Date: Sat, 7 Apr 2001 02:10:49 -0400
Reply-To: David Beierl <dbeierl@ATTGLOBAL.NET>
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: David Beierl <dbeierl@ATTGLOBAL.NET>
Subject: Re: Wet weather power loss
In-Reply-To: <092f01c0bee9$843b2560$175ec0d8@oemcomputer>
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 06:32 PM 4/6/2001, Helen Fahy wrote:
>wet weather, though it does not have to be actually raining, it can occur in
>heavy fog.
Ok, almost surely electrical
>tachometer is being driven, of course the engine is slowing down, but there
>is nothing erratic about the motion of the tach needle.
Ok, still getting some kind of ignition pulse at the coil
>turning ignition switch off an on brings back normal engine performance for
>a short time <<<Why does this work and is this the key to the analysis of
>the problem???>>>
This resets the ECU from whatever strange mode it may have gotten
into. That means it's not a fuel supply problem. It *may* also put the
system into open-loop briefly. ECU inputs are Coolant temp, Air temp, AFM
opening, rpm and TDC from Hall sender, lambda sensor, throttle switch (WOT
and closed throttle [should] give same signal). Outputs are fuel pump
drive (depends on Hall pulses), injector open pulses, ignition primary pulses.
Things that can go wrong -- bad inputs: Temp II open gives very rich mix
(lambda voltage will stay above half volt), shorted is too lean to run
cold. Lambda sensor grounded or low output (could include bad ground
connections at engine) gives very rich mix. Lambda sensor leakage from
heater wire should give very lean mix. AFM noisy or intermittent output
could give lean mix (but not rich, I think) and could put ECU in a panic
state where van loses power -- this is cured by cycling power to
ECU. Shorted/leaky cap at AFM would give lean mix. Note that AFM is
ignored above some rpm -- not sure what but 4500 should be safe. Temp I
(air temp) -- dunno how much effect it has, i.e. whether it can overpower
the lambda sensor which is normally the final arbiter on mixture. In any
case, open would make the mix rich I think, short would make it
lean. Shorted throttle switch should give rich mixture above some
rpm/throttle combination, severe surging below that but above 1500
rpm. Open throttle switch will make lean at full throttle, possible
backfire on decel with closed throttle (normally it shuts off the fuel on
decel until engine rpm goes below 1500 rpm)? Intermittent Hall sender
should show on tach if interruption is long enough.
Bad outputs -- wrong ignition timing, fuel pulses too long or short,
*alternate fuel pulses skipped* -- this is how the rpm limiter works.
Lambda sensor -- bad sensor (unplug it should make situation stable as ECU
goes closed-loop -- voltage at ECU should remain steady near half
volt). Bad wire (short/leak to shield) -- mix very rich, unplug sensor no
difference, voltage at ECU (probably) steady below half volt. Could be a
very high-resistance leak, this is a high-impedance circuit. Bad engine
grounds -- sensor is grounded through exhaust system to engine, then to ECU
pin 19 from engine head. ECU also grounded to body at pin 13; this is a
big wire and appears to be the power ground, where the other is the signal
ground for the lambda sensor and ?other signal inputs? -- AFM, Temp I, Hall
generator all grounded to pin 6 of ECU. The major concern is the lambda
sensor -- it runs from 0-1 volt prox. at very high impedance and the ECU
uses it as a threshold detector. If it gets stuck either above or below a
half volt, ECU will exhaust its adjustment range trying to shove the
mixture the other way. AFM and temp sensors are analog inputs with range
0-5 volts and much lower impedance. To be confident of what the ECU is
seeing, measure lambda voltage at the ECU btw pins 2 and 19 (not 13) using
10 megohm-impedance voltmeter (or scope with x10 probe...). However, if
lambda voltage measured anywhere is jumping up and down within a 1-volt
range then ECU is in closed-loop mode and is controlling mixture correctly
or nearly so.
Quickie description of digifant operation -- ECU takes rpm and AFM voltage
to develop a basic pulse length for injectors (all injectors fire at once,
once per revolution, and amount injected is controlled strictly by open
duration since fuel pressure is controlled to be constant relative to
manifold pressure). If rpm is too high for AFM it uses an internal table
based on rpm. It then corrects for air temp and water temp, and opens
injectors for corrected time. When running in closed-loop mode it then
adjusts successive pulses until the lambda sensor flops the other way, then
adjusts again until it flops back, forever. If throttle switch is closed
it ignores lambda sensor and richens mix by precalculated factor.
At the same time it uses internal maps of rpm vs AFM vs idunnowhat to
advance and retard ignition timing. When idling, the idle stabilizer unit
constantly adjusts input airflow to maintain constant rpm. It uses an
electronic control that drives an air valve that bypasses the throttle body
but not the AFM.
david
David Beierl - dbeierl@attglobal.net
| 
