O2 Sensor measurements, experiments etc...
Posted: Sun Oct 25, 2009 9:42 pm
I thought i would post this stuff up here, some people might find it interesting:
(1991 completely standard NA6)
I had a wide band sensor and an LC1 controller sitting idle in my cupboard from a previous project, so i thought instead of gathering dust i would do something with it.
Fortunately it was the same thread as the OEM sensor, so i did a switcheroo:
Then a quick drive down the block to see if it works: (I ended up tagging the high side of the relay for the controller off IG2 instead of IG1 so it gets cut during crank, ill have to fix that)
Without an O2 sensor on steady state cruise I guessed the ECU would be in some degraded mode and will revert to a conservative tune. From driving i didn't notice much difference, i figure the original O2 sensor has probably never been changed and at 160k its getting pretty old. One thing that was somewhat noticeable is that in 5th at 100kph the car did have abit of pickup if you gave the accelerator a gentle squeeze.
Had to go for a lengthy highway drive so i took some data:
Its probably hovering a bit from minute throttle changes, as the car sits on 100kph.
The next thing i did was program one of the sensor controller analog outputs to emulate the OEM sensor, and fed that back into the ECU:
Now on highway cruise you can clearly see the difference, the average cruising AFR is much higher, and if you zoom in on the data you can see the AFR oscillate as the ECU jumps between rich lean based on the feedback from the sensor.
Have only done 120km like this will be interesting to see how many k i get out of this tank. (BP ultimate for all experiments)
So now you could say i have used $300 piece of equipment and a weekend to achieve exactly what the OEM ECU was doing perfectly fine before I started.
I'm interested to know whether this is leaner or richer than what the standard configuration would track. Obviously it can be adjusted either way by changing the lookup table for the analog output.
Its fractionally hesitant at 100kph now... and one thing that was defiantly noticeable, after a long trip on the highway I had to stop and idle. It was rough and possibly misfiring a little, but seemed to get better after a bit of stop - start typically crummy Melbourne urban speeds.
I haven't pulled or tamed the O2 feedback yet to conclusively prove that the lean mixtures are causing this, it could just be a coincidence. But I'm thinking possibly the lean mix for an extended period of time takes its toll on the plugs? I need to pull a plug out and have a look, perhaps i can change a heat setting or switch to iridium or some other sort to improve?
The next thing i wanted to do was try adjusting the clock spring on the AFR a few clicks at a time to lean out the rest of the map, the AFR has a calibration sticker dating to 2000 and I'm sure they would have set it on the conservative side
Ill probably need to log a few other signals to help correlate the data.
Just to conclude I want to add that of course I understand the potential consequences of running an engine too lean. I am trying to be careful and will make small changes at a time.
Before i fit the MegaSquirt (also a survivor from a previous project) I am trying to do some hands on learning about how far you can take a tune by tinkering with an OEM setup. My last EFI project was a carburetor conversion, so i didn't quite have the availability to benchmark that i do now.
If the worst does happen, then i have successfully determined what "too far" is.
(1991 completely standard NA6)
I had a wide band sensor and an LC1 controller sitting idle in my cupboard from a previous project, so i thought instead of gathering dust i would do something with it.
Fortunately it was the same thread as the OEM sensor, so i did a switcheroo:
Then a quick drive down the block to see if it works: (I ended up tagging the high side of the relay for the controller off IG2 instead of IG1 so it gets cut during crank, ill have to fix that)
Without an O2 sensor on steady state cruise I guessed the ECU would be in some degraded mode and will revert to a conservative tune. From driving i didn't notice much difference, i figure the original O2 sensor has probably never been changed and at 160k its getting pretty old. One thing that was somewhat noticeable is that in 5th at 100kph the car did have abit of pickup if you gave the accelerator a gentle squeeze.
Had to go for a lengthy highway drive so i took some data:
Its probably hovering a bit from minute throttle changes, as the car sits on 100kph.
The next thing i did was program one of the sensor controller analog outputs to emulate the OEM sensor, and fed that back into the ECU:
Now on highway cruise you can clearly see the difference, the average cruising AFR is much higher, and if you zoom in on the data you can see the AFR oscillate as the ECU jumps between rich lean based on the feedback from the sensor.
Have only done 120km like this will be interesting to see how many k i get out of this tank. (BP ultimate for all experiments)
So now you could say i have used $300 piece of equipment and a weekend to achieve exactly what the OEM ECU was doing perfectly fine before I started.
I'm interested to know whether this is leaner or richer than what the standard configuration would track. Obviously it can be adjusted either way by changing the lookup table for the analog output.
Its fractionally hesitant at 100kph now... and one thing that was defiantly noticeable, after a long trip on the highway I had to stop and idle. It was rough and possibly misfiring a little, but seemed to get better after a bit of stop - start typically crummy Melbourne urban speeds.
I haven't pulled or tamed the O2 feedback yet to conclusively prove that the lean mixtures are causing this, it could just be a coincidence. But I'm thinking possibly the lean mix for an extended period of time takes its toll on the plugs? I need to pull a plug out and have a look, perhaps i can change a heat setting or switch to iridium or some other sort to improve?
The next thing i wanted to do was try adjusting the clock spring on the AFR a few clicks at a time to lean out the rest of the map, the AFR has a calibration sticker dating to 2000 and I'm sure they would have set it on the conservative side
Ill probably need to log a few other signals to help correlate the data.
Just to conclude I want to add that of course I understand the potential consequences of running an engine too lean. I am trying to be careful and will make small changes at a time.
Before i fit the MegaSquirt (also a survivor from a previous project) I am trying to do some hands on learning about how far you can take a tune by tinkering with an OEM setup. My last EFI project was a carburetor conversion, so i didn't quite have the availability to benchmark that i do now.
If the worst does happen, then i have successfully determined what "too far" is.