Select Monitor Data and Lambda trouble

[trac]

Good day,

This is about some Select Monitor Data that I've collected from my Subaru with 1992 Legacy Turbo 2.0 engine.

I've made an overview of all the menu's available, what values I should be getting and in red my actual measured values (ignore these for this problem).

All these values are coming from the Legacy I FSM




In chapter 2.7C also this overview of the 4 switch input menu's:



My active switches (Led On) with warm engine idling, transmission in neutral:


FA0:

1. IG

7. NT


FA1:

1. ID

6. FP


FA2:


FA3:

10. OR

Some questions and remarks on these values:

- I put in a supposely good o2 sensor in my car. When I start the car, the A/F (lambda) voltage goes from 0.30V steadily to 0,0V.
- With this happening, my ALPHA value, after climbing to say 18%, also steadily drops to 0,0%
- After the A/F and ALPHA gone back to zero, my LoadData ends up at 12, way under the specified 30-50.

- Regarding F11, why is my injector pulse width lower than specified?

- Regarding FA03; LED 10 is on, indicating a rich mixture, even when my lambda voltage stays at 0,0V. Why is it indicating a rich mixture while my lambda voltage is dropping to zero?


I could use some Select Monitor Guru input here, or some verification of someone with some data of a Legacy I Turbo or MY95/96 WRX.

For some reason my car's been running crap for some time now, and after checking and replacing many things, this lambda, ALPHA and LoadFractor value is the best I can find what's currently clearly out of spec.

Hope someone can help me on this. Has anybody earlier see his lambda signal falling back to zero during warm up? Let me know.

gr. Bram

[Legacy777]

Hi,

You may want to check out this thread of mine.
http://bbs.legacycentral.org/viewtopic.php?f=19&t=24376

Regarding your numbers. The things that stand out to me are the engine load data and the injector pulse width. In my opinion, I'd be suspect of the O2 sensor. Is the O2 sensor you installed brand new? Was it a Subaru sensor or aftermarket Bosch, NGK, or something else?

Is your car an automatic or manual trans?

What I'd maybe suggest doing is unplugging the O2 sensor and see what the car does and what readings you get from the select monitor. You may get a check engine light, but it should force the ECU to use its lookup tables, which tend to run a little rich, so your injector pw should go up.

Have you tried resetting the ECU? i.e. disconnecting the battery for an hour or so?

[trac]

Hi,

You may want to check out this thread of mine.
http://bbs.legacycentral.org/viewtopic.php?f=19&t=24376

Regarding your numbers. The things that stand out to me are the engine load data and the injector pulse width. In my opinion, I'd be suspect of the O2 sensor. Is the O2 sensor you installed brand new? Was it a Subaru sensor or aftermarket Bosch, NGK, or something else?

Is your car an automatic or manual trans?

What I'd maybe suggest doing is unplugging the O2 sensor and see what the car does and what readings you get from the select monitor. You may get a check engine light, but it should force the ECU to use its lookup tables, which tend to run a little rich, so your injector pw should go up.

Have you tried resetting the ECU? i.e. disconnecting the battery for an hour or so?

Thanks for your reply, to answer your questions:

This sensor is a Bosch, a used one that I bought for testing purposes. It got the right connector, so I figured it should work on my car.

I can indeed run a similar test with the sensor unplugged and see what it does. However, is it correct that the ECU doesn't give a fault code when unplugging the 02 sensor? I don't think mine does looking at earlier tests.

The car is a manual trans.

I also can verify my reading with your values. Next weekend I will look into it, but still can't figure out why the lambda mV signal is dropping steadily to zero, and at the same time the led light is indicating a rich mixture. Curious to find out what's behind all this.
I can also test another, good working sensor from another Leggy Turbo, to exclude the possibility this sensor is also faulty.

greetings, Bram

[Legacy777]

Bram,

You will probably find that the ECU will not give a fault code when unplugging the O2 sensor. When doing this, the ECU won't take into account the data from the O2 sensor. I'm curious to see what type of data you see when you unplug it. Like I mentioned, it will help rule out the O2 sensor or maybe point you to something else.

[trac]

I tried another lambda/o2 sensor.
This is a Bosch pn. A24-A71 107 48310 (printed on the shell).
The previous old sensor also was a Bosch, pn. A24-631 209 22272.

Both these sensors have the correct 4 pin plug with the three wires connected. I assume I got the correct sensor.

