Crowbar Methods Protect Overvoltage Transients

June 1, 2002
Sensitive circuits are prone to failure if stressed by overvoltage transients. In some cases, the damage can be extensive and expensive.

For the JPEG version of this article, click here.

Sensitive circuits are prone to failure if stressed by overvoltage transients. In some cases, the damage can be extensive and expensive. The source of the overvoltages is legion and unpredictable, ranging from nearby lightning strikes to regulation system failures. Fortunately, the power regulating system often protects the load from the transient — but not always.

To meet the protection needs of sensitive circuits, overvoltage “crowbar” methods are often used. Fig. 1 shows the normal crowbar arrangement. In the crowbar method, a fast-acting voltage sense circuit detects the overvoltage and acts to turn on a powerful SCR connected across the supply lines to the sensitive load. Should a voltage transient occur, the SCR essentially pulls the voltage to near zero and protects the load from further stress. A fast fuse in the supply line will clear, should the power supply fail to current limit correctly. To prevent the SCR from failing, the I2t rating of the SCR must exceed the I2t rating of the fuse.

A conflict exists with this method. If the protection is made very fast to fully protect the load, it has a tendency to fire spuriously on small, low-energy transients, shutting down the load unnecessarily. The supply must then be cycled off/on to reset the SCR. On the other hand, slowing down the response time may permit a damagingly large transient to get through to the load. Normally, an unsatisfactory compromise between the two extremes is necessary.

Fig. 2 shows a method providing full and fast protection, while eliminating overvoltage and nuisance shutdowns.

R1 and R2 set the clamp voltage: in this example 7V. U1 (a TL431) detects any overvoltage and turns on progressively to pull down on the gate of a power P-channel FET, Q1. As the FET turns on, current flows in Q1 and R6 (R6 is 0.1Ω). Q1 and U1 act as a shunt regulator, with negative feedback from R4 and R5 to prevent oscillation of Q1.

The clamping action provided by U1 and Q1 is sufficient to effectively clamp low power transients to a safe value, without the voltage on R6 becoming high enough to fire the SCR. For example, a value of 0.1Ω for R6 would allow a clamping current of up to 6A before the SCR turns on. Even at 6A, the R7-C1 time constant delays the SCR action. If the current is greater than 6A, the voltage on R6 increases further, and the SCR turns on more rapidly, removing the stress from Q1. Finally, for currents above 12A, the alternative path through D1-R9 opens up to fire the SCR immediately, bypassing the R7-C1 delay.

We can maintain a fast-acting clamp action up to 6A in this example, preventing shutdowns for lower stress conditions, while retaining the powerful crowbar protection action for high-stress conditions above 12A. For lower voltage requirements, a small “N” type FET replaces U1.

Keith Billings is the President of DKB Power Inc., ([email protected]), and author of the Switchmode Power Supply Handbook. He will present Abe Pressman's “Modern Switching Power Design Course” this fall. For information visit the Web at www.apressman.com.

For more information on this article, CIRCLE 339 on Reader Service Card

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!