduct significant current and heat up. In this case, 3kΩ seems to be the limit, so we picked a 5kΩ pot for VR3 and made sure not to turn it higher than 3kΩ.

Testing The Darlington Push-Pull Follower

Let’s test the circuit shown in schematic Figure 5 on a breadboard and see how well it performs.

With a proper bias network, when driving a 100Ω, the output peak-to-peak voltage shows a 10.38Vp-p signal with no sign of “notch” or discontinuity in the waveform, as shown in Figure 6.2.

The signal does not exhibit visible noise or instability. It is also unaffected by a 100Ω load. Although the temperature of the power transistors (BD140 and BD139) rises a bit during prolonged operation, a small heat sink would remedy this issue.

With the addition of a push-pull Darlington output stage and improved bias network, this push-pull follower can now drive significantly heavier loads while maintaining good waveform quality and stability.

Conclusion

In summary, the final push-pull follower circuit shown in Figure 5 functions well in practice. The circuit is also on a rather simple side, using only a few transistors and common components.

However, there are a few drawbacks worth mentioning, namely its limited frequency response. At lower frequencies (<500Hz), the output amplitude begins to fall because of the coupling capacitor acting as a high-pass filter. Its low-frequency reactance reduces signal transfer.

On the other hand, at higher frequencies (>3kHz), the output waveform starts to distort due to parasitic capacitances. They introduce a phase shift and slow down the circuit’s response. In a feedback amplifier system, these effects can skew a sine wave signal, forming a sawtooth-like wave instead.

These parasitic capacitances may come from the breadboard, less-than-ideal wiring, or the junctions of the transistors themselves. As we test all of these circuits on a breadboard, it’s likely to be a major source, since it is known to introduce a lot of parasitic capacitance.

On the next occasion, we may use these ideas to create a proper audio-ready amplifier for wideband or higher-frequency applications.

---