Actions
Forum Thread Details
  • Replies 63 replies
  • Subscribers 48 subscribers
  • Views 4658 views
  • Users 0 members are here
Related

Engine Management

jack.chaney56

Hi Ben,

I am a programmer with a very small amount of skill with circuits, and am looking to create a platform for an engine management system, using an Arduino Mega 2560. I had done a bit of the coding, when I ran into some timing issues with the built in Arduino manager, so I switched over and started using AVR Studio and a programmer to go directly to the chip itself.  The code looks like it should work ok, but now I need some additional circuits to handle the energy levels of coils and injectors (Something like IGBTs). Sensors are being run through simple dividers (no protection yet), and cam and crank inputs are through a simple comparitor

 

Let me know what you think,

Jack

  • Something that I wanted to point out, because there have been several suggestions for using a variety of other processors. If you look at the code provided, the great majority of the operational components are not dependent on any particular processor. As long as there are A/Ds available, and some mechanism for timer control, all the rest is fully portable.  In fact with the exception of four files, (with minor differences) the same code runs on ATMega328 and ATMega2560.  The code for the Cam and Crank operation only requires an interrupt vector point for a trigger line for the signals.

     

    Next part will start to cover exceptions, changes, and asynchronous operation.  Also, how to manage calibration and monitoring.

     

    Jack

  • in reply to jack.chaney56

    This is (all of it) quite a brain dump, thanks.  When this is all done, I'll start designing engines (no, not really, is joke . . .)

     

    I understand and agree about the integer/floating point discussion.  When optimizing program execution time, integers are soooo much faster.   And multiplying y=x*x*x is way faster than using y=x^3.  [I learned this on a motorola 68k processor (PLC, not macintosh)]  Do the integer math for all the intermediate calculation.   When (if) you need precision, yes, do floating point at the end.

  • in reply to mp2100

    Yes, thank you Allen.

     

    I have a general 16 bit rule book that I use.

    The 16 bit solution
    All data elements have like conversions, and a single set of units.

    All conversions are done externally to the device, or at the UI level.

    Values are all maintained as 16 bit integers. types are:
      - Temperature -> absolute degrees K * 100 (0 degrees C stored as 27315)
      - Angle in binary radians 360 converts to 32768
      - Ratios are all fixed point * 1024 or 100% == 1024
      - Pressure is stored in kPa * 100 absolute (1 atm stored as 10129)

     

    Limits of accuracy for equations are imposed based on 16 bit values
      - angles (error +/- 0.0109 degrees)
      - ratios < 6399.90% for unsigned or 3199.90% for signed (error +/- 0.049%)
      - temperature < 655.35 degrees K or 382.20 degrees C (error +/- 0.005 degree C)
      - pressure < 655.35 kPa or 6.47 atm (error +/- 50Pa)

     

    This manages a significant amount of the measurement. Any cases where the values or accuracy are exceeded becomes a special case and happens very rarely.

     

    Jack