Just about all analog engineers (and even a few digital ones) are familiar with the standard op amp. Originally developed to perform mathematical operations in analog computers (hence the name), it’s become one of the most widely-used electronic devices and is available as a standalone component, integrated into other analog chips, and even as a peripheral in a microcontroller.
For many small-signal applications, you can treat a current-generation op amp almost as an ideal device, thanks to seventy years of continuous improvement. But an op amp designed for high power – ah, now that’s a horse of a different color. Apex Microtechnology offers power op amps that can deliver up to 9A from a supply voltage of 350V and op amps that can deliver up to 50A from a 200V supply. At those power levels, you need to pay attention to a few key details if you don’t want your design to suffer a (literal) meltdown.
Let’s take a look at some of the factors that can contribute to a successful design.
Stay Within the SOA
The SOA (Safe Operating Area) graph for a power op amp defines the acceptable limits of its power handling capability. If operated outside the SOA, the op amp is likely to be permanently damaged or destroyed.
Figure 1: The SOA graph displays the limits of an op amp’s power capability (Source: Apex Microtechnology).
Figure 1 shows a typical SOA graph. The horizontal axis shows the difference in voltage between the supply voltage VS and the output voltage VO; the vertical axis shows the output current drawn from either the positive or negative supply.
There are three basic power limitations:
- Current handling capability. This horizontal line at the top of the graph represents the limit (5A in this case) on the output current imposed by the current handling capabilities of the bond wires, the die junction area and thick-film conductors.
- Power dissipation capability. This refers to the amplifier output stage. Note that the product of output current on the vertical axis and VS–VO on the horizontal axis is constant over this portion of the SOA curve. For TC = 25°C, this line represents the maximum power dissipation capability at maximum junction temperature using an “infinite heatsink.” If you find one, let us know. The line varies with case temperature; the power derating curve in the datasheet gives the maximum internal power dissipation for a particular case temperature.
- Second Breakdown. This limitation applies only to bipolar transistors When a bipolar transistor simultaneously experiences high collector-emitter voltage and high collector current, non-uniform current density in the emitter causes localized heating and “hot spots” at the junction. The temperature dependence of junction current results in increased current density at the hot spots; this increases the temperature, which further increases the current density and so on, eventually leading to thermal runaway and transistor failure. MOSFET power transistors do not suffer from second breakdown because their ON-resistance increases with temperature.
Let The Heat Out: Thermal Management
It’s also important to practice good design practices for thermal management to remove as much heat as possible from the power op amp die to minimize the operating temperature and increase reliability. The failure rates of both bipolar and MOSFET power transistors increase significantly for junction temperatures above the 25°C baseline. A further goal is to minimize the effects of the removed heat on other devices.
Many components contribute to effective heat removal. The heatsink is a key component, but read its datasheet carefully. There are multiple variables that can affect the thermal ratings including power level, air or liquid flow rate, and even orientation.
The heatsink is only one link in the chain. Non-compressible thermal washers, for example, are mounted between the op amp and the heatsink to improve thermal conduction relative to a bare joint; a substrate made of Kapton (a temperature-stable polyimide film) adds electrical isolation, and a compression washer is used under the mounting screw.
Figure 3: Removing the heat from a power op amp is a multi-faceted process that requires several thermally-efficient components. The recommended mounting technique for a PCB is shown here (Source: Apex Microtechnology).
The maximum allowable internal power dissipation of the power amp depends on a number of factors including the maximum allowable junction temperature (TJMAX); the actual ambient temperature (TA); and the thermal resistances of the heatsink (ѲHS), the isolation washer (ѲHSC), and the power amp junction to its case (ѲJC). The equation is:
PMAX = (TJMAX — TA)/(ѲJC + ѲHS + ѲHSC)
The Apex Power Design software tool (described below) includes thermal interface ratings between cases and heatsinks.
Stabilize The Loop (And The Non-Loop)
A power op amp often must drive a load such as a solenoid or a motor that is not simply resistive and has capacitive and inductive elements.
