[Q] EFI computers on different engines

[Craig Pugsley]

******************************************************************  
* FIRSTLY A GENERAL QUESTION:                                    *
* DOES ANYONE HAVE A WIRING DIAGRAM FOR A TOYOTA FITTED WITH A   *
* 3T GTEU 1.8l Twincam turbo? I HAVE A COMPUTER (AGAIN ANALOG)   *
* TO SUIT THIS ENGINE, AND AM WONDERING WHAT IT'S EXTERNAL       *
* CONNECTIONS TO THE SENSORS ARE SO I CAN MAKE SOME MEASUREMENTS *
******************************************************************

Thanks for your reply Tom, interesting reading.      

> > He wants me to make up a box that tweaks various inputs to get the
> > fuelling correct.
> 
> [rest snipped just to save bandwidth]
> 
> Will this be an analog box to precondition the signals?

That was the initial idea (a bit easier to fiddle with rather than
having to 90% design a digital system before anything works).
See below for what I intend to do now.. 

> Have you thought about a uC box to replace the original? 
> Have you done any uC work?

I am about 1/3 of the way thru designing a microcontroller version (8051
family-the 332 stuff is a bit too advanced for me :-)
However, he wants a cheap and quick system up & going, hence trying to
use an OEM box and tweak it's inputs to make it behave appropriately.
Besides, the extra experience with oem boxes will help my understanding
for doing my own from scratch.

> [wants to remove air box vane meter... and replace it with
> a map sensor]
> 
> The early Nippon Denso should be the BOSCH L jetronic analog
> design. Contact switches for idle and full throttle should be on
> the throttle body. 

Yes, it's operation is virtually identical to an L jet. I got a generic
EFI repair book (lots of manufacturers, condensed info of connections,
voltages etc). Happily, it seems that these ecus are virtually
identical, with the diagram in the book conveiniently matching what I've
got on hand.

I got my hands on this box last night, it's got 5 analog ICs (Custom..
consecutive part numbers), several dozen C+R, a few transistors and a
diode or two. There is about 10 resistors on PCB pins, which I thought
was intriguing.. Then I realised these would almost certainly be the
calibration resistors. Hmm.. (light globe above head comes on) I could
make up a panel with a pot for each resistor, do a bit of reverse
engineering and figure out what each one does - so I should be able to
come up with something in a few weeks rather than #*&$ months/years!
If anyone is interested I'll keep the group informed of any progress.

OTOH, I've also got a digital ECU (from an 89 Mazda MX6 2.2l non turbo), so
if the inputs to that could be tweaked it might self learn enough to
work quite well.

An importer of japanese engines here have piles of 2nd hand computers
(at $US 15 each!! :-) and pumps, injectors etc. So, if I can get a workable
solution with one particular type of computer that's in ready supply I could
sell about 1/2 a dozen straight away (a few freinds are displaying keen
interest in EFI). 

> > My understanding of these trapdoor type systems is
> > that almost all of the fuelling calculation is based on the trapdoor
> > potentiometer, along with idle curcuit (switched by AFM or throttle
> > position), 
> 
> Yes.  Ideal would be to "bench test" the donor computer, with all the
> parts to get an idea of the functions of how much the air vane
> position modifies the amount of fuel injected.

Exactly what I plan on doing. I've already got a set of readings from an
aftermarket computer tuned for the sort of engine I'm trying to run
(Mazda 12A turbo rotary), so I've got a half decent start point.

> > Questions:
> > i/ If I have a MAP sensor and alter this signal ONLY and feed it into
> > the trapdoor potentiometer input on the ECU (ie use a MAP instead of a trapdoor
> > AFM), and leave all of the other sensors STOCK (unmodified), is this
> > system likely to function successfully?
> 
> If possible, you may want to make a series of measurements to get an
> approximate idea of the voltage vs. airflow from the air box from the
> donor system.  Once you know that function, then it would be NICE to
> have an actual system to compare manifold pressure vs. air vane voltage
> readings.  This should give an idea of the function between the two.

Hmm.. funny you should mention that ;-)
I was planning to get a throttle body, trapdoor sensor on one side, MAP
on the vacuum side, Vacuum cleaner/leaf blower to simulate the engine.
Then I'll vary the throttle and the vacuum to simulate different engine
conditions and take some readings from the AFM and from the MAP. If
there is a straight line corellation then an opamp solution would be
adequate, but if it's a curve then I think the LM3914 solution would
offer a much better degree of tunability. (I HATE non linear analog
electronics) 

Regarding the pulsing in the manifold, I think that could be smoothed
electronically or with a small damping chamber in the vacuum line.

> There are those on this list who are much better qualified than I,
> but thought I would at least toss out a few ideas to get the ball
> rolling. 

Hey, sometimes the 'grass roots' approach is easier to understand
than the hi tech approach. Your input is very much appreciated.

Cheers,
Craig.