Subaru Impreza 1996, MT, EJ20, sensor troubleshooting

[Tremmor]

Hi again!

Excellent learning situation manifesting itself in form of : Impreza with warm engine cranks but doesnt start until some time (5-15 mins) has passed.
Its been happening intermittently in +20, 0 and -30 degrees of Celsius. By reading this wonderful forum found that probable cause could be engine coolent temperature sensor, I
replaced it with new one and cleaned crankshaft and camshaft position sensors from oil and dirt. Car ran much better and responded well to throttle, but it
didnt stop the problem.

Instead of replacing stuff I Want to diagnose stuff like a boss when problem occurs next time and be able to pinpoint problematic part. I removed passanger side metal panel from floor and found ECU with text 2G, 22611, AC540, A18-000 D15, 6212, Unisia Jecs Corporation , Made in Japan. Then I went to look for ECU pin order and found few wrong ones but photos from Ausubaru.com were promising.


http://www.ausubaru.com/forum/archive/index.php/t-2765.html

Thanks for user jzk25 for these,  
http://i65.photobucket.com/albums/h219/AL_RS/96EJ20NAMan2002.jpg
http://i65.photobucket.com/albums/h219/AL_RS/96EJ20NAMan2005.jpg
http://i65.photobucket.com/albums/h219/AL_RS/96EJ20NAMan2006.jpg
http://i65.photobucket.com/albums/h219/AL_RS/96EJ20NAMan2007.jpg

I tested TPS with usb-oscilloscope and laptop and it drew smooth curve as I pushed pedal.
Gonna later post screenshots of crankshaft and camshaft position sensor results. But I didnt get logical results from O2 sensor pins, if I recall they were something like
12 voltages but gonna verify pins and measure again to make sure. When I disconnected o2 sensor I was suprised that it didnt light CEL.

Could someone verify if this statement is true or does it apply in my case.
A faulty oxygen sensor can reduce your fuel economy by 30 to 40 percent and drastically reduce performance.

 

[Fairtax4me]

O2 sensors can cause poor fuel economy and power loss. They do not affect starting because the sensors do not work until they're hot. The ECU doesnt get any O2 sensor readings until the sensors are warm. Once the sensors start responding the ECU will use those readings to adjust the air/fuel ratio, and will go into closed loop operation.

 

O2 sensors have 1 signal voltage wire, 2 wires for the heater element in the sensor, and sometimes 1 signal ground wire.

 

The signal voltage wire will show 0.0-1.0v and will fluctuate.

 

The heater element wires will be 12v. One is supply voltage, the other is ground. These wires will be the same color. The supply voltage for the heater element usually turns off after the sensor is warmed up.

 

O2 sensor codes are not typically set immediately if the sensor is unplugged. Most of them are two trip codes (the fault has to be present for two drive cycles), or the O2 sensor monitor has to run (O2 emissions monitor has certain requirements which are not always met in one key cycle) before the code will set.

 

Typically a crank with no start is the ECT sensor, or a crank or camshaft sensor. But the crank and camshaft sensors will set codes immediately if they have a problem that prevents the engine from starting.

 

Get a spark tester and the next time it won't start check it for spark.

 

When you replaced the ECT sensor, did you replace the one wire sensor or the two wire sensor?

[Tremmor]

Thank you for good explanation Fairtax! I replaced 2 wire ECT sensor, the 1 wire sensor next to it was for dashboard temp gauge and it works ok. I fiddled with oscilloscope and here are the results:

 

Crankshaft sensor has a crack where its bolted into engine. Is it just cosmetic or cause of ittermittent no start.

 

Crankshaft idle:
http://oi57.tinypic.com/9tih5t.jpg
Crankshaft 2000rpm
http://oi61.tinypic.com/25uhe7q.jpg


Camshaft idle
http://oi60.tinypic.com/2rpuwb7.jpg
Camshaft 2000rpm
http://oi59.tinypic.com/23suop2.jpg
 

 

From the youtube videos where people test O2-sensor, I dont think results should look like these, hopefully i tested right pins. Sensor has 2 white wires and 1 black.

 

o2 idle
http://oi58.tinypic.com/2m848yw.jpg
o2 idle 2
http://oi59.tinypic.com/osaf44.jpg
 

[Fairtax4me]

For the O2 sensor set time/div to 200ms and see if the signal looks better.

Front sensor when its warm should swing between about 0.3v and 0.8v. Should swing a couple times a second.

 

Also are you checking the front sensor or rear sensor?

 

Or is the front sensor the first pic and rear sensor the second pic?

[Tremmor]

Its number 21 in

 

http://opposedforces.com/parts/impreza/en_g10/type_57/exhaust_system/exhaust/illustration_3/

 

 I think its only one there is. Gonna get better pictures.

Edited March 6, 2014 by Tremmor

[Tremmor]

Had to verify if I was measuring right pins at ECU by disconnecting O2-sensor, I was looking to see voltage to drop to 0 but it didnt change. Then I hooked scope clips into sensor black and white pins, beautiful graph but here it makes absolutely no sense: it measured 35V. Reconnected probes and scope, restarted scope program, still 35V, and then measured  about 1.6V from AA-battery. This must be some extraterrestrial technology or invention from secret Subaru laboratory.

 

 

edit: If two white wires are for heating, black is signal wire I should be probing black wire and engine ground, right?

Edited March 6, 2014 by Tremmor

[Fairtax4me]

Yeah. Black wire is signal. You can ground to the engine or to the battery -. The O2 sensor creates voltage so it doesn't matter if its hooked up to the ECU or not. It doesn't even have to be on the car. You can heat it up with a torch in a vice and it will create a voltage signal.

 

I don't know where you got 35v from. Which two wires were you checking that gave 35v?

Edited March 7, 2014 by Fairtax4me

[Tremmor]

Either of two whites and black wire gave that 35V. Lets see if I get same results today =)

[Fairtax4me]

I think you must have just confused your scope by not having a ground.

 

The ECU controls the ground side of the heater circuit, so when the heater is off there will be 12v on both wires.

Connecting to either of those and to the signal wire with the same scope channel will give you strange results because both wires will have voltage. The scope will think its reading some kind of strange AC pattern.

This is why its very important to have a good ground for scope testing.

[Tremmor]

Got more logical results by connecting black wire and actual ground, at least scale was 1V, but it was pretty much straight line even if engine was warm at idle, if i revved a bit voltage got to 560mV but nothing to describe as smooth swinging so guess its time for new sensor.

[Tremmor]

Tried to measure spark plug wire resistance for cylinder #1 and found no continuity nor resistance with multimeter so I ordered new plug wire set. Spark plug testlight shows spark in all plugs. I couldnt get resistance even for coil pack posts (Diamond F-569) when testing it even after using contact spray and light scuffing with sandpaper. So it seems kilovolts jump over two gaps or more for cylinder # and minimum of 1 gap for others. Does coil pack ground itself with 4 bolts to intake manifold? Underside of coil pack had paint flaking off around bolt holes. How long lifespan do coil pack have anyway? Wonder if it would be safe to get one from wrecking yard or get new one that is.