Oxygen Sensor Failure

09/18/2005: Problem.

While taxiing out the other day, I noticed what I thought was a substantial drop in RPM for no good reason. Every few minutes I would lose about 150-200 RPM, then the engine would slowly come back up to where it was.

Back at the hangar, I hooked up my scantool to a PC and ran the engine for several minutes, hoping to capture whatever events that would cause this RPM drop. Fortunately, the problem appears to be fairly regular and repeatable.

During the first half of this data capture, I was adjusting the throttle setting a bit, so there isn’t a good reference for RPM. However, near the end, I made sure the throttle was set to 1500 engine RPM, which resulted in about 15″ MAP. When I was convinced I saw the RPM drop, then ramp back up to 1500, I shut the engine down so I could see what happened.

Here is a plot of the normalized data:

Plot of bad O2 data
Plot of bad O2 data


The purple sawtooth pattern is the O2 sensor voltage.

The blue square wave pattern is the short term fuel trim.

The yellow line indicates engine RPM.

09/25/2005: Solution?

This morning I replaced the O2 sensor with a new one. The Bosch part number of the replacement is 13077 and cost was $59 including tax. I checked with every major auto parts chain in town, plus several online retailers and could not find the part much cheaper.

I did another engine run and did not see the same RPM drop as before, so I flew around the pattern and landed. Of course, I captured the data!

Here is a plot of the normalized data:

New Sensor Data Plot
New Sensor Data Plot


The purple line is the O2 sensor voltage.

The blue line is the short term fuel trim.

The yellow line indicates engine RPM.

Comments on this plot to corresponding to the number annotations:

(1) – Start up.

(2) – What the hell is this?

(3) – This was idling at about 1500 engine RPM outside my hangar.

(4) – This was during taxi to KGEU runway 1.

(5) – Swing around, pull to idle, runup check, pull to idle.

(6) – Advance to the hold short line and get clearance. Yes, I waited THAT
long until I could go. It was busy in the pattern today.

(7) – Position and hold, then clear for takeoff. Full power and prop set to 2700. Peak engine RPM was 4900. Interesting fuel trim setting; did it go open loop?

(8) – Top of pattern, set prop to 2200, pull throttle back to keep distance behind a Cessna dragging it around in a 747 pattern at 80 knots.

(9) – Pull to idle over the threshold for landing.

(10) – Taxi back to the hangar and shut down.

Other interesting things in the captured data –

  • The timing advance was always -64.
  • The absolute throttle position was always 0.
  • Have yet to figure out what that means.
  • Also, the ‘vehicle speed’ parameter seemed to follow the RPM, so I wonder how they calculated it. Obviously there is no vehicle speed sensor hooked to the ECU.