I have designed this circuit that is designed to accept a 0-5V analog signal from a typical automotive pressure sensor, I also want to use a thermistor (a automotive coolant temp sensor) on the same input pin with no external components (voltage source and resistor) so I have added the left hand side of the circuit that can switch a 5V reference voltage through either a 4k or 560R resistor, the reason for the two resistors is because I want to use different sensors of which the specs are vague or no existent and use a wide temp range -10 - 100c.

So when the fets are turned off I want to accept a 0-5v analog signal and when at least one is turned on I want to accept a thermistor that is connected between AIN1 and AGND.

I am using a STM32F which is obviously 3.3v. When I first designed this and asked for help I was told to use a 2.5v reference and supply the VREF+ pin with the 2.5v, so that is why my resistors are set to 10k and 12k. But I am not sure why I can't just connect VREF+ to +3v3 and use a 17k and 33k resistors to get a full 3.3v from 5v?

The input is protected by a SMBJ5.0A as I want to protect it from short circuits to battery positive.

I'm also not sure whether the left hand side of the circuit is going to interfere with my 0-5v signal when not in use?

Ignore the lack of ground symbol, its because I have multiple analog grounds and I kept forgetting what symbol I used for each.
AINB is the output.

 

I think your thermister readings are going to scale in a very strange way with the circuit as drawn. You've got one voltage divider (R74 or R82 in conjunction with thermistor) feeding into another one (R122 and R125) which complicates calculations quite a bit. Maybe workable, but certainly not straightforward. A voltage follower op amp buffer been AIN1 and R122 would help with that.

I wouldn't expect the left part of your circuit to have much effect on the rest when the MOSFETs are off. In theory there are leakage currents from the 5V reference through the MOSFETs even when off, but I doubt they're enough to cause major problems.

These are just my personal first impressions. I'm no expert by any stretch of the imagination. I'll be interested to see what others think.

 

You mean like this?

Am I best to use a 2.5v voltage reference and keep the resistor values the same or am I best to supply Vref+ with 3.3v and use 17k & 33k resistors? What is the benefit of both options?

 

I think that op amp setup should work. The op amps will need to be powered from the 5V side in order to safely handle direct 5V input in case a MOSFET is activated with no thermistor connected... or clamping diodes and series resistors could be used, or both. Not sure. I'd have to think about it more, but short on time right now. Sorry.

I'm not sure how much difference 2.5 vs 3.3 reference will make. Hopefully someone more experienced can give a definitive answer.

My gut feeling is that whenever you have a choice, it's better to keep voltages higher (within reason) as long as possible to stay father away from the noise floor. So that would lean me towards 3.3V.

When I think about analog reference sources, I'm usually focused on accuracy and stability. So, using a 2.5V reference from a really stable source might be better than using a 3.3V reference that's shared with lots of IC power connections and could get noisy. Then again, when you really want accuracy in delicate situations, the reference voltage is also used as the starting point for generating your sensor signals (like in wheatstone bridges, etc.) and that's not going to work here regardless, so not sure it matters.

As you can tell, I have a lot of loosely connected facts floating around my head, but I'm not 100% sure what they mean for your situation. My guess is that, if the rest of the design and setup is staying the same, 2.5 vs 3.3 won't matter much.

 

Will that mean powering each op amp from different supplies, i.e. the first IC11B with 5V and the IC11A with 3.3V?

 

Will that mean powering each op amp from different supplies, i.e. the first IC11B with 5V and the IC11A with 3.3V?

I don't think so. Seems like both powered with 5V should be fine. If IC11B sees 5V max in/out, then R122/R125 should guarantee that IC11A in/out stays within acceptable limits. In that case powering it with 5V does no harm, and its output is limited based on its input signal.

Please double check my thought process on all this, and don't just take my word for anything. I think it all makes sense though...