Anyway, when hooking up the new A24-A71 107 48310 I can see sensor values (after starting and idling engine) on my select monitor of 0,20 to 0,30 V for a number of seconds, then it slowly starts to drop to an o2 signal of 0,0 V.
Also my ALPHA value then drops to 0,0 % and stays there. LoadFactor remains on 12/13.

I again tested all the o2 sensor wires, even the shield, but nothing wrong there.
The LED light on the select monitor keeps on indicating ‘rich mixture’.

Has anybody seen such symptons of a dropping lambda/o2 signal?
Any clues what may be wrong here?

In a few weeks I can test another o2 sensor and another turbo ECU, but I’m still trying to figure out what’s really going wrong here.
Can it be that my injection system (injectors) are faulty? I don’t get any fault codes whatsoever.

Next weekend I'll try the select monitor with the 02 sensor disconnected.

Thanks Bram

[Legacy777]

If your alpha % is 0, that's fine and really where it should be. The O2 sensor values should cycle back and forth and from your description, it sounds like everything is working fine. The O2 sensor LED light is supposed to blink intermittantly. That's the way the narrow band sensors work.

Not sure about the load values, but the O2 sensor data you described sounds normal.

[cj91legss]

Sorry for not understanding what type of hardware you guys are talking about.... But seriously what are you talking about?

[Legacy777]

The select monitor is the Subaru tool used to read the ECU parameters. It's basically the same thing that Vikash's scan tool reads from the ECU, but instead of needing a computer, it's a stand alone device.

[cj91legss]

Ok I understood about the select monitor stuff, just wanted more info on this device. Is it expensive?

[Legacy777]

You really can't buy them any more. I picked mine up on ebay for around $300 or so and several cartridges. You have to have the cartridges for the specific application you want. The select monitor can talk with the ECU, TCU, cruise computer, air suspension computer, and automated climate control.

They have a new select monitor that Subaru uses, and it's a couple thousand dollars.

[trac]

I bought mine for e 50,- with a few cassettes. There are quite a few Subaru dealer cleaning up their inventory, so many old manuals and tools like select monitor becone available then.
As also many people not know what the select monitor is and what you can do with it, you can pick up a nice deal on those. Really helpfull diagnosing the engine.

But regarding the lambda; it's not normal I think. As said, it starts at 0,20/0,30V and then per second drops by say 0,4V so after a number of seconds it's dead still at 0,0V. There's no cycling of the voltage whatsover.
Together with the dropping of the lambda, also the Alpha drops to zero gradually with increments as the lambda does.

I learned from my manual that a constant value of alpha=0,0% means a broken 02 sensor. I may try another sensor, but can't explain the symptons of that dropping voltage.

Bram

[Legacy777]

If your O2 sensor values are not changing (oscillating) back and forth between .1 & .9, then I would agree you likely have a problem with the O2 sensor, or possibly the wiring or ECU. I'd start with the O2 sensor and check the wiring.

The 0% alpha correction is fine, trust me. Alpha correction is an indication of how much fuel the ECU adds or removes to the base fuel calculation to reach the proper stiochiometric air fuel ratio of 14.7 to 1.

Honestly, I think you need to focus more on the problem of how the car is running, vs. the numbers. When troubleshooting, it's common to try and fix a symptom vs. trying to fix the problem. Can you describe your problem a little more. I know you said it has been running like "crap", but if can provide a little more details, that will help.

[trac]

Hi josh,

Thanks for giving me some feedback and things to go on. Much appreciated on these hard to solve problems!

Regarding your last post and questions:
Indeed i reported a lousy running engine before, that was however on the previous engine which had some bad compression on two cylinders and a faulty coil.
I good a good deal on a fresh fcd engine with sti internals and such. That was installed  with fresh coils about 1000 kms ago.
Broke the engine in with open wastegate and not going above 2500 rpm.
Then gradually ran on wastegate pressure, but I already noticed it was still slow on boost. This was kind of confirmed looking at the lambda voltage not going above 0.75v when accelerating. Later btw i also noticed my throttle valve was not reaching it 's limits, so the ECU might not have gotten the 'pedal to the metal' input.
Anyway, similtaniously i noticed the lamdba behaviour and to be honest, kind of affraid to do some engine damage when these values acting so weird.
I still would like to see everything work as the manuals and experts say.

Still gaining knowlegde on this and i wil work it out i'm sure, but a little clueless at the moments as it comes to seeing the values and not finding any errors in my connections.

