How completcated does Efi HAVE to be?

[Will McGonegal]

> > >sensor, which means you need the flywheel, transfer case and the actual
> > >sensor.
> >
> > Surely it doesn't have to be *that* accurate? Why not use the coil? 2 per
> > Rev should be enough, you only need speed not accurate position?
>
> No, it does.  You have to have a very good accuracy (I think the minimum
> is 6 "steps" per revolution) because otherwise the system will get
> confused when the engine changes speed quickly.

I did some calculations to check into what kind of errors you'd get on your speed under constant acceleration if you used a single pulse per revolution

Speed [rpm]=> V1 =  60.0 / (T1 - T0)

Case 1:

Engine accelerates from 600 to 6000 rpm in 0.5 seconds (no load?!)

The first revolution takes 0.063597837 seconds.  The speed change is 600 to 1287 rpm.  With that time period you'd calculate the speed to be 943 rpm.  A 27%, or 344 rpm error.

After 27 revolutions (0.494977295 s) the engine turns at 5946 rpm.  One more revolution (0.010000405 s) later the engine is at 6054 rpm.  We calculate that speed to be 6000 rpm (1 %, or 54 rpm error)

Case 2:

Engine accelerates from 600 to 6000 rpm in 5 seconds (0 -60 mph?)

The first revolution takes 0.092327988 seconds.  The speed change is 600 to 700 rpm.  With that time period you'd calculate the speed to be 650 rpm.  A 7%, or 50 rpm error.

After 27 revolutions (4.989990984 s) the engine turns at 5989 rpm.  One more revolution (0.010009016 s) later the engine is at 6000 rpm.  We calculate that speed to be 5995 rpm (0.09 %, or 5 rpm error).

Since case 1 is probably not normal operation, the error there is probably acceptable.  If the case 2 error is still too high for a DIYer, then one could also calculate the rate of acceleration (from two (or more)
time periods) and factor that into the speed calculation to get a value that is closer to the actual speed.  I tried a simple acceleration calculation which reduced the error of the 1st revolution in Case 2 to 3
rpm.

More pulses per revolution will increase accuracy but do all applications need it?  I think you could do it with a pulse every second revolution and have it run better then a carburetor.

Will McGonegal