5 seconds of cranking to start

[Cody Forbes]

cobram at juno.com wrote:

>  It's normal for the turbo to take up to 2.5
> revolutions to start, Phil the UR guru said it takes those first
> couple turns for the ECU to figure out what's going on before it'll
> let the engine run.
>

I got this ;-).

On the turbo engines the ECU uses 3 inputs to figure out whats going on. On 
the flywheel there's the RPM sensor and the TDC sensor, then in the 
distributor is the 1 window hall sensor for cylinder #1 TDC compression 
stroke, but you knew this much already ;-).

RPM sensor counts the flywheel teeth and very accurately determines the 
RPM's even when the engine is speeding up quickly enough that the RPM 
changes significantly before 1 revolution has taken place.

The TDC sensor uses a pin on the back of the flywheel to determine, again 
super accurately, when the crankshaft is at 0 and 180 degrees top dead 
center. Of course 2 crankshaft revolutions per one ignition event (per 
cylinder), so this is not enough information to figure out when to fire the 
spark plug.

The hall sensor in the distributor is a reletively *IN*-accurate sensor that 
just lets the ECU know when cylinder #1 is somewhere arround the top of it's 
compression stroke. This is only used once during starting, after that the 
ECU knows where cylinder #1 is and knows it's time to fire cylinder #1 every 
other time the flywheel TDC sensor triggers.


Remember the turbo cars are very intelligent comprared to the natually 
aspirated counterparts. The N/A cars used a 5 window hall sensor with vacuum 
advance that actually tells the ECU "FIRE THE COIL NOW DAMNIT!" - a "dumb" 
ECU that doesn't do many calculations, just relays information. The turbo 
cars' 1 window hall sensor say "Hey, heres cylinder #1 ready for you to fire 
the plug as you see fit", and the ECU determines from then on out what the 
actual timing should be based on the programmed curve and also based on 
input from the knock sensor. Cylinders 2-5 ignition event is based PURELY on 
math derived from the cylinder 1 inputs, there are no inputs at all for 2-5. 
This is why you can turn the turbo distributor a good 20 degrees with no 
change in ignition timing... the distributor is only there to spin the hall 
sensor and to transfer voltage from the coil to the proper plug wire, it has 
no internal advance mechanism. The only reason the engine will eventually 
stumble and die is if you turn the distrutor far enough that the rotor isn't 
aligned with the proper cylinder when the ignition event happens.


So to get to the point (finally, right?) you are correct in saying that the 
engine does need to spin over just a teeny bit before it will fire. The ECU 
has to wait for the hall sensor to trigger that magical cyl#1 TDC 
compression signal before any fun can begin. Depending on where the engine 
came to rest when you shut it down this could be an instant, or it could 
take a revolution to figure out. Don't forget that each change in pitch of 
the starters noise is NOT a revolution, just one cylinder on compression 
stroke. I think the misconceptions there are what leads some to believe that 
it takes a number beyond 1 full revolution.

Somebody should Wiki-fy this ;-). If nobody beats me to it I'll get it on 
there this weekend.


-Cody Forbes
http://www.5000tq.com
'87 5ktq - Fast.
'86 5ktqCD
'86 5k
'86 5k