TO-264 packaged BJTs are commonly found in audio amplifiers and power electronics. We occasionally encounter transistors with more than three legs such as the onsemi NJL3281DG (NPN) / NJL1302DG (PNP) as shown in Figure 1. These transistors have five legs: three for the conventional transistor connections and two additional legs for the integral thermally coupled feedback diode.
This engineering brief reviews the operation of a Class-AB amplifier biasing with an emphasis on the thermal coupling between the output transistors and the biasing network. This feedback path is necessary to prevent thermal runaway of the amplifier’s output stage.
Figure 1: Image of the onsemi NJL3281DG NPN transistor featuring three legs for the transistor and two legs for the stabilizing thermal feedback diode.
Tech Tip: The output transistors in a Class-A amplifier are always on. A Class-AB amplifier is biased so the output transistors are slightly conducting at the Q point. In a Class-B amplifier both output transistors are turned off at the Q point.
Class-A amplifiers are inefficient as considerable power is dissipated by the output transistors. Class-B amplifiers are efficient but have high crossover distortion. Class-AB amplifiers provide a good compromise between high efficiency and low distortion.
What is thermal runaway?
Thermal runaway is a hazard for semiconductor devices such as the large transistors in an audio amplifier. The problem is an intrinsic positive relationship between temperature and conduction. Current causes the temperature to increase, which leads to more current. The result is an out-of-control spiral with the potential to destroy the transistor or trip the overtemperature or overcurrent protection if present.
How do we prevent thermal runaway?
A classic solution is to place a thermal sensing semiconductor on the heatsink as close as possible to the output transistors. This semiconductor is then used to automatically bias the output stage. This is a negative feedback loop that tends to stabilize the amplifier by counteracting the uncontrolled positive runaway thermal feedback loop.
Figure 2 presents the classic solution using transistor Q1 as the temperature sensing element. Think of transistor Q1 as a temperature-dependent variable resistor. As the temperature increases, the voltage as measured from Q1’s collector to emitter decreases. This results in a lower voltage as measured from base to base on Q3 and Q4. The reduced bias voltage will thereby reduce the current passing through Q3 and Q4.
Figure 2: Representative output stage for an audio amplifier. Transistor Q1, Q3, and Q4 must be thermally coupled.
Tech Tip: Bias is set (trimmed) when the amplifier is quiet with no output signal. For a class AB push-pull amplifier, the DC quiescent point (Q) places the transistors just above the conduction point. A multimeter will read approximately 1.5 VDC base to base across Q3 and Q4.
Observe that current flows from rail to rail through output transistors Q3 and Q4. The Q point is therefore a balancing act to place the transistors in conduction without entering thermal runaway with increasing wasted heat generated in the transistors Q3 and Q4.
We should mention that the Q point has a profound impact on amplifier distortion. It’s another balancing act. We could certainly reduce the Q point for Class-B bias to lower Q current and reduce the risk of thermal runaway. However, this would increase crossover distortion as there is a moment in time when both Q3 and Q4 are turned off.
Alternative biasing using diodes
Figure 3 presents another form of thermal regulation using a pair of thermally coupled diodes instead of a transistor. In this example, the diodes have similar characteristics to the transistor’s base-emitter junction. Just like the transistor example in Figure 2, the diode’s conduction will increase with increased temperature. This tends to lower the bias voltage applied to the output transistors, thereby thermally stabilizing the amplifier.
Figure 3: Representative output stage for an audio amplifier. Diode D1, D2 as well as transistor Q3, and Q4 must be thermally coupled.
Problems with thermal coupling
Both Figure 2 and Figure 3 have a thermal lag problem. The sensing element Q1 and D1 / D2 respectively are always thermally behind the output transistors Q3 and Q4. This is a simple recognition that the sampling points are in different locations. The transistors and thermal sensing element(s) are often separated by several inches on the heatsink. Consequently, the transistor die itself can become hot long before the sensing element knows there is a problem.
This can lead to instability where the system thermally oscillates about the setpoint, potentially increasing the crossover distortion for low-level musical passages. Biasing is also a time-consuming process. It’s taken me up to an hour to properly set the Q point with multiple set – wait – check cycles.
Thermal stability solution using integral diodes
The onsemmi JL3281DG (NPN) / NJL1302DG (PNP) transistors featured in Figure 1 provide a unique thermal stabilizing solution. Instead of using independent diodes, the diodes are integrated into the transistor’s body, thereby ensuring optimal thermal coupling between the transistor die and sensing diode.
The situation is further improved when we recognize that the onsemi designers have carefully controlled the characteristics of the thermal sensing diode to match the transistor. This eliminates the need for variable resistor R1 as shown in Figure 2. The need for time-consuming bias trimming is also eliminated.
Tech Tip: The integral diode design solution is accompanied with an intrinsic predetermined biasing, where the Q point is determined by the onsemi engineers matching between the output transistor and sensing diode. This may be problematic for users who desire greater bias control such as when an amplifier is placed into full Class-A mode. In these situations, designers may prefer traditional transistors such as the closely related MJL3281AG (NPN) and MJL1302AG (PNP).
Parting thoughts
The transistor with integral thermal sensing diodes provides an excellent case study for amplifier design and may be perfect for your next circuit design. With tight thermal coupling, the thermal lag problem is minimized leading to improved amplifier stability and reduced service time.
This article assumes a conventional output stage configuration. Perhaps another day we can explore the Sziklai pair.
Kindly give a thumbs up if you found this article useful. Also, let us know if there are any related topics you would like to explore.
Best wishes,
APDahlen
Related information
Please follow these links to related and useful information:
About this author
Aaron Dahlen, LCDR USCG (Ret.), serves as an application engineer at DigiKey. He has a unique electronics and automation foundation built over a 27-year military career as a technician and engineer which was further enhanced by 12 years of teaching (interwoven). With an MSEE degree from Minnesota State University, Mankato, Dahlen has taught in an ABET-accredited EE program, served as the program coordinator for an EET program, and taught component-level repair to military electronics technicians. Dahlen has returned to his Northern Minnesota home and thoroughly enjoys researching and writing articles such as this.