ECU pulse width calculations...?

[Andrew Brownsword]

I finally got to try my modified program in the car's ECU today.. and 
(unfortunately) did NOT get the expected results.  The car is a '93 Ford
Probe GT (Mazda 2.5L DOHC V6 engine).  I had modified the program by
locating all of the places that the injector firing routines are called
from, tracing backwards to locate the last place that the calculations are
scaled by a ROM value before being passed to the firing routine, and then
scaling back the injector values by the ratio of the flow rates of the old
and new injectors.

Doing this, as I understand it, should have reduced the pulse widths
controlling the injectors and therefore decreased the amount of fuel being
injected.  We cut the numbers by fully half and measured the effect using a
(high end) A/F meter.  We expected to see the engine lean out significantly
at idle... instead it leaned out slightly.  From 11.5 up to 13.2 or so.  It
would momentarily jump up to around 14.5 and then slowly drop back down to
13.1 or so.

Does anyone have theories, advice, hints, etc?

Thanks,
  Andrew