I am using the circuit below, which is comprised of 4 Tekscan Flexiforce sensors (https://www.tekscan.com/products-solutions/force-sensors/a301). In the circuit below, the sensors are Rs0, Rs1, Rs2, and Rs3.

The first op-amp is to dynamically control the force range of the sensors and output a signal from Vref (0.5 volts using the voltage divider 9k and 1k) to 5V.

The second op-amp subtracts the offset (Vref) and amplifies to 0-5V.

I also have MOSFET's attached to each sensor to control which one is active. I built this circuit, but noticed that the sensors that are pulled LOW using the MOSFET are still introducing some signal in the output. Basically, when I press the sensors that are supposed to be inactive, they are still outputting a small signal.

I am assuming a MOSFET was not the right choice, and there are better switch devices to pull the sensors to GND harder. What other kind of devices should I include to better isolate each force cell? Any other thoughts would be much appreciated!

enter image description here


1 Answer 1


Those 2N7000 are a reasonable choice, but the specified leakage current (1 uA at room temperature) would cause the effects you observe. Do you have a way to measure if in fact the FETs are leaking that much ?

In any case, you can make the circuit insensitive to FET leakage by adding additional FETs between the drain of each FET and the opamp's non-inverting ('+') input. When you switch off each sensor (e.g. M1), turn on the corresponding new FET also. The new FETs should have drain = '+' input; source = junction of sensor and existing FETs.

This works because the opamp creates a 'virtual ground' at its '-' input. By switching the bottom node of each sensor to the '+' input, the unused sensor will have nearly zero V across it and therefore will not affect the behavior of the circuit. The leakage in the additional FETs will have a negligible effect on anything else.

  • 1
    \$\begingroup\$ Also, keep in mind that your reference voltage at the + input of the op amp needs to be extremely stable, and generating it as you do is asking for trouble. Power supply noise and variation need to be accounted for. \$\endgroup\$ Dec 18, 2015 at 4:59
  • \$\begingroup\$ @WhatRoughBeast Thanks for the info. Would you suggest using a reference diode? I'm using one currently for an my Arduino ADC. \$\endgroup\$
    – tabchas
    Dec 18, 2015 at 5:00
  • 1
    \$\begingroup\$ @tabchas - It all depends on your power supply. How quiet is it, and how stable? It's your call. \$\endgroup\$ Dec 18, 2015 at 5:01
  • 1
    \$\begingroup\$ Actually your whole opamp circuit is a little strange. How 'perfect' you need to be depends on the accuracy you are trying to achieve. Normally R10 would be 100k also to make OP3 a proper difference amplifier; then VOUT2 would be VREF*rf/rs. Also best to add 1k in series with R9 (make it 101k) to match impedance from R1//R2. \$\endgroup\$
    – jp314
    Dec 18, 2015 at 5:07
  • 1
    \$\begingroup\$ @tabchas - The obvious place to start is a cap across R2. However, you may find the nonlinearities in the sensor are more important than power supply issues. You'll just have to pay attention to what you're doing. \$\endgroup\$ Dec 18, 2015 at 5:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.