I have come across a circuit (I am not allowed to share it) with an opamp that uses a special output. The opamp's output is loaded with a resistor and has no other connection. It is the current drawn by the opamp that is used as output instead. This output current is then fed into an optocoupler, which controls other circuitry and feedback is then brought to the opamp's input to close the loop.

Does this solution have any advantages apart from possibly a lower component count? Are there any risks (one could be different values of supply current for different opamp types)?


  • 4
    \$\begingroup\$ This same structure can be used to control complementary output transistor in rail-to-rail output stages, e.g. look here: electronics.stackexchange.com/a/625828/237061 . I think it is not too exotic. Stability might be a bit harder than for more conventional circuits, because the negative feedback loop is probably a bit "longer". \$\endgroup\$
    – tobalt
    Jul 20, 2022 at 9:22
  • \$\begingroup\$ At least tell us which op amp. It may be important. \$\endgroup\$ Jul 20, 2022 at 11:05
  • \$\begingroup\$ It is OPA237NA. \$\endgroup\$
    – Hyp
    Jul 20, 2022 at 11:30
  • \$\begingroup\$ If this question has been answered then you should formally select an answer @Hyp else raise a comment for clarification. \$\endgroup\$
    – Andy aka
    Sep 2, 2022 at 13:54

2 Answers 2


Are there any risks?

Absolutely. Nearly the whole of the data sheet of an op-amp is given over to stating the power supply conditions for a whole bunch of specifications and measurements. Those conditions will imply a stable power supply that is adequately decoupled with 10 nF or 100 nF capacitors on both supplies.

So, if you insert components in the power feeds to the op-amp then, you are likely degrading every test condition in the data sheet and this invalidates any reliance on those specifications.

Does this solution have any advantages apart from possibly a lower component count?

The primary disadvantage stated above trumps any advantage that might be gained.

Of course I've seen this being done in some audio amplifiers and it looks neat because it can self bias the output transistors in a push-pull amplifier but, there are other ways to skin a cat and, those other ways will not degrade the performance expectations specified in the data sheet.


Et cavete a user (let the user beware)


This is a common technique, and there’s nothing too proprietary about it. Plenty of old app notes show some examples of it. Generally, it’s not the best idea, since the modern parts were never tested for such use not specified for it. So you may get something that works for a year and then won’t ever work again with newer production lots. Are any savings worth the trouble? No, they are not. It’s a clever hack for hobby and boutique use (say analog synthesizers), but in a regular product I’d stay away from it.


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