# Isolated power supply for measurement

I have to measure potentiometer sensor value in a noisy environment (car). The sensor is about 3 meters from my main processor board, my idea was to simply use this sensor as voltage divider and read on the Analog input of microcontroler the value.

My main problem (perhaps not !) is to send my clean 3.3 V on the potentiometer, I am afraid to get ESD/EMI or whatever on the 3.3 V lines along the wire and get back to my card where microcontroler and other sensitive component are.

So how could I isolate the 3.3 V? Can any DC/DC converter do the job ?

• Add another DC-DC convertor if you need isolation. Without more info on the sensor and what it does and how quickly the output changes, it's hard to add anything to this answer though. This is a bit of an XY question. – Trevor_G Dec 16 '17 at 13:21
• The sensor is here novotechnik.com/pdfs/TX2_e_2010.pdf sorry for mistake , it has to measure the position of the rear wing. assuming I will put the electronic close to the sensor and send information via CAN Bus. How to read carefully this Pot value with microcontroler (avoiding trivial voltage divider) ? – Delphes Dec 16 '17 at 15:01
• Is it a stable position or does it move around rapidly? – Trevor_G Dec 16 '17 at 16:03
• It is very stable. – Delphes Dec 16 '17 at 16:25
• THen follow @MarcusMüller 's advice and use shielded cables and filter the crap out of the signal. You can probably treat it as "almost" DC. You need to accept the value with a high impedance buffer too... the spec sheet implies you should not take more than 1uA out of the wiper. – Trevor_G Dec 16 '17 at 16:29

As a minimum I'd do something like this.

The first op-amp basically sets up a constant current into the sensor pot. The current will be 3.3V/Sensor Resistance. For a 1K pot that will be 3.3mA. R2 and C2 provide you some isolation from the noisy 12V. Between the two of those it should provide a more stable voltage at the sensor that is less affected by the vehicles DC.

If you have some other regulated supply though, use that for the op-amps instead and adjust the resistor values accordingly. The 10M resistor R4 and capacitor C3 filter the sensor output and should be pretty much impervious to system noise. With a time constant of 1S only slow changes of the sensor will be accepted.

Adding transorbs or zeners to protect against any nasty automotive voltage spikes entering this circuit would also be prudent. Shielded cable is advised.

simulate this circuit – Schematic created using CircuitLab

• MANY thanks , the 12 Volts I made from the Battery is already safe, no transient, no overvoltage, I will try this schematics. – Delphes Dec 16 '17 at 17:24
• @Delphes Q3 may be overkill for this though. – Trevor_G Dec 16 '17 at 17:45
• @Trevor datasheet says "recommended wiper current $\le$ 1 µA"; I'd much rather have a "possible range", but my wild guess is that long term, it's probably smarter to imprint something in the order of 1 µA, rather than three orders of magnitude above? – Marcus Müller Dec 16 '17 at 18:02
• By the way, while we're at this: Maybe having a current output rather than a voltage output might make especially much sense here. – Marcus Müller Dec 16 '17 at 18:03
• @Delphes yet another contradiction, by the way: you care about EMI and EMP, but "the 12V from the battery are clean": that makes no sense, either. Because if the battery voltage was clean, then there would be no supply stability problem in the first place. And, hint: the battery is the same device to which the alternator (a DC machine!) and (if not Diesel) the ignition coil is connected to. It's pretty much the opposite of a clean voltage source. – Marcus Müller Dec 16 '17 at 18:05

ESD doesn't "just happen". Use sufficiently isolated cable, so that a spark wouldn't even get to the cable.

EMI might be interesting; you'd need shielded cable against that (but also, a noise and signal model. Wild guess is that your potentiometer's signal has a few Hz of maximum frequenzy, whereas your EMI would have kilo- to Gigahertzes and can very easily be filtered out – like you'd filter the output of a potentiometer, anyway, since you don't want to deal with the "scratching" of the pot, anyway).

But then again: there's plenty of electronics in your car. So digitize your potentiometer reading as close to the sensor as possible and send the values over a digital link. Hint: CAN bus was invented for car-internal communication, and there's a lot relatively cheap microcontrollers who can do that. I'm pretty sure you were planning to buffer / amplify the potentiometer sensor output right at the sensor, anyway, so why not directly convert it to digital there?

• Thanks for the answer, I can unfortunatly not put any electronic near the sensor, and If do this I will be oblige to put a power line for this electronic so the problem will be the same. I am not concern by noise for pot reading, I am more concern by my 3.3 V clean voltage supply (the one I made on my board) going along wire in the car. FYI this is a brand new car and the pot will read the position of the rear wing. – Delphes Dec 16 '17 at 11:33
• Your statement doesn't make sense. You're worried about the noise from the 3.3V, but you're not worried about the actual measurement noise – which will be worse. And how is "getting 3.3V to the sensor and a measurement voltage back" any less effort then "getting the supply voltage to the digital part and a digital signal back"? That, also, makes no sense. My answer stands. – Marcus Müller Dec 16 '17 at 12:12
• You are right, there are multiple problems that I need to be concern. So let restart at the begining I have to read the value from this sensor : novotechnik.com/pdfs/TX2_e_2010.pdf this will be 3 meters far from my main board, on this board I have very safe 12 volts powersupply (no overvoltage, no transient etc), I also have 5 volts and 3.3 volts, CAN bus etc. What is the best way to read this pot on my microcontroler ? – Delphes Dec 16 '17 at 12:24
• Hey, this would be a great addition to your question itself (all relevant info should be in the question itself, not in some comment to an answer that noone but me will read...). Also, I've given you an answer to exactly the question on the end of your last comment. – Marcus Müller Dec 16 '17 at 12:33
• @Delphes This really depends on the sensor and what it does too. If it is a measuring a static quantity... like say water temperature this is fairly trivial filtering, If it's measuring some thing that has a high frequency, perhaps, injector signal or something, it is trickier. – Trevor_G Dec 16 '17 at 13:19