I'm considering re-implementing my thermistor measurement circuit to be isolated, since it's destined for an industrial environment. The thermistor is a simple passive sensor and is not grounded on the far end.
U1 is a 5V/5V regulator rated for 3kV isolation. The line out to the thermistor is connected to protection shottkys, a TVS zener, and caps that set the time constant at (I think) ~1s.
U3 is a "linearized" optoisolator rated to 3.75kV isolation.
Back-of-the-envelope calculations for the following nominal values:
- the input and output of U2 at 2.5V,
- the LED current at 8.7mA,
- Irm1 = Irm2 = 90uA,
- line time constant = 853ms
simulate this circuit – Schematic created using CircuitLab
Questions:
- Have I made any mistakes in how the circuit is connected or the passives are chosen?
- Is it even necessary for this isolation to be done, given that the far end does not have its own ground?
- Are there any rules I should follow with regard to PCB layout (especially shuffling around the diodes and caps, etc.)
Best Answer
Thermistors are relatively high level devices and are not generally grounded so it's not usually necessary to isolate them. You can add a low pass filter and use shielded wire to the thermistor if you expect significant EMI.
You should check the reverse leakage of those Schottky diodes- if they are the type I think I see-- at -4V and 75 degrees C they will typically leak 200uA!
Edit:
simulate this circuit – Schematic created using CircuitLab