Design Ideas: June 9, 1994

High-speed ADCs sometimes place very strict limits on their allowable input-voltage range, limits that generally require the use of input clamps. Some applications require very temperature-stable clamps that don't interfere with the speed of a 50-MHz amplifier. Fig 1 shows one such circuit that clamps positive-going voltages to 1.25V at 10 mA, and you can easily adapt it for negative-going voltages and different voltage and current levels. Recovery from the clamped state occurs in about 1 nsec.

This circuit improves on the conventional diode clamp in three major areas. First, the clamping voltage drifts at only a fraction of the -2 mV/C figure of a simple diode clamp. For even lower drift, you can replace the transistors with a monolithic matched pair. Second, by using the emitter of a transistor as the clamping element, the circuit easily obtains a low clamping impedance over a very wide range of frequencies. Third, Q₁ recovers from the clamped state much faster than does a PN junction diode because there is no excess charge storage.

The op amp in the reference-circuit servo controls Q₂ to establish 1.25V at the emitter while carrying exactly 10 mA. The circuit also applies the base voltage of Q₂ to Q₁. Assuming reasonable matching of the transistors, Q₁ clamps V_OUT at exactly 1.25V when the clamp reaches 10 mA. This occurs when V_IN reaches 2V. (Transistor base currents are assumed negligible in this analysis.) The 47-pF capacitors reduce the clamping impedance of Q₁ at high frequencies.

You can easily modify the clamping voltage and current by changing V_CLAMP and R_BIAS. The circuit can clamp negative-going voltages using NPN transistors in place of Q₁ and Q₂ and changing the supply-voltage polarity. You can combine NPN- and PNP-based clamping circuits for precision clamping in both directions. When using this circuit, be sure to consider the emitter breakdown voltage of Q₁. (DI #1445)