Cheap extra protection for output stages?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
A thought I had about failure modes of output stages. When heavy current is drawn from the output (for instance during a short circuit or if one output device fails), the voltage across the emitter resistor can be momentarily quite large (until the filter cap has discharged at least). Hopefully a fuse fails before the devices carrying the current suffer secondary breakdown.

This situation with the high voltage across Re can directly reverse-bias the base-emitter junctions of the opposite devices leading possibly to their failure as well. Thus damage can cascade to more transistors than necessary. Adding two signal diodes should prevent this and give a very cheap way to gain some extra robustness:

D1 and D2 provide this protection by allowing C2 to charge up. There need to be resistors between the Vbe multiplier and the drivers for this to work, as the multiplier is a shunt regulator. Its worth simulating the whole VAS/bias/driver setup into a short circuit load anyway to see if other devices are at risk in this situation - cascade failure back to the small signal part of the amp is also worth protecting against.

If the emitter resistors have upto about 2V across them at normal full-swing output then ordinary silicon diodes should be OK for D1/D2, if more than 2V happens in normal operation then you could use a cascade of signal diode and LED to increase the forward voltage. Bare LEDs have a low reverse breakdown, note.

CFP output stage has somewhat different voltage constraints of course.

Another approach at greater cost is large diodes in parallel to emitter resistors to limit the voltage (as well as improving output swing into heavy loads).
Last edited:

Your idea does a bit more than you have bargained for, I think...
It also current limits.

Assume the bias+ and bias- nodes are connected by a Vbe multiplier that has a quiescent value of Von.

If we drive these nodes high to make "output" high, then bias- gets nailed at output+Vd2(on), and bias+ limits to output+Vd2(on)+Von. So, the maximum positive output current is limited to about Vd2(on)/R1.

IIRC, I think Roberge shows this in his opamp book, or maybe it was Williams or Pease...
Great idea Mark, but you are late to the party. That circuit has been around for ages. The first time I saw it was on the ADI module op-amps that Ray Stata designed a long time ago. May even pre-date those though.
But hey, rediscovering neat circuits by yourself is incredibly satisfying though.
Good job (I am not being sarcastic, I mean it).
- Sandro
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.