Bob,
Why did you replace that O2 sensor? Have you considered taking the new one back under warranty (if there is any)? Or perhaps trying a different brand O2 sensor? What brand was that new one? Was it OEM with the OEM plug or a universal O2 sensor? I ask these questions wondering if it might not be the new O2 sensor that is off, or even more interesting that they are, or were both off, biased on different sides of the .45 V mark!
Answer: (In retrospect) STUPIDITY. It had over 80,000mi and most sources recommend replacement at 80k. Further, about every month or two it would throw a MIL for "slow response" (if I idled in a traffic jam for more than 5 minutes, but never "on the road") which is indicating that it's getting toward the end of its life. It didn't cost me anything to reset the code, but I found it annoying, never guessing that replacing it would cost me that 25% in fuel economy.
The one I put in was, I think, a Bosch, and it was an "OEM type" having the proper plug to plug right into the harness.
For the Record my jeeps are Renix and there are many of us with the Renix system also interested in this thread. Perhaps we should just refer to our comments as Renix, and Post-Renix, O2 tricks for treats! LOL.
Speaking to an earlier thread, I am wondering where the information came from that says that "Most ECUs have an internal voltage divider that will set a nominal voltage if the O2 sensor is not connected or goes open". I am wondering why they would bother.
That is from a Bosch book on fuel injection I used when I was regularly servicing BMWs that used Bosch designed FI. That circuit supplies that voltage (which is overridden by the actual O2 output) to enable the vehicle to be driven if the O2 fails completely both as a "limp-home" for the driver, PLUS as a safety device for the catalytic converter, so the computer would not go too rich w/o an O2 and harm the cat. The BMW 528 model regularly ran lean w/ the O2 connected, and would hesitate or surge under slight throttle. Most of us simply disconnected it (nice green wire with an inline plug right on the firewall screaming "disconnect me!" and got MUCH better performance, and better fuel economy. Running lean actually caused drivers to add throttle, which caused MORE fuel usage. I could get 33mpg disconnected with great throttle response, but only 30mpg and poorer throttle response connected. NO MIL light to worry about in those 1980's e28 circuit!
From what I have read if the O2 sensor input is not acceptable, the ECU ignores it and uses the data from the other sensors plus a map table (burned into the memory) of empirical dynomoter test data to control the A/F ratio which is referred to as open loop, but is in fact another closed loop that simply runs rich all the time!
Maybe true now with the JTEC PCU (which I have not studied). Technically, any operation w/o feedback (as from the O2 sensor) is OPEN loop. The o2 is the only sensor that provides feedback, and is the "Master Sensor" if you will.
Also I am quite sure that the O2 sensor data input to the ECU must oscillate back and forth at about 1 second intervals across the magic 0.45 V (post Renix) mark or the ECU will decide to ignor the O2 sensor data. In other words the system always has over shoot and undershoot on the 0.45 V set point and the computer expects that, so it better see it.
What concerns me is that the slope on the real O2 sensor output (narrow band) is so steep that anything more that a 0.05 V bias addition or subtraction may throw the system into a range where it can not measure the true O2 concentration anymore, that it will only know that it is over range (high or low) and the signal gets stuck way off in the high or low range and then the bias circuit "trick" does not ever see 0.45 Volts again which it needs to add to the 0.05 to get the fake signal value. To make matters worse, if there is a small variation between O2 sensors, then they could already be slightly biased one way or the other from sensor to sensor increasing the risk of getting beyond the O2 sensors measurement limits.
Yep, that is a possibility that only experimentation will show. Make no mistake, the PUC (or ECU in Renix) will TRY to drive the O2 input back to 0.45v, which is actually what we WANT, because that will mean that our O2 sensor WILL be at a greater voltage (sum of actual sensor output
PLUS the bias voltage we are adding to it).
As to whether it can do it in a way that will satisfy the computer's expectations with regard to slew rate (that 1 oscillation per second) is something I won't know until we try.
I DO have a dual trace oscilloscope packed somewhere away (since we moved to TX lots of my shop tools are in a storage unit still), and if I ever can get organized in our new place I might actually FIND it and have time to experiment.
Am I making this clear, what I am saying about the steep slope, error problem?
Perfectly, and I agree 100%. The only question is whether it would defeat us, or restrict how much we could fool the computer, and if that would be enough to make a measurable difference in fuel consumption.
Also I read and quote from below "
The .450 reference voltage that can be measured on the O2 output circuit, with the ignition on, engine not running, is used by the ECU to determine if the O2 sensor is operating", and I ask where this came from? as it makes no sense to me?? If the power is on it should also be heating the O2 sensor thus turning the O2 sensor on in less than 20 seconds such that it generates the volatge itself. Also being lean, engine off, and plenty of excess air, a working Post Renix O2 sensor should be biased to the lean side in about 20 seconds and should not read 0.45 Volts?
I don't know where that came from. I didn't write it. I HAVE read methods to "test" an O2 sensor w/ a propane torch (off the vehicle), or on the vehicle by introducing propane into the intake (to make a temporary rich mixture). Of course, the only reason to have the ignition on is to heat the sensor, though the heater has no effect on operation if the sensor is hot (600 deg F or above) from combustion gasses (though the computer looks for the heater draw and will throw a code if it doesn't see it).
Now with all that said, I see no reason that a good working, highly accurate narrow band O2 sensor can not be biased slightly. Also I thing a high impedance analog volt meter could be rigged up and used on the dash to monitor the actual O2 sensor values to insure that the op-amp, potentiometer, or what ever magic black box is used to bias the voltage between the O2 sensor and the ECU. That would allow one to see the actual real time values while the ECU is being biased to lean out or enrichen the A/F mixture.
Thanks, that is the FIRST such comment I have gotten, and that's exactly what I propose.
Bob, one last thought, not to kill your efforts to build one of these gadgets (I am all for it), but to firshaget at what caused the 5 MPG mileage drop. When you changed the O2 sensor, is it possible that the sensor/harness wire connectors might need a little cleaning? If the connections or ground are not perfect, and adding just a few ohms of added resistance, that would lower the voltage signal to the ECU thus making it run richer, right?
