Newbie extends greetings

[Neil Bradley]

Greetings fllow DIY EFIers...

I thought I'd take a moment to introduce myself and my DIY/EFI
project. I'm quite a bit of a newbie at this, but I've been
studying/reading up for a month or so now and would like to bounce some
ideas off the "experts" here. ;-)

I'm working on an ignition/fuel injection/turbo control system for a 3
rotor rotary engine. The engine is mounted in a dragster and is using a
bank of MSD boxes, along with having 3 turbos (one per rotor).

It must run 9 fuel injectors and 6 spark plugs. The system must tolerate
15000RPMs without any misfirings. It uses a throttle position sensor,
manifold absolute pressure sensor (for boost/vaccum), water temp, oil
pressure, and air temperature. It uses a crank angle sensor with 20 degree
marks and a TDC sensor as well. I intend on wiring the 20 degree marks to
an interrupt on the main CPU so I can get very accurate timing.

My current plan is to use a Philips XA-G49 CPU as the main "drive" CPU and
an Atmel AVR-8535 (8MHZ version - can't remember the specific model #) to
handle the A to D and the pulse width generation. I looked at the Motorola
parts, and the TPU is way attractive, but the cost of the chip is too
expensive, the manuals are a mile thick, and the tools for them really
suck. I decided to go with the XA-G49 for a couple of reasons:

* On-chip flash programmable (just throw a bit and off you go)
* Built in multiplies and divides
* Serves as a programmer to the AVR device

And the AVR:

* Built in A to D
* Fast and can generate really accurate pulse widths for doing ignition
  and fuel injection. This could be done with the XA-G49 but the XA part
  only has 3 timers, and some of the ignition and injector firings will
  overlap at high RPMs, making multiplexing the timers quite a bit
  tougher.

The XA-G49 will tell the AVR to "Generate a 0-1 or 1-0 pulse for x clock
cycles RIGHT NOW", so all the timing in terms of when it happens is
controlled by the XA part, and the work of changing the pulse's states is
done by the AVR part.

We plan on using a National Semiconductor LM1949 for the injector drive
circuit, along with the National LM9011 for the ignition interface from
the crank angle sensor. We also want to include a knock sensor, and so far
the HIP9010 (Honeywell/Intersil?) part seems to do what we need. If we get
a knock, we want an interrupt to occur so we can back the timing up
immediately, because on a rotary, knocks tend to destroy engines if not
caught extremely quickly.

So my questions are as follows:

* What do you think of my approach/design for my application?
* Know of better interface parts?
* What is the shortest digital pulse that would ever be needed for an
  injector? Would 250 microseconds be long enough? I'm trying to nail down
  injector pulse length granularity so we can figure out our requirements
  for the "slave" pulse width generator.

Thanks in advance! Looking forward to contributing somehow...

-->Neil

-------------------------------------------------------------------------------
Neil Bradley            "Devo was exquisitely stupid. But we knew we were being
Synthcom Systems, Inc. 	 stupid, whereas a lot of people are stupid and don't
ICQ # 29402898           realize it." - Gerald Casale - DEVO Cofounder


----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org