Subject: New Megasquirt mother board rotary
From: rotaryeng
Date: 7/16/2012, 9:46 PM
To: AAAA Put this in the To box


 I've always been an indifferent programmer, and I have 0
  experience with 68 asm. I   could probably muddle through it and end
  up with something that works though. Are you   looking for someone
  to make modifications to the MS source?

     I've been thinking, and it may make more sense to use something
  like the TPS input.   MS has a neat built in tuning feature if you
  hook up a wideband. It would be a shame   to give up that
  capability. The TPS on the other hand probably isn't necessary for
  AC   use as it is mostly concerned with throttle response.  I am
  assuming that you have no   other inputs that are unused in your setup.

     Josh

     Mostly for some one to collabrate. Two heads are better than one.
  Most of the time   the other guy has a better idea :) I have got to
  come up with a routine that   calculates the coil dwell time
  compensation as a function of RPM. The idea is turn on   the current
  in the coil 2 or 3 ms before you need the spark to fire.

     We can connect the O2 sensor as long as you are running no lead
  auto gas.   A switch somewhere can turn it off and on. 100LL
  destroys O2 sensors.

     We got 8 channels of A to D and we are using only six.

     Paul Lamar

    I've run wideband sensors on cars with leaded race gas for
  considerable time with  little ill effects.  To be sure, eventually
  the lead will coat the sensor, but if  you're only using it for
  tuning, you can remove it once you're finished and plug the  bung.

    Isn't there some built in dwell time compensation in the ms
  software? I was using the  ms+spark-extra version. Have you checked
  that one out? It was several years ago, they  may have integrated it
  with the baseline ms by now.

    http://www.msextra.com/doc/index.html#ms1


http://www.msextra.com/doc/ms1extra/MS_Extra_Basic_Configuration_Manual.htm#rotary

    Josh

    Thats what I had in mind. By rearranging the chips I am going to
  wind up with a lot  of real estate left over. The cost will be lower
  as well. Use of the CD4050 buffer  instead of transistors  helps a
  lot as well.

    Yes on the dwell but I haven't found the source code yet. I
  suspect that was done in  C version. I have to support this so I am
  sticking with asm.

    Paul Lamar

   Paul,

   Open this file and search for "dwell".  They've noted it fairly
  well. I can't make heads or tails of the actual programming without
 studying  asm.

   Josh

   I did and this is what I came up with.
    ldhx    #0
           sthx    dwelldelay1
           sthx    dwelldelay2
           sthx    dwelldelay3
           sthx    dwelldelay4
           sthx    dwelldelay5
           sthx    dwelldelay6

   Aha they used a look up table.

   clr     lowresH                        ; low res (0.1ms) timer
           clr     lowresL                 ;
           lda     dwellcrank_f
           sta     dwelldms                ; initial dwell period
           mov     #$10,dwellush           ; } high speed dwell delay,
                                           ; default of 4.1ms
           clr     dwellusl                ; } until calc in main loop
           bset    SparkLSpeed,SparkBits   ; At boot turn on low speed
  ignition         clr     RevLimBits

   From the looks of it they struggled with the problem a bit :)

   They are using a timer that increments or decrements every .1ms.
  Probably what happens is they preset the counter in the timer as a
  function of RPM. Low count low RPM. High count high RPM I think Then
  the timer interrupts the back ground program to fire the plugs when
  it times out. Presto bingo.

   Thanks Josh. I think I got it.

   Here is how the Timebase Module works in the MC68HC908 :
   Quote the MC68HC908 data sheet.

  Paul Lamar
   ----------------------------------------------------------
   Timebase Module (TBM)
   16.1 Introduction
   This section describes the timebase module (TBM). The TBM will
  generate periodic interrupts at user selectable rates using a
  counter clocked by the external crystal clock. This TBM version uses
  15 divider stages, eight of which are user selectable.

   16.2 Features
   Features of the TBM module include:

   [UTF-8?]• Software programmable 1-Hz, 4-Hz, 16-Hz, 256-Hz,
  512-Hz, 1024- Hz, 2048-Hz, and 4096-Hz periodic interrupt using
 external 32.768- kHz crystal (We have)

   [UTF-8?]• Configurable for operation during stop mode to
 allow periodic wakeup from stop
   16.3 Functional Description

   NOTE
   This module is designed for a 32.768-kHz oscillator. (that is what
  we are using)

   This module can generate a periodic interrupt by dividing the
  crystal frequency, CGMXCLK. The counter is initialized to all 0s
  when TBON bit is cleared. The counter, shown in Figure 16-1, starts
  counting when the TBON bit is set. When the counter overflows at the
  tap selected by TBR2:TBR0, the TBIF bit gets set.

