design review request

A circuit is using a N-channel MOSFET Q3 to switch off supply to part of the circuit using a microcontroller output: the source is connected to the actual ground, drain to the switched ground and the gate is driven by the microcontroller which is connected to the actual ground.
The reason for that has to do with the purpose of the circuit, for power saving and for "historical reasons".

This circuit has worked well in actually fairly harsh conditions (hight temperatures) for most of the year. The last few days however (cold and no more sunshine powering the batteries through its solar panel) some erratic behavior is seen (data not properly transmitted). When recovering the unit from the field I notice that LED3 (battery) keeps lighted, even with Q3 forced closed (gate to 0V). When measuring conductivity between switched ground (GND) and actual ground (AGND) there is no conductivity.
No shorts, or bad soldering, or bad connections are seen. Q3 can be switched properly by connecting the gate to either AGND or to Vcc.

Do you have any suggestion or explanation for the bizarre behaviour of LED3?

Has some part of the external wiring developed an earth fault?

Your sensor terminals are directly connected to inputs on the HC132; an external ground plus the input diodes on the IC could cause some voltage on the LED.

For AC resistance sensing you should be able to use capacitors in series with the probes to avoid DC problems.
By preferred method is a grounded probe (or metal casing in a fluid system) plus a signal probe fed from an AC source or oscillator via a resistor and capacitor, then another capacitor from the signal probe to a load resistor and precision rectifier.

Right off the top of my head...

JP4 shorted??

What's the output status of pin D9 of the Arduino?

What is the on resistance of a 1N60 with a gate voltage of 5V? All the figures in the datasheet are with Vgs at 10V.

Mike.

Solved! Last sleepless night I realised that LED3 is connected to a higher potential than the switched load. Q3 is perfectly fine, no current through source.
But since LED3 is connected to a higher potential there will be a stray current reversing through the load to Vcc.

I have currently no solution to light up LED3 only when Q3 is conducting.

EDIT: well, it just occured to me that the MCU can drive another N-mosfet which switches LED3 to AGND. Extra component extra cost.

EDIT2: attached drawing shows Vcc = 3.3V, should be 5V. Since the battery supplies 6V only currently this may explain somehow why only now I start to see the erratic behaviour. And since my initial testing did occur at 6V input only that explains why I never before did see LED3 actually light up, the light from it was too faint.
Conclusion: prototype testing can never be too thorough.

cowboybob said:

Right off the top of my head...

JP4 shorted??

What's the output status of pin D9 of the Arduino?

Click to expand...

JP4 is not shorted; Arduino output switches nicely between 5V and 0V.

Pommie said:

What is the on resistance of a 1N60 with a gate voltage of 5V? All the figures in the datasheet are with Vgs at 10V.

Mike.

Click to expand...

When Q3 is on there is no voltage drop across D to S.

earckens said:

When Q3 is on there is no voltage drop across D to S.

Click to expand...

By "on" do you mean has 3.3V (on your updated schematic) on the gate?

Edit, missed your post higher up.

Mike.

Pommie said:

By "on" do you mean has 3.3V (on your updated schematic) on the gate?

Edit, missed your post higher up.

Mike.

Click to expand...

"on" = conducting