Background
I'm designing a signal acquisition board and one of its parts has several differential op-amps which are not very good from PSRR standpoint. So they need a very clean power supply, and since my main power rails (5 and 12V) come from a SMPS, I slapped a LM2931 LDO to get a clean 3.6V rail for the op amps.
Now I'm tempted to use this same rail for the supply of some 4 temperature probes. They are wired like this:
simulate this circuit – Schematic created using CircuitLab
The problem
The thermistors are changeable/pluggable and are moved around often; therefore it's not impossible that external wiring fault gets me +5V or +12V in the "To ADC" node.
The ADC itself is protected, so it won't be toast. I've added a TVS parallel to the thermistor to guard against short transients (ESD as well).
However in a sustained +12V scenario, assuming the TVS fails open, current will overflow through R1 and potentially raise the +3.6V rail, since the LDO won't be able to sink that ~6mA of incoming current, ruining the other 3 temp sensors, and potentially damaging the LDO.
Desired effect
I want the acquisition board to be rugged against such faults. The wiring would get corrected eventually, but I don't want this to be an immediate show-stopper for the other measurements.
Proposed solution
One way would be to add a low-value resistor on the +3.6V rail to provide ample sinking capability, should it be needed.
However, I'm also thinking about adding an op-amp buffered second +3.6V rail, which would be capable of sinking current:
The idea being that
- The LM358 serves as a kind of isolation between the important +3.6V rail (that I really care about), and the thermistors power supply (not that high-priority).
- The buffered 3.6V can sink current, if external faults arise.
Questions
Is the op-amp buffered idea worthy (would the second +3.6V line still be clean from noise), any caveats I might be missing? Some other approach entirely?