The characteristic of such a load varies with frequency: in particular, a load which is has a large capacitive component can pose a problem. In a pure capacitor, the voltage lags the charging current by 90 degrees, which introduces a large phase delay in the loop response of the op amp’s feedback circuit. This can contribute to instability, which can lead to ringing or even oscillation.
Following standard feedback theory, the closed-loop op amp circuit can oscillate if the loop gain (1/ß) is greater than or equal to one (0dB) at a phase shift of 180 degrees. A robust design will change the response of the feedback circuit or op amp to reduce the phase shift so that this condition never occurs. A good rule of thumb is to design for 135 degrees of phase shift at 0dB gain; this constitutes a phase margin of 45 degrees and allows for safe operation during transient conditions.
These circuit modifications are referred to as compensation. Phase compensation, for example, adds a capacitor to reduce the op amp’s open-loop gain (AOL) vs. frequency curve. Many small-signal op amps include phase compensation internally, so you have limited room to make changes, but Apex power op amps usually allow you to add an external phase compensation capacitor, as shown in Figure 2.
Other compensation methods include feedback zero compensation, noise gain compensation, and isolation resistor compensation. These are discussed more here.
Figure 2: Phase compensation stabilizes the loop by adding a capacitor CC to move the first pole of AOL, the open-loop gain (Source: Apex Microtechnology).
Testing Loop Stability
There are a few standard tests that can determine if your design is stable. For example, a small-amplitude square wave can be used as the source voltage, or a small-signal AC square can be superimposed on a low-frequency sine wave to test dynamic stability.
Of course, for best results you should perform the tests over the full range of both output current and operating temperature. You can find out more about techniques to stabilize the power op amp loop in Apex’s application note AN47.
Even if you take all necessary steps to ensure the stability of your feedback loop, other factors such as printed circuit board (PCB) layout, power supply bypassing, and grounding can still trip you up. These are often termed “Non-Loop Stability” factors; you can explore these in AN19, “Stability for Power Operational Amplifiers.”
Optimizing Your Design For Power
To deliver the most output power and achieve maximum efficiency, you should minimize the power amp’s internal power dissipation. The internal power dissipation of a power op amp is the sum of its quiescent power plus the product of output current and the supply-to-output differential voltage, VS – VO. Remember the SOA graph?
The application usually defines the power requirements of the load, but calculating the power dissipated inside the amplifier is not always simple. For a purely resistive load, the maximum power dissipation occurs when the output voltage equals half the supply voltage. When performing analysis, this is the worst-case condition if the amplifier does not have to withstand short circuits.
One parameter that you do have the freedom to change is the power supply voltage.
In general, the power supply voltage should be the minimum necessary to produce the required output, but be careful. A small-signal rail-to-rail op amp driving minimal current may be able to swing to the supply voltage, but a power op amp has a minimum supply-to-output differential voltage, which is a function of the output current. The 150MHz 5A peak PA09, for example, can swing to ±VS – 7V for IO = 2A.
Make sure your power supply voltage meets this specification but is otherwise no higher than needed. Each extra volt here is one more watt that must be dissipated per amp of output current.
Along the same lines, many applications don’t require driving the load with a symmetric signal. If that is the case, consider using an asymmetric power supply. Most power op amps can operate well with single supply or asymmetric power supplies.
More information on optimizing output power can be found here.
Conclusion – Need Help?
If you’re new to analog and power design, you’ll notice that there a lot of factors to consider, and this article only skims the surface of a very deep ocean.
Luckily, help is at hand. In addition to a comprehensive library of application notes, Apex offers the free Power Design software tool that helps automate many of the calculations you need when designing with high power linear and PWM amplifiers.
The design tool includes multiple modules that let you plot load-lines and current limits directly onto the SOA to ensure circuit stability; calculate the internal power dissipation and heatsink requirements; dynamically select a part, and assist with other power analog circuit design tasks.