This is part of a larger circuit, but problem area is shown below. In simulations (of the larger actual circuit not shown), the base voltage of Q1 drops to about -6.35V. Can this damage Q1 (2N3904)? What if the base voltage drops to -15V? Current flowing through emitter base with base at -6.35V is in pico amperes (~25pA).

If this is a problem, how can I limit base voltage between 0 and -5V, without limiting op amp supply voltages?

PS: This is for a bench power supply. I need a +15V since there is a differential amplifier for voltage sensing and I lose accuracy at low output voltages if op amp +V supply is low. The -15V tracks the +15V, otherwise I lose accuracy at low voltages and current sensing (due to greater offset).

enter image description here

  • \$\begingroup\$ Reverse biased diode across B-E. Normally off, but will cut Vbe off at -0.6V, max current determined by R1. \$\endgroup\$ – Brian Drummond Mar 11 '18 at 14:01

The absolute maximum Vbe voltage of Q1 is -6V , as you can see from a cursory look at the datasheet. So, -6.35 is (slightly) outside the allowable range.

Typically they won't break down until 7-9V so you're probably okay, but if you want to be sure just slap a diode (eg. 1N4148) from base to emitter, or change the divider values to limit the base voltage to -5V.

It's bad juju to cause reverse breakdown in the B-E junction, and a long term reduction in beta has been observed as a result of higher currents being forced through junction.

  • \$\begingroup\$ Thanks. I'll see what spice says with a diode, or a zener. Never thought of that. I'm not sure I can limit voltage with resistors. What is the minimum reverse current that can damage the BE junction? Datasheet says 10uA @ 6V, maybe 25pA is OK? Or maybe simulation limits itself, and is not to be trusted? \$\endgroup\$ – Indraneel Mar 11 '18 at 13:49
  • \$\begingroup\$ The data sheet says that the junction will not leak more than 10uA at 6V at 25'C. Since 25pA < 10uA, the transistor meets specifications. You are not supposed to put more than -6V on the base. SPICE may not model reverse breakdown, or if it does, it will be typical and not worst-case (neither the stock LTspice 2N3904 or Circuitlab model reverse Vbe breakdown). \$\endgroup\$ – Spehro Pefhany Mar 11 '18 at 15:46
  • \$\begingroup\$ I added a 1N4148 from emitter to base and all seems to be good. Minimum base voltage is 0 to -1V. Spice is showing some peculiar oscillations at some voltages, but I think that may be a LTspice problem (oscillations when I set both Iref and Vref at 0.9, but no oscillations at 8.99 or 9.01). Thanks again, I accepted your answer. \$\endgroup\$ – Indraneel Mar 11 '18 at 18:15
  • 1
    \$\begingroup\$ Folklore indicated the Noise Figure is degraded, once the emitter-base junction gets into breakdown. I suppose the avalanching of charges into the junctions will inject more charge-traps which will absorb and release charges randomly. So ..... avoid EB breakdown in a low-noise design. \$\endgroup\$ – analogsystemsrf Mar 12 '18 at 5:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.