Correct Way to Limit Op Amp Output Current

Question

I'm planning to use an AD817 op amp in the output stage of my function generator (datasheet). I'm worried about how the part would stand up to a short, so I ran some simulations. The simulation below sources 70mA through R3 when the output (probe tip) is shorted. This makes sense and tallies somewhat with the datasheet which gives a max current of 100mA @ 25degC, but less at higher temperatures. Since the part (according to spice) is dissipating 800mW it's likely to be hot. Although 800mW is within the max spec for this device (around 1.4W @ 25degC), I'm not comfortable with 800mW of dissipation. Which brings me to the question.

Is there a best practice way to limit the output current of an op amp under short circuit conditions?

Basic Amplifier:

Here are some options I've tried, with comments:

1) Push-pull Amplifier to Relieve Op Amp Output Stage

Works well to limit op amp output current but could deliver 10V/47 = 200mA through R3 @ 1W. The transistors would need to dissipate 500mW each.

2) BJT Based Current Limiter

BJT based current limiter based on some sources I came across (see here and here). This is more like what I'm looking for, but I couldn't make it work!. Am I correct in thinking that the current diverted away by Q3/Q4 ends up at the base of R3?

Other options:

3) Reduce the gain. Reduced signal amplitude isn't desirable.

4) Limit the power supply current. Would (needlessly?) effect other parts of the circuit.

Answer 1

EDIT - IGNORE C1 - unfortunately this is the only circuit that I could find with the Rout component shown within the feedback loop so, please ignore C1 - it is not meant to be there.

This technique is used (with care) a fair amount: -

Ignore C1 and just analyse how much current the op-amp can supply to a load in the place of C1. If the maximum output of the op-amp is (say) 4 volts (+/-5 volts supply), there can be no more than 53mA delivered into a short circuit.

On normal loads, the op-amp has to work a bit harder because the 75 ohm forms a potential divider BUT, the good side is the feedback loop compensates for this.

Answer 2

You can use your method, but just add some more op-amps in parallel and increase the resistors. For example, use 100R resistor to get an output Z of 50 ohms.

I believe this method is used in Keysight (née Agilent, née HP) arb function generator output stages.

^{simulate this circuit – Schematic created using CircuitLab}

Answer 3

The 50 ohms output impedance helps as it will limit the output current. There are opamps that can drive 50 ohms (to ground) BUT the output voltage will be limited to a couple of volts. Usually too little for a function generator especially if you would also like to be able to shift the DC level at the output up or down (this adds to the voltage).

The easiest and most common choice is to implement and amplifier like in option 1), the amplifier should be designed to drive any voltage into 50 ohms. Current limiting is not needed as the 50 ohms resistor will limit the current.

Ad option 2) I think the output current never reaches its limiting value because you have the 50 ohms at the output. Anyway, if you insist on current limiting I would suggest to use a more traditional solution with current sensing resistors in series with the emitters like your first link. Many audio amplifiers use this type of protection.

Another observation: your amplifiers appear to be class B amplifiers read this if that does not ring a bell. For a function generator you want low crossover distortion so I would suggest at least class AB or even better: class A. But it all depends on your needs and how efficient you want your design to be. Oldfashioned analog function generators get warm, they use a significant amount of power, that is (partly) because their endstage operates in class A !