   If the TBIE bit is set, an interrupt request is sent to the CPU.
  The TBIF flag is cleared by writing a 1 to the TACK bit. The first
  time the TBIF flag is set after enabling the timebase module, the
  interrupt is generated at approximately half of the overflow period.
  Subsequent events occur at the exact period.

   -------------------------------------------------------------------------
   End of quote.

   Piece of cake :) If I get this working that will allow using cheap
  LS1 or RX8 coils. MSD CD's are expensive. Thanks for your help Josh.

   Paul Lamar

  Josh,

  I think they use a dwell time delay timer to start the actual dwell
  angle. So higher rpm actually uses less delay, and lower rpm should
  be more delay.

  Years ago I got totally confused when I first found "timing" used in
  relation to "angle".

  Since engine is mechanical, but ignition is electronic, they
 actually live on two different "time" domain. Mechanical devices
 base all timing on distance traveled. Therefore, its timing is based
 on angle traveled, rather than time as in seconds or fractions thereof.

  Since ignition coil needs time "K ms" to build up its energy
 content, when translated to mechanical timing in degrees, we have to
 change the amount of degrees in order to get a consistent K. Thus
 the advance in Dwell angle at higher rpm to keep "K" a constant time
 to charge up the coil.

  The timer routing uses absolute time. Each revolution has a variable
  amount of time depending on RPM. As RPM goes up, time becomes less.
  To get the proper Dwell Angle, or a fixed time of K, we have to calculate
  the time needed for a revolution at a given RPM, then subtract the
  necessary Dwell Angle Time "K" to get the Dwell delay timer. I think
  you are pointing to the Dwell delay timer thus the reason why it is
  should be smaller at higher rpm.

  Hope this makes sense and may be useful when someone tries to tune
  their engine. As I think this is also true for fuel injection timer.

  Henry Nee

  Hi Henry,
  How's it going. I am thinking of coming over there for awhile. How
  much would an apartment  cost me in Shenzhen by the month?

  Here is a simple QB program that calculates the numbers.

  CLS
  PRINT "RPM     RPS       Ms Per Rev    Deg per Ms "
  FOR RPM = 2000 TO 8500 STEP 500
  RPS = RPM / 60
  MsPerRev = 1000 / RPS
  PRINT USING "####"; RPM;
  PRINT USING "######.###"; RPS;
  PRINT USING "#######.###"; MsPerRev;
  DegPerMs = 360 / MsPerRev
  PRINT USING "#########.#"; DegPerMs
  NEXT

  Also check out the attached table file. The trouble with the Mac is
  they don't automatically display attached files so you missed this
  table in my last message :)

  Paul Lamar

  Could you use one pair(2 rotor) of Hall sensors at 97* BTC to turn
 the coil charge current on and another pair to shut off or fire the
 coil BTC. My thinking is this; 22* BTC for the timing and 45* for
 the charge time at 7500RPM as per your chart. The dwell at low RPM
 is high but should not be a problem as the amount of run time at low
 RPM in AC use is minimal.

 Dale Davies

 Right now there is only one external interrupt in the MC68HC908.
 That would work but we would get too much dwell at lower RPM with
 just two and the RX8 coils are susceptible to burn out. The dwell
 should be in the range of 10 to 2 ms. I could use my SuperKim
 interrupt encoder and have 8 external interrupts so there could be 8
 hall effect sensors on different radii. Then one would select which
 int to turn on depending on the RPM.

 I might just put that on the board anyway as that would greatly
 expand the power of the MC68HC908.

 BTW I heard from a guy in Oz recently that used the SuperKim to
 control a robot way back when.  reubenh@cyberneticzoo.com

 http://cyberneticzoo.com/?p=5372

 Paul Lamar



 I figure two radii. One for low speed and idle of 2000 RPM with sensors at
22* and at 67*. This gives 3.75ms at 2000RPM. By 3500RPM it would be down to
2.14ms dwell time. Then switch to a radii with 22* and 102* sensors which are
90* apart. This will give 2ms at 7500RPM and at 3501 RPM the dwell will be
4.29ms. The coils should not heat up at these dwell times.

Dale Davies



-- The Rotary Engine News Letter. Powered by Linux. ACRE NL web site. http://www.rotaryeng.net You Tube key word PaulLamar2 Copyright 1998-2012 All world wide rights reserved.