This month's caller had a burning issue. Well OK, no open flame, but there was a severe red spot on his finger where he had touched the device. The conversation started with the brief description, “I built this breadboard and the amplifier is getting awful hot.”

“Where does all this heat come from?” he continued.

In all cases there were two elements in the heat system. As we discussed last month, the self-heating associated with the device quiescent current can be significant. This is especially true in the some surface mount packages. The second source of heat in the amplifier is from the product of the current delivered to the load times the voltage across the output stage.

Most callers that have this type of problem get all tied up in the power being delivered to the load, where the problem is in the power left in the amplifier. To illustrate this point look at the graph in Figure 1. This shows the three powers associated with any amplifier circuit. As the voltage across the load increases, the power in the load increases as the square of the voltage. Remember the relationship: P = V2 /R. This describes the load power and is seen as the dotted line. Note that this is true only for resistive loads, reactive loads demonstrate a more complex result.

The power from the power supply is simply the product of the voltage times the current. This is shown as the straight, solid line. The power of concern is the power left in the amplifier. This is given by the relationship: P = (Vs -VL )2 /R. Notice that this curve peaks at 50% of the supply voltage. Just to confirm this result, consider the boundary conditions. At zero load voltage the full supply is developed across the output stage but there is no current. At full output the load current is maximum but the voltage across the output is minimum. In both cases the power in the output stage is minimum.

Is this a concern? It can be, as it was in the case of my caller. The solution is in the techniques available to get the heat away from the device. The classical power devices were packaged in the TO-3 metal can or an extension of the TO-220 plastic package with the heat tab. In both cases the metal surface of the package was usually bolted to a part of the chassis. Modern power packages are using the printed circuit board to channel the heat away from the device. In classic designs, where the power devices were fastened to a chassis piece, the ambient temperature was usually the surrounding air. With the newer power packages that use the PCB as the heat radiator there are a lot more concerns, such as the heat generated by neighboring parts. Additional concerns include the topology of the components on the board. Can neighboring devices shadow the power device from cooling air flow? I have seen cases where the final card works well in the lab but once it is placed in the final package a shielding layer provided insulating action and the heat is trapped.

The new power packages allow smaller packages but also need more attention to the escape path for the heat.

klein_bill@ti.com
And be sure to copy, what's-his-name, sohr@cmp.com