At work we have a machine that is used for testing and calibrating force sensors. It basically consists of a huge lever that is actuated by an electric motor with a gear reduction and a calibrated reference pressure load cell. Right now, the machine is driven by hand by turning the voltage knob of a benchtop power supply that is directly connected to the motor, while reading the mV output of the pressure load cell on a benchtop voltmeter. For each calibration, the required forces are converted to the mV reading of the reference load cell by an Excel spreadsheet. The sensor under test is hooked up to a power supply and it's output is measured with another benchtop voltmeter. The machine is then driven to each point and the result is noted by hand.

We want to automate this machine. The measurement part is pretty straightforward. Our Fluke benchtop multimeter has an RS-232 output which I can use to capture the results.

The part I'm not too sure about is the position control loop for the machine. I imagine some kind of PID loop controlling the machine. I've done some PID control before, but mainly for heating systems. I also know just enough about control theory to be dangerous ;)

I imagine a µC that reads the reference load cell via an ADC and drives the motor via a current DAC. Why a current DAC? While driving the machine by hand, I quickly learned that one needs to give regular "pulses" to the motor to keep the machine at a specific force value to ensure a stable measurement. This is especially important when calibrating sensors with integrated amplifiers, because those amplifiers need to be held at fullscale load for quite some time (15..30s) while they do their thing internally. So I think a constant current needs to be applied to the motor to ensure it holds the required torque.

Does this sound plausible to drive this kind of machine? Is this kind of system straightforward to control by PID? Any suggestions on what properties I need to check to ensure a smoothly running control system? Is a current DAC the right choice for this application?

  • \$\begingroup\$ A PLC sounds like it would be a better fit to what you are after, no? \$\endgroup\$
    – Tyler
    Commented Aug 19, 2019 at 19:02
  • \$\begingroup\$ What's a current dac? You could omit most of your description and ask more simple questions. Do you want to control motor torque with load cell feedback or what? \$\endgroup\$ Commented Aug 19, 2019 at 19:55
  • \$\begingroup\$ Although we're not supposed to recommend products, there are emerging a few HMI's with built-in PLC functionality. Such as these Pro-Face series with digital and analog I/O. \$\endgroup\$
    – rdtsc
    Commented Aug 19, 2019 at 19:57

1 Answer 1


You could use a microcontroller, but as mentioned, a PLC would be a better way to automate this, or perhaps a motion-control gizmo such as are made by Galil.

You don't want a current DAC so much as a motor driver that'll take whatever signal is produced by your PLC (probably a voltage) and command a current to the motor.

Galil motion control or National Instruments* should be able to recommend a path forward, by buying stuff from them at what would be a staggeringly high price if you were developing a product, but which should (if you use the stuff right) save you way more money in engineering time than you spend on them.

* Whom I recommend as I cringe, because I don't like they way they think - but if you ever hire a manufacturing engineer, they'll be very happy with the setup.

  • 1
    \$\begingroup\$ So please educate me on why I would want a PLC, especially considering I'm not a control automation guy and I would need to learn the hardware and software side and also how to choose the right PLC. I have a background in embedded systems engineering and I'm totally fine with designing and programming a custom controller for this. The only thing I wanted to check first was if I'm on the right way regarding PID for this type of machine and how to exactly drive the motor that is right now controlled by the operator with a DC power supply. \$\endgroup\$
    – Crazor
    Commented Aug 20, 2019 at 4:41
  • \$\begingroup\$ The two biggest reasons that I would justify doing it that way (and I'm in the same boat as you) is, first, that if you go the PLC route then there's a whole ecosystem of modules (like amplifiers and data acquisition modules) that are already built and waiting to be roped together (or plugged into a rack), and second, that if you ever want to shake loose of being the guy who maintains that (instead of being a design engineer) it'll help for it to be built using parts that a typical manufacturing engineer will be familiar with. \$\endgroup\$
    – TimWescott
    Commented Aug 20, 2019 at 15:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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