Gr. Bram

[Legacy777]

You could try resetting the ECU's memory, which should eliminate any past learned values that may be causing you issues.

Can you provide a little more information on this statement:

Later btw i also noticed my throttle valve was not reaching it 's limits...

Without a wide band O2 sensor, the information being fed to the ECU is only as good as the narrow band O2 sensor, which isn't that good. Running a wide band O2 sensor even if only temporarily will give you an idea of the AFR's and how the engine may be running.

[trac]

You could try resetting the ECU's memory, which should eliminate any past learned values that may be causing you issues.

Can you provide a little more information on this statement:

Later btw i also noticed my throttle valve was not reaching it 's limits...

This I noticed by checking the F10 function on the Select Monitor. From fully closed to fully open it should read a signal voltage of 0.9 to 4,7V. I noticed my fully closed signal was a little off, and my fully open signal didn't reach the 4,7V by far. Adjusted both.

Without a wide band O2 sensor, the information being fed to the ECU is only as good as the narrow band O2 sensor, which isn't that good. Running a wide band O2 sensor even if only temporarily will give you an idea of the AFR's and how the engine may be running.

Next step when trying another 'approved' lambda, is indeed install a wideband (which can also submit narrowband signal) and see what is measured.

Today I noticed btw, after the car is parked for a few days, it starts instantly upon turning the key. Perfect.
Then trying to start it after a few hours, few seconds cranking, quite some blue smoke out of the exhaust.

Is this a clue or a whole different problem

Bram

[Legacy777]

Out of curiousity how old is the coolant temp sensor? When these things go, they typically don't give a code and just feed the ECU incorrect informaiton which will either cause them to run rich or lean. There is A LOT of additional fueling corrections and other stuff that are based upon the engine temp. So if it's off, it will cause you a lot of problems.

Regarding the throttle signal, it's normal for those values to be off a little bit, it doesn't matter one bit. The ECU adjusts things based upon the startup value when the throttle is closed. The only adjustment to the throttle position sensor is to adjust the idle switch. You set things for the idle switch and you'll be good.

Here's the scan for the idle switch
http://main.experiencetherave.com/subar ... sting3.jpg

and the begining part of the tps diagnostic if you desire.
http://main.experiencetherave.com/subar ... sting1.jpg
http://main.experiencetherave.com/subar ... sting2.jpg

[trac]

I did a test session using in total 3 Legacy Turbo ECU 's and two lambda sensors.

Basicly came out with more questions than answers.

In different combinations it came out that:
- lambda value in one combination started at 0,30V and dropped incremently to 0V after say 30 seconds.
- other combinations showed a lambda value of a constant 0V or maybe a little movement around 0V to about 0,06V, further no cycling seen on no combination.
- Alpha value being steady at 0.0% or climbing together with revving the engine or showing irratic beheavior. One time it was following the revving of the engine (going up and coming down to 0,0% then stayed at 0.0% while revving. Put contact on and off and it was coming up and down again with the revs...very strange.
- Basicly all the other values where the same and showing good values. The load data at idle was 12 and also climbing and coming down with the revs.

I'm gonna get some data from someone else his Legacy 2.2 and his 2.0 turbo and later also going to test with some wideband sensor to see what data that brings.
So far still hesistant to drive this thing, affraid a wrong mixture will cause my engine damage.
Also still having a little hard starting, have to crank it for several seconds before it fires up.

Bram

[Legacy777]

Out of curiousity how old is the coolant temp sensor? When these things go, they typically don't give a code and just feed the ECU incorrect informaiton which will either cause them to run rich or lean. There is A LOT of additional fueling corrections and other stuff that are based upon the engine temp. So if it's off, it will cause you a lot of problems.

[trac]

Out of curiousity how old is the coolant temp sensor? When these things go, they typically don't give a code and just feed the ECU incorrect informaiton which will either cause them to run rich or lean. There is A LOT of additional fueling corrections and other stuff that are based upon the engine temp. So if it's off, it will cause you a lot of problems.

Hi josh, regarding that coolant sensor and select monitor read out, it shows perfect values constantly, never got any blinks or strange values when checking it in the select monitor.

I do can check that coolant sensor connections again. If I'm correct, there's a thermistor (3 wires) and a temo. sensor (2 wires). The temp. sensor is also used for the dash read out and I think for the coolant temp. value in the select monitor. Why would that thermistor be there. The connections in the diagrams aren't completely clear in the manual.

