here's the scenario - need to test a temperature sensor (a solid state device on a PCB in this case) for short to 5V and short Gnd faults. What would be feasible methods to creat a fault? For instance if I connect one side of the sensor to an external 5V, that should be able to give me a short to 5V fault ... is that right assumption?
Also, what are the implications that I should be aware of, e.g any EMI effects, circuit imedance etc while trying to check for short to 5V and short to Gnd faults.
Thanks

 

That depends on the nature of the device. If it is powered, it is already attached to the 5 volt supply. A fault would be more like the device opening up, so it had no output. That would be true of a purely passive device like a thermistor as well as something more sophisticated like a three terminal sensor with a scaled output.

Shorting to ground is not something you want to do just to see what it looks like, unless you can be assured that the device won't cause some other component failures due to excessive current.

Doubt if you will see anything radiating from an open or a short. Opens are characterized by no current in that branch of the circuit, so full supply potential exists across the failed device. A short may cause overheating due to excessive current.

 

actually the requirement is to test the micro controller for short to power & short to gnd. The sensor is powered to 4.96V .... and if it sees a 5V or higher, the micro controller would set the fault for short to voltage ... yep thats rite!

 

That sounds like the ADC is simply going to the maximum value. Why not change to programming to reflect a conversion out of range?

 

can you explain a bit when you say "Why not change to programming to reflect a conversion out of range"

 

I obviously have no knowledge of the device or the significance of the sensor in the overall system. If this is just for testing purposes, why not just have the uC indicate the voltage is out of range? Is the sensor sourced by a voltage greater than 5? Is thre some potential for harm? If it's a test setup, surely someone can change some code.

 

I see what you saying. Yes, there's a runtime param to read the current value of the sensor voltage. That would be an easy & non-intrusive way to test this boundary case - thanks for the tip!