I want to thank everyone for their input and the alternate method of using MOSFET packages. I really appreciate the help! Plan on using the BJT version with crutschow's modifications. But the MOSFET choice is also probably just as good. It is always a pleasure getting input from you all on this forum.

 

If you could manage with only half the A/Ds range - after all it is just the battery voltage you are measuring - 9 bits instead of 10, then this might work.

It needs a MOSFET with a guaranteed Vgs(on) of <1.5V - such as RTR020N05TL.

 

If you could manage with only half the A/Ds range - after all it is just the battery voltage you are measuring - 9 bits instead of 10, then this might work.
View attachment 245343
It needs a MOSFET with a guaranteed Vgs(on) of <1.5V - such as RTR020N05TL.

Thanks. I have noticed that specific part number has built in ESD protection that appears per the datasheet to increase the gate to source leakage current up to 10 microamps.
Also has a Drain to Source 'off' leakage current max of 1 microamp. The one major difference I have found between FETS and BJTs. All the FETS I have looked at have leakage currents up to 1uA. With BJT's the 'off' leakage current can be in the picoamps or nanoamps depending which part is selected. For BJT's is goes up considerably for high temperature applications. I have seen some at 15uA when at 150C. But for ambient temperatures below say 80C leakage current for BJT appears better than FET's. Though I prefer to use FETs when possible.

 

The datasheet for the chosen BJT's have an leakage current specification of 15nA at 25C.

But that is with the Emitter open. The emitter is grounded (NPN) so 15nA * hfe.
The mosfet will be much lower than 1uA operating at 9v.
But either will work..its up to you.

 

Thanks. I have noticed that specific part number has built in ESD protection that appears per the datasheet to increase the gate to source leakage current up to 10 microamps.
Also has a Drain to Source 'off' leakage current max of 1 microamp.

It was the first one I found - I bet there are others without zeners.
The gate-source leakage current was specified at 12V, not 3.3V. When the gate is at 0V, then there source will also be at 0V and there will be no leakage.

 

But that is with the Emitter open. The emitter is grounded (NPN) so 15nA * hfe.
The mosfet will be much lower than 1uA operating at 9v.
But either will work..its up to you.

What you are saying seems to make sense. In LTSpice simulation here I used 2N3906 and 2N3904. The OFF current through R2 is 10nA. The OFF current through R6 appears to be practically zero. Is there an issue with the Sim?

 

The FDG6332 spec states Max VGS(th) of 1.5v. IDSS is 1uA Max but that is at 16V and will be much less at 9v or 3.3v
I don't think you can get a "guaranteed" <1.5v on any mosfet unless you pay for QA inspection.

 

What you are saying seems to make sense. In LTSpice simulation here I used 2N3906 and 2N3904. The OFF current through R2 is 10nA. The OFF current through R6 appears to be practically zero. Is there an issue with the Sim?

I wouldn't use a sim for that. I'd test using a HW prototype if it's that important.

 

I wouldn't use a sim for that. I'd test using a HW prototype if it's that important.

You have a point. Sim's are not the real world and have limitations. I am using SMD to DIP converters to build a prototype on my solderless breadboard before I make the actual PCB.

 

I wouldn't use a sim for that. I'd test using a HW prototype if it's that important.

True, LTSpice has limitations. I wired up your circuit with the part you recommended. It had only 20pA of leakage current. That seems highly unlikely. Also noticed even with 3.3V on gate it was not turning the circuit entirely off. Attached is the sim:

 

True, LTSpice has limitations. I wired up your circuit with the part you recommended. It had only 20pA of leakage current. That seems highly unlikely. Also noticed even with 3.3V on gate it was not turning the circuit entirely off. Attached is the sim:

That's not my circuit.

 

Why not using a voltage divider with high value resistors (in the 10M range) buffered by an opamp with fet input and shutdown?
A quick simulation with ltspice (LTC2063, 100M and 10M resistors) gives a consumption lower than 1uA (0.5 measurement, 0.2 idle).

 

Here is the circuit from post #11 with some modifications to reduce current consumption.

 

Here is the circuit from post #11 with some modifications to reduce current consumption.

View attachment 245348

Appreciate you taking the time to wire up and sim this. I am considering using it.

 

Note that the bipolar transistor circuit uses only leakage current when not measuring the voltage.
Is the current used during measurement, critical?

 

Note that the bipolar transistor circuit uses only leakage current when not measuring the voltage.
Is the current used during measurement, critical?

Not really, the microcontroller is in sleep mode and wakes up every minute to take a temperature reading over I2C device. It then goes back to sleep. Everyday or so, the data is downloaded via wireless control to a PC for data analysis. During this process the PC request the battery status. So the circuit turns on long enough to allow the voltage to stabilize and then takes a measurement and turns that circuit back off.

 

Perhaps you can use a system similar to mine. The PIC operates on 6V (actually 5.5) and the ADC connects to a constant 3.3V even as the battery voltage begins to drop. As it drops, the ADC reading will actually start rising. Experiment with the ADC numbers to flag your low battery condition. Sorry about picture size.