A critical aspect of power supplies is dc supply-voltage stability. In addition to powering circuits, a precision dc voltage is often used as a reference to which voltage signals are compared for decision making. If reference voltage fluctuates wildly, so does the decision making of the system.

To solve this problem, one large voltage regulator is often located at or near the power source. However, one problem with this approach is the voltage drop along the supply line caused by wire or printed-circuit resistance. The resulting variable voltage throughout the system often degrades regulator performance. Another problem with using a single regulator is the various impedances in the system. Common impedances between terminals provide paths for unwanted coupling. This requires added decoupling and bypass circuitry which, in the long run, can degrade local regulation. These effects can be reduced or eliminated by using individual regulators at each critical circuit location.

The need for a voltage regulator is determined by the load and voltage supply, rather than supply alone. Selecting the correct regulator requires a thorough knowledge of load requirements. Several questions should be answered:

  • What is the maximum range of voltages over which the system must perform?
  • How much current does the system require?
  • How much ripple voltage can be tolerated?
  • How much voltage drift with temperature is acceptable?

IC regulators: A temperature-compensated reference voltage is developed on the IC and compared with the output voltage of an error amplifier in IC regulators. This amplifier has low-temperature drift to maintain good output-voltage stability with changing temperature. The error amplifier drives an output stage consisting of a Darlington pair. The chips also contain the necessary bias and protection circuits including short-circuit protection, thermal shutdown, and pass transistors.

Unlike other ICs which operate at minimum power levels, voltage regulators are generally operated at or near their power limits. For this reason, the maximum power dissipation PD(max) is a crucial parameter. The power dissipated is in the form of heat and is a function of the device package, junction temperature, and ambient temperature. (Junction temperature is the average temperature of the monolithic IC chip located inside the regulator package.) Maximum permissible junction temperature TJ(max) is specified in device data sheets.

Other related thermal specifications are thermal resistance from the junction to the case θJC and to ambient (room air) θJA, thermal resistance from the case to the heat sink θCS, and thermal resistance from the heat sink to ambient θSA.

For optimum circuit performance, junction temperature of a semiconductor must be kept at or below its maximum rating. Since reliability of semiconductors improves as the operating junction temperature decreases, heat sinks are often used to help dissipate heat. To determine if a heat sink is necessary, first determine junction temperature of the device from

TJ = TA + θJAPD = junction temperature, °C; TA = ambient temperature, °C;JA = junction-to-ambient thermal resistance; and PD = power dissipation, W.

Actual power dissipation PD is obtained from PD = IOUT (VIN - VOUT) + VINIQ, where VIN = input voltage; VOUT = output voltage; IOUT = output current, A; and IQ = quiescent current, A (standby current, no load).

It is good practice to keep TJ about 25°C below the maximum value specified by the manufacturer. Typically, these maximums are 150°C, therefore, TJ should be kept at or below 125°C.

Maximum recommended power dissipation is determined by PD = (TJ - TA)/θJA.

Therefore, external heat sinking is necessary if IOUT (VIN - VOUT) + VINIQ (TJ - TA)/θJA.

To determine the amount of heat sinking required, substitute (θJC + θCS + θSA) for θJA and solve for θSA.

Junction-to-case thermal resistance, θJC, is supplied by the semiconductor manufacturer. Case-to-sink thermal resistance, θCS, and sink-to-ambient thermal resistance, θSA, are generally furnished by the heat-sink supplier. This is difficult to determine analytically because it is a function of so many variables, such as contact area, contact pressure, interface materials between the IC and the heat sink, and whether convection or forced-air cooling is used.

Power supplies should not be connected in parallel to provide higher output current unless allowed by the manufacturer. A few millivolts difference in output voltage of two power supplies connected in parallel can cause an unbalanced condition: the power supply having the higher voltage delivers most or all of the current while the other power supply idles. Moreover, current from the higher-voltage supply may flow into the other supply and damage it.

Power-supply efficiency is the ratio of output to input watts. The difference between input and output power is dissipated by a power supply as heat. Unregulated, ferroresonant, and switching-regulator power supplies have efficiencies as high as 85%. Series-regulated supplies are considerably less efficient (typically 40%), but offer better regulation and lower ripple than other types.