Short circuit detection algorithm that doesn't false trigger?

[e100]

I'm trying to make a power supply that has short circuit detection, but ignores the brief current transient you typically get when charging up the decoupling caps on the device that is being powered.  For inspiration I looked at the data sheets of various voltage regulators but these give little or no information on how this is implemented within the device.

Currently to avoid false triggering I continuously calculate a rolling average of the current over a half second window. This seems to work ok on the bench. Is this the sort of algorithm used in commercial devices?

[Phoenix]

What sort of power supply is this? Most DC bench supplies have a constant current mode that effectively protects against short circuit.

[e100]

What sort of power supply is this? Most DC bench supplies have a constant current mode that effectively protects against short circuit.

It's essentially a car battery with a solid state switch that powers a bunch of sensors at the end of 50 meters of cat5. There's a 0.2 ohm resistor in the low side that I use to measure the current with a microcontroller.

[SiliconWizard]

Well, hoping that your PS has a current limiter for safety reasons, detecting a short-circuit condition may be as simple as monitoring the output voltage of your PS instead of monitoring the output current. If it's a 12V supply, you could set a threshold of 1V or 500 mV. Looks like a sure way of detecting a short.

[donpetru]

In this case it would be useful to attach schematic because I think you have wrongly placed that 0.2 ohm resistor as a current sensor !!!

[e100]

In this case it would be useful to attach schematic because I think you have wrongly placed that 0.2 ohm resistor as a current sensor !!!

The car battery is in my garden, not a car so there's no grounded metalwork to warrant a high side shunt resistor.