spike747

I want to document a problem I was having and what fixed it. It was a strange electrical problem that took me a long time to diagnose (didn’t have the money to just throw parts at it) and want to pass on what I learned so someone else doesn't have to go through the same experience...

I have a problem with my 2001 Subaru Legacy 2L B4 EJ20 auto which I just bought from an auction. Engine light came on the way home from the sale (Grrr!)...
P0328 KNOCK SNSR 2 CKT HIGH
P1593 TCM COMMUNICATION CRT
P0136 O2S CKT (B1S2)
Mechanic cleared codes, but the P0328 kept coming back. Replaced knock sensor and the codes/engine-light went away, however problems persisted...

When cold, I get sudden bad power loss when accelerating out of the first few corners. It lasts 2-3 seconds then recovers and continues to accelerate like normal. It feels like I get poor acceleration sometimes when hot as well.

Cam belt, water pump, thermostat, plugs, auto fluid/filter, oil, oil filter, air filter ...all done. Checked plug leads, air leaks, fuel pressure, compression ...all fine. LTFT+STFT -3% at idle and about the same at steady 2000rpm. Car's done 180,000kms. O2 sensors are original.

Rather than go on and on about what I did. This is the crux of it...

The rear o2 sensor was intermittently spiking high voltages back to the ECU (5.115 volts). When this happens several sensor readings go crazy... knock sensor (the actual voltage at the sensor changes from the normal 2.5vDC to over 3vDC), temp sensor drops 5degC, The OCV Values drop to half and near 0, and the throttle angle goes up almost 2 degrees (when you aren't even touching it).

I was eventually able to reproduce it “on command” after driving the car hard (lots of bursts at near WOT when the car was already hot) and got these codes to appear in the ECU...
P0328 Knock Sensor 1 Circuit Malfunction (High input)
P1307 Oil Flow Control Solenoid Value (OCV) 1 Signal Circuit Malfunction (Short)
P1309 Oil Flow Control Solenoid Value (OCV) 2 Signal Circuit Malfunction (Short)
I noticed from the logs that the rear o2 sensor was going to 5.115 volts at the same time the engine light was coming on. I unplugged the rear o2 sensor and tried to reproduce it... I couldn't. Plugged the rear o2 sensor back in, then immediately got the ECU errors back again.

Put a new rear o2 sensor in ($423!! NZD) and now the car runs so much better. Hot and cold has been improved remarkably. The car feels much smoother and responsive now. I still get a very quick stumble on the first corner when cold, but it is so insignificant to the problems beforehand. I think replacing the front o2 sensor might clean that last bit up.

Anyway, I hope this information can help someone someday.

PS. If you need any more info… just ask.

spike747

A week later and car is still running better when warm/hot (smoother, more consistent, and more torque in the 2-3000rpm range) with the new rear (post cat) o2 sensor, although I should mention that I can still get significant power loss on the first few corners from a cold start, if I push on the accelerator more than just a little.

On average, the new rear o2 sensor reads rich, while the old front o2 sensor and AFR reads lean… 0.588v, 1.625v, and 14.93 (on a similar-ish test using the old rear o2 sensor, the averages were.... 0.469v, 1.540v, and 14.7). From monitoring a consistent idle I can see that approx. 1.545v on the front o2 sensor results in a calculated AFR of 14.7 (data collected using the ecuExplorer program). I think the ECU is compromising the AFR ratio to appease the new rear o2 sensor. A new front o2 sensor will probably bring all the numbers back into line; however I’m worried that it will do so by removing more fuel from the mix. My current thoughts are that the stumble/loss-of-power problem, when cold, is due to the fuel ratio being too lean. The problem is significantly improved when both o2 sensors are disconnected (idle from cold is MUCH better as well). When the o2’s are disconnected, the injector pulse time is longer in the same situations.

I’m not sure what I should do next (if anything). The stumble problem is reasonably minor and a new front o2 sensor may make it worse. Does anyone have any constructive suggestions?