0
\$\begingroup\$

I was doing some research into RPM governors for diesel engines and could essentially find two different approaches. One was based on some NE555 based PWM circuit that based on the input RPM from a hall sensor creates more or less current in a solenoid that will actuate the throttle of the engine. It was usually rather complicated circuits with a lot of elements.

On the other hand I also found quite a few approaches where an arduino was used that took the signal from the hall sensor and essentially controls a servo that increases or decreases the throttle. It ended up being pretty much "Take arduino X, connect hall sensor to pin Y, servo to pins XYZ and run this code".

Objectively this is much simpler than a circuit with roughly 30 elements. So I was wondering, are there any downsides of using a "digital" approach for an RPM governance device compared to the purely "analog" version with just ICs and some RC components? Or is just about costs?

\$\endgroup\$
2
  • \$\begingroup\$ Most things automotive are about costs. That, and ruggedness. \$\endgroup\$ Aug 27, 2020 at 20:00
  • \$\begingroup\$ Funnily, as an end customer usually the easier solutions are also the cheaper ones, an arduino nano clone for 1-2$ is cheaper than ordering all the parts for the analog circuit with 100pcs for each part :) \$\endgroup\$ Aug 27, 2020 at 20:21

2 Answers 2

1
\$\begingroup\$

The analog version has the advantage of no software, relatively easy-to-understand circuit and instant boot-up.

The micro-controller version has the advantage of being able to do some fancy stuff such as PID control to prevent instability, overshoot or to adapt better to varying load conditions.

In the harsh environment the microcontroller version may offer long term stability as it is less likely to suffer from component degradation.

\$\endgroup\$
2
  • \$\begingroup\$ As a full time programmer the "no software" at least for me is not really a point ;) \$\endgroup\$ Aug 27, 2020 at 20:25
  • \$\begingroup\$ Thanks for accepting my answer but I recommend that you un-accept to while the Earth rotates once or twice to allow the whole of humanity a chance to answer. You'll then encourage some different viewpoints which may help to build a better picture. You might get a much better answer. \$\endgroup\$
    – Transistor
    Aug 27, 2020 at 20:59
1
\$\begingroup\$

If you are just controller RPM, slowly, you only need one Hall feedback sensor. with more you can respond faster 3x120 deg. the control is about anticipating load changes to torque and speed by estimating load regulation error and rate of change with power added.

  • by anticipating changes (computing derivative or changes of time interval for frequency , you can assign a gain to D, by simply comparing desired f to measured time interval averaged,the error gain is assigned to P and to get exactly the right frequency you can remove the P gain which creates some residual error,myou can integrate the time interval errors with some gain feedback called I.

  • Now the PID range of values for gain correction are all critical to startup and load step responses and with the wrong values, it can be unstable, ring or oscillate.mSo you start with P,and small, and medium D gains then learn more about PID autocalibration and fault detection ,should feedback have noise or get open circuit or shorted then correct itself with alerts.

The code may do some of this ,but without specs!! i can only guess. The proper way is to start with IO specs and functional error tolerances. Then create a block diagram with gain values for each. Then add details for sensors, indicators and fault management or operator attention.

\$\endgroup\$

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.