op-amp short circuit protection

[chancs]

In a design that I am working on, I sometimes want to connect the output of an op-amp (ADA4522-1) to the ground via an NMOS (2N7000). The duration of this event can be as short as a few ms to even hours. In the datasheet, it listed Short-Circuit Current Source of about 22mA and in the absolute maximum rating section, Output Short-Circuit Duration to Ground is rated for Indefinite. My question is is it safe to short the output to the ground for a long time?


Link to ADA4522 datasheet: https://www.analog.com/media/en/technical-documentation/data-sheets/ADA4522-1_4522-2_4522-4.pdf

[Manul]

Based on datasheet you can do it. But it just sounds non-elegant. If we set aside estetical aspects of such design, one thing to note will be increased power dissipation and temperature. In the design where some degree of precision is required it is good to keep things at a stable temperature. If you short it randomly, you may have noticeable temperature fluctuations which may affect your op amp parameters and your circuit performance. I would do this only if there is a guarantee of very little output voltage at the time of being shorted. This is only suggestion.

[Siwastaja]

General-purpose advice: don't rely on the short-circuit protection features of ICs in normal, intended operation, unless the parts are especially specified for indefinite short-circuit operation (which they usually are not).

For example, the thermal protections implemented tend to trig so close to the temperature of destruction that accelerated semiconductor degradation may happen.

The ICs that exhibit short circuit protection also tend not be specified as functional. Of course, during the protection, you wouldn't expect that, but if you expect the part to be functional right after the short is removed, you may be wrong. It may take cooldown time for the device to resume in-spec operation.

You could use an analog SPDT switch IC if you need to set the output to ground: selecting between the opamp output, or ground.

[Zero999]

Yes, it's not ideal. If it's driving a high impedance load, you might be able to get away with putting a resistor in series with the output, so it's not shorted hard.

[T3sl4co1l]

Can't pull down on +in, or up on -in, for the same result?  It's rail-to-rail output, forcing the output to a rail does the same thing.

If the inputs need to be essentially isolated (no injected current allowed), you might add another stage as a buffer, then a series resistor and a pull-up/down there.  Instrumentation amplifier configuration for example.

Can also look into gated type amps, though I don't know offhand if anyone makes one in zero-drift, or with nearly as high voltage (how much voltage do you really need?..).

Shorting should be okay as it's limited and thermally protected, but again, it's not a good idea to rely on.  They don't give min/max current limits, perhaps because they're all over the place.  So you might not be able to design in adequate heatsinking, to avoid the thermal limit, for example.

LDO regulators do this sort of thing with their current limits; it's a ballpark range, a safety feature, not an operating feature.  It just means it won't explode instantly and you have some time to deal with the fault, but that time might be a hundred microseconds, it might be seconds.  Obviously, this amp is delivering a lot less power than an LDO, and as long as the thermal protection does its job, it shouldn't release the magic smoke, but thermal cycling isn't great for recovery time or longevity.

Can also consider a series analog switch, instead of brutishly shunting it.

Tim