Voltage divider with MCU - On/off circuit

In order to save battery powered device as measuring voltage is every minute or so, my idea is to use MOSFET or transistor as a switch. However, I'm not certain this is possible.

In this case MOSFET. MCU is 3.3V powered. Divider is made of 10K and 820 ohm, then impedance should be suitable for ADC and measured voltage range from 0-40V. At max measured voltage of 40V, voltage to ADC should be 3.03V and current 3.7mA, which should be safe for MCU...

Now, what happens if involve N-ch MOSFET in this example? When MOSFET is off, the 40V (for instance) with drop of 10K will go to ADC and destroy it...

Then, P-ch MOSFET variation circuit is much more logical, however is this approach is suitable?

Thanks.

The problem with the P-channel approach is that for very low input voltages there will not be enough Vgs to turn on the FET. If the goal is to reduce the steady-state current coming from the 40 V source, two options:

An N-channel gate driver IC and an N-channel MOSFET in series with the 10K (as in your P-channel schematic). This has a charge pump so that Vgs is not dependent on the input signal.

A optical solid state relay with the switch in series with the 10K. This has an internal voltage generator and can switch millivolt sources. It's like a pre-packaged low power version of the first option. Digi-Key, 4-pin DIP package, 60 V or more:
**broken link removed**

ak

I prefer to buffer an R divider with a Rail-Rail CMOS OA.


Or if you know the input impedance of the ADC

simply passive

Thanks to all, helpful info indeed...

The MOSFET solution then seems to not be suitable at all, involving more complexity and inaccuracy. Thanks.

Solution with op amp also solves the problem regarding over voltage protection of MCU's ADC.
Passive solution is questionable, as input impedance into ADC pin should be less than 10K.

Very helpful info for me, indeed.
Thanks.

One of your issues may be because your using a 2N7000. There are much better FETs out there with much lower RDS(on) characteristics.

Yes, however this is actually more generic idea than real usage and 2n7000 is just generic representation of the N-ch MOSFET.

LTspice IV seems to be free software and much easier way to test suitable circuit before actually make real prototype...

Thanks.