I can also test the car myself with the CO meter to see what mixture I'm getting while idling and revving.

Bram

[Legacy777]

There are two temperature sensors. The brown connector sensor is a two-wire sensor that the ECU uses. There is also a one-wire sensor that is used to display the temperature on the dash.

What temperature readings are you seeing at normal operating temperature?

[trac]

There are two temperature sensors. The brown connector sensor is a two-wire sensor that the ECU uses. There is also a one-wire sensor that is used to display the temperature on the dash.

What temperature readings are you seeing at normal operating temperature?

Upon idling and some revving without the car driving normal temperatures between say 83 and 93C. Fan coming on nicely when reaching approx. 93C.
Looked at FSM ch. 2.7C, page 77 and this is what it should be. I can do the test described there, but I think the sensor is working fine as it is.

As we're on this subject, maybe you can shine some light and take a small sidestep how this temp.gauge is or should be wired.
This because it seems like there is some confusing info in the FSM how the temp.gauge should be wired.

As said, on the thermostat housing there are two sensors. A one pin sensor and a two pin sensor.
See following pics, seems like different ways of hooking up the temp. gauge have been used in the FSM.

ECU wiring with two pin water temp sensor with spliced (RedGreen/RG) wire going to temp. gauge (chapter 6.3 page 28.


A little different on chapter 6.2 page 60


Or direct from one pin sensor (thermistor), WhiteGreen/WG wire. Connections shown on chapter 6.2 page 58.


Connections dash shown on chapter 6.3 page 68.


And a combination as shown on chapter 6.3 page 69.
Temp. gauge diagram showing WG wire going to the gauge and RY wire coming back to ECU.


All the wires are in my car, even the spliced RG one, but the WG wire is connected to my temp.gauge now.

What do you think?

Bram

[Legacy777]

Your low end temperature is a little on the low side. What was the ambient temperature when you took those readings?

Can you take a reading of the coolant temp sensor via the select monitor when the engine is "cold" before you start it up after it's been sitting overnight. If you put the key in the "ON" position and don't start the car, you will be able to still connect to the ECU and review the ECU values. Compare the ECU water temperature value against a good known ambient temp where the car has been sitting.

Regarding the wiring for the dash temperature gauge. I believe the extra wire coming from the 2-wire temp sensor to the dash temp sensor wire is an ECU reference ground. Two wire temperature sensors use a +5v source connected back to a reference ground in the ECU. This is a separate ground that is not tied together to your typically "chassis" ground.

The dash temp sensor appears to be using a wheatstone bridge circuit to calculate the correct temperature. However, one side is referenced to the ECU reference ground, while the other is referenced to the chassis ground through the sensor body. I really don't know why it's done this way....but it is.

Here's a little background on wheatstone bridge circuits
http://en.wikipedia.org/wiki/Wheatstone_bridge

[trac]

Thanks for your answer and explanation.
But I still don't exactly know now how my temp. sensors and gauge should be wired now.
Looking at some of the diagrams, you'd only need one sensor then. But why there's two then?
From the above diagrams, which one should be correct? This is basicly the only uncertainty I have in my wiring.

From what I remember, I think the spliced RG wire coming from the temp.sensor is not connected. So these two wires only go directly to the ECU.
And the WG wire coming from the thermistor is connected to my gauge now. So no connection between ECU and gauge if I remember correctly.

Bram

[Legacy777]

Did you alter how the coolant temp sensor(s) were setup when you swapped the engine? If not, I wouldn't really worry about it. There are two sensors, because that's how the factory set it up. The two wire goes to the ECU. The extra wire going to the stock gauge is done somewhere in the wiring, and really has nothing to do with the ECU coolant temp sensor. It was just a convenient place to get an ECU reference ground.

Maybe I'm missing something or misinterpretting how you have things setup.

[trac]

Oh well, this is indeed a project car with an installed EJ20G.
All is/was well, never had any fault codes or funky behaviour, it's still that some of these small issues like lambda value I'd lilke to get working as the manual describes.
So far I found nothing in the wiring job I did, and so far can't find any clues the ECU is missing an input or something...

I remembered when checking the way the temp.gauge is wired, I was a little confused to find different info in the FSM. So I connected it what seems the most logical way, and all seems to be working as intented.

How would missing that reference ground be noticable in the ECU behaviour?

bram