This simple dc circuit breaker can protect a power supply from the ever-present screwdriver or even isolate a satellite's dual power system from a short circuit in one subsystem. The breaker resets itself whenever the fault goes away or whenever you cycle the power.

Fig 1a shows how the heart of the breaker works as an on switch at normal currents. When the supply applies a small dash of forward current to the breaker's plus and minus terminals, the current turns on Q₂ a smidgen. Q₂ then turns on Q₃ a lot, which turns on Q₄ full blast. The whole circuit acts as a glorious saturated-transistor short circuit between the two terminals. The resulting forward voltage across the terminals is little more than one V_BE-the standard forwarding fee for turning on a transistor.

This simple circuit poses nasty analytical problems. The transistors' arrangement in Fig 1a is 3/2 of a Middlebrook connection, which alternates pnp and npn transistors driving each other for a large overall current gain. Its new on-resistance-roughly R₁ divided by all three transistor current gains-can be tiny. But, the transistors are well into saturation, reducing their gains by distributing current in complex ways. Rather than asking what resistance you get from these unknown reduced gains, ask yourself what equal gains (B) a given resistance would require-say 1(ohm).

Saturation curves for a gain of 15 can give you a rough idea of what to expect. For more accuracy, consider that Q₄ may not be quite as saturated as the other transistors because it drives no following V_BE.

Fig 1a is like an SCR switch that's on all the time. To make a circuit breaker, you need only turn the switch off at the right current level. Fig 1b shows the whole scheme, complete with turn-off circuit and minor enhancements.

Voltage divider R₂/R₃ at the "sense" end of the chain applies half the terminal voltage (or whatever portion you want) to Q₁'s base. A large-enough current through the circuit's plus and minus terminals eventually forces the terminals' SCR-like saturation voltage to rise to a threshold value where the voltage divider R₂/R₃ turns Q₁ on a bit. Q₁ starts to steal current limiter Q3's current from Q2, turning off the whole works in a flurry of inspired circuit-breaking.

Only tiny current paths remain through R₃ and CL₃. The open-circuit terminal voltage holds off the circuit breaker via R₃ until the worker removes his/her screwdriver or you turn the power off, allowing the terminal voltage to drop back to a small value.

The current-limiter CL₃, although not vital, provides a lower on-resistance than a resistor without added open-circuit power. Adding or subtracting one transistor from the cascade lets you switch transistor sexes at one end of the chain. The collector of sensor transistor Q₁ never "sees" more than the V_BE of the next transistor, Q₂. Therefore, Q₁ can be a low-voltage, high-gain, small-signal device. Base resistors R₄ and R₅ absorb the small collector leakage currents from each previous stage so that each stage stays off when the circuit breaker trips. Leave out the resistor if you must take chances, but the cascaded current gain can be so large (as high as 1M) that a little I_CEO could turn the breaker on when you want it off.

The last transistor in the chain, Q₄, carries the most current; yet, it may be furthest from saturation because no following state's V_BE diminishes its collector voltage. If the trip point must be more predictable, a small resistor in series with Q₄'s collector, R₆, in Fig 1b guarantees that the V_CESAT is where you want it to be-more or less. Don't put a wimpy transistor at the end of the chain and a power-hungry, low-V_CES transistor just before it. The powerful stage could hog most of the current. Resistor R₇, in Fig 1b, helps guard against "current hogging" by increasing the V_CES and the V_BES of Q₃. You may not need resistor R₇. Zener diode D₁ saves the circuit from the person who wires it in backwards and inductive kickbacks from the load

The uncompensated trip point tracks the 0.3%/°C decrease in Q₁'s V_BE-a good thing sometimes. Ordinary circuit breakers and fuses display much worse temperature-related behavior. The designer could replace R₂ or R₃ with a clever thermistor array or a band-gap circuit.

C₁ reduces trip speed. If EMI gets into this high-gain, cascaded chain, the EMI could turn the breaker off before its time. If you need a slow-blow breaker, be my guest and make C₁ as high as you need. Redesigning this circuit for MOSFETs is left as an exercise for the reader. (DI#1672)