cold start weirdness

[SJ]

> From: Dan Cordon <cord4530 at uidaho.edu>

> My 87 5k has just the opposite problem. Long starts when normal temps
> outside, but super quick starts when cold. My problem seems to be leaky
> injectors. When it's cold outside (anywhere from ~30° to -13°) it fires
> right up, but runs slightly rough for the first second or two until the
> main injectors start going.
**** First and foremost . . .. you have a predictable problem. It happens
this way every time.
My problem(no start after short warmup) is intermittant. With environmental
conditions the same . . .sometimes I have the problem, most of the time I
dont.

 I dont understand why leaky injectors would cause this problem. Are you
saying that residual fuel pressure is the problem . .the injectors bleed off
the pressure over time?
If you turn ignition on, but dont start the engine. Does'nt the fuel pump
pressurize the system?
If you wait a few seconds before starting the engine, is not the system
fully pressurized now? Why would a leaky injector have an effect on not
starting? Leaky injector might provide a richer mixture . .. but that is
exactly what you want . . .firing the CSV adds that additional fuel and the
engine fires up. If the injectors leaked more, you wouldnt need the CSV to
start the engine.
If I remember correctly, leaky injectors will cause hard starting when
engine is hot and then shut off and left to sit for a while. The drop in
pressure causes vapor lock in the lines.

Your problem appears to be no signal to the CSV for extra fuel. The ECU is
not grounding out the CSV. Its like a switch shut off in the ECU. The temp
sensor should read about 5000 ohms at 30 degrees F, 14,000 ohms at -2 F. At
60 F, it should read 3,000 ohms.
Easy enough to check the resistance of the temperature sensor.

> As a test, last summer I hooked up a relay to ground the CSV whenever I
> had 12V to the starter relay. Really quite simple.....using a 4-prong
> relay, connect the starter signal to trigger the relay, and ground the
> other three pins. It worked great for me. Started right up on warm days.
**** This proves that the ECU is not firing off the CSV. Maybe all you have
is a broken ground wire going to the ECU?
Another thought. The ECU also controls the Differential Pressure Regulator.
In effect, that acts like a choke to richen the mix on startup. When its
warm, you might not need the CSV. But you need the DPR to richen the mix a
bit to start.
The DPR has a limit to how much it can richen the mixture. At that point,
you need the CSV to cut in for a shot of extra fuel.
You said that the engine runs rough for a few seconds. That would indicate
such a lean mix. Might make sense to see what the DPR current is doing at
startup for various temperatures.

I can see the temperature sensor as the common link. The ECU fires the CSV
based on what the sensor reads. The ECU varies the DPR current based on what
the sensor reads.

>
> Either way, if you're okay with the hack, it seems to work fine, though
> a momentary push button would be a lot quicker to instal, and give the
> user more options. I tend to agree with Hue though. Get a noid light set
> and diagnose what's going on with the CSV, or use a multi-meter/spark
> tester to see if ignition is the real problem.
**** We have two different conditions. Yours and Huw's . .. which is
predictable and constant. Mine, which is intermittant. And when my car
fails, I'm on the road with no tools.
 My method of troubleshooting requires not finding the actual problem, but
eliminating what doesnt. This is like electrical trouble shooting. Divide
the circuit in half. If you get a signal in the center . .the problem is
after the test point. If you get no signal , the problem is before the test
point. Then you divide the circuit in half again . .. etc etc. Eventually,
you narrow it down to a very specific area.
If hitting the pushbutton gets the car to start . . .I know its not
ignition. I just eliminated half the system.

I just thought of another idea . . .attach an LED to the pushbutton switch.
The LED bridges the CSV terminals . . . you need 3 wires to do this. When
ECU shorts out the CSV to fire it, the LED  lights up. If the LED dont light
up, you know what the problem is . .and hit the switch.
The LED would be useful to see how long the CSV stays on for different
temperatures.

Whats a multimeter/spark tester? I did a search and all I can find is an
in-line spark tester for around $7. I might get one of these . .. easy way
to determine if you have spark. Looks like it works by lighting up a neon
bulb.

SJ