Daydreaming at Cone Shaped Hat Headquarters

[Joe Boucher]

After looking at some of the documentation from some of the GM turbo ecm's,
I think just figuring out the tables in the 747 is a daunting enough first
step task.  After that, redoing the code could be a follow on project.
Let's gang up on a simpler task.

Joe Boucher
'70 RS/SS Camaro  '81 TBI Suburban


At 04:07 PM 3/26/98 -0800, you wrote:
>"Hey You!" <nacelp at bright.net> wrote:
>>   To me, the perfect plan is this take a 1227747, and a minimum
>> chip for it.  Why a 747, here's why.  It has a pair of P+H outputs
>> it's MAP, can be used with 6s or 8s, has a Knock sensor.  With
>> a little work I think we could interface a ford MAF for those that
>> have to have that.  It would be a P+P for alot of applications.
>> It has room for TTC (and some oem appl didn't use that), so it
>> would work fine with a manual trans.  It's cheap, and easy to find.
>> It's been well hacked.  There are other advantages to this one,
>> but for BW I'll refrain.  IF this goes well then we can try the larger
>> prom'd ones, or do a turbo ecm, and I got a bunch of ideas for
>> further projects this being the leap off one!.
>
>It's not clear if you want to use the existing GM code in the '747, or
>to "roll your own."  If the former, you're pretty much limited to
>applications similar to the OEM usage.  It becomes just a table hacking
>project.  You haven't moved away from the status quo where everyone
>picks the OEM ECM best matched for their project.
>
>If you want to "roll your own", there's 8K bytes of soldered in ROM in
>the '747 that wants to do things it's way.  [Stare at disassembly]  You
>_could_ make an EPROM that intercepts the ROM and never returns.  After
>key-on, before interception, the ROM will have run it's init code and
>one pass of the spark function.  Your EPROM would have to dedicate a few
>locations to keep the ROM happy until you gain control.  All of your
>code and data would have to fit in the less than 4K bytes of remaining
>EPROM, though you can use a few subroutines in the '747's ROM.  You also
>won't be able to use the (one) hardware interrupt, though the GM
>hardware will let you poll for it.
>
>If you want to "roll your own" and not be bothered with any of the ROM,
>there's a way.  The 1226869/1226870/1227170/1227302 qaudruplets have the
>EPROM (aka MEMPAK), CALPAK, and CPU on a little daughterboard.  There is
>also a way of electrically disabling the ROM.  I've thought of making a
>replacement daughterboard that turns off the ROM and provides a socket
>for a 27256 (32K*8) EPROM.  Then the ECM would be fully at my command. 
>I suspect the same thing is possible to do with the '747.
>
>But, I've decided that the daughterboard project is a stupid idea.  For
>the expense and time to make a daughterboard, I could obtain several
>1227730 ECM's.  They already have the ability to use a large EPROM to
>hold all the program code.  Plus, the CPU is faster, there is more RAM,
>more and better I/O, and a much faster ALDL interface.
>
>If you still want to write custom code for the '747, I've got the
>registers of the C3 chipset mostly figured out.  I'm willing to share
>this data after verifying it some more.
>
>As to the MAF, why use a Ford?  The C3 chipset in the '747 already has
>the ability to handle a GM MAF.  Worst case, you'd have to connect a
>CMOS buffer/interface between the harness connector and one of the 40
>pin DIPs.
>
>Could someone with a '747 trace out the following parts of the circuit
>board?  Find the 40 pin dip marked 16023263 (aka 23263.)  Trace out
>where pin 30 is connected.  (That's the MAF input.)  It may go to pin 14
>of a chip marked 16043538 (aka 43538.)  If so, trace out where 43538 pin
>15 comes from.  It should go through a resistor network and then to the
>harness connector.  If 23263 pin 30 instead connects to a chip marked
>16004773 (aka 4773), that is the output of a 4049 CMOS buffer.  Find the
>corresponding input and continue tracing it back.  There is also a
>chance that 23263 pin 30 is simply grounded.
>
>               unsigned long BinToBCD(unsigned long i) {unsigned long t;
>Ludis Langens     return i ? (t = BinToBCD(i >> 1), (t << 1) + (i & 1) + 
>ludis at cruzers.com            (t + 858993459 >> 2 & 572662306) * 3) : 0;}
>
Joe Boucher
'70 RS/SS Camaro  '81 TBI Suburban