I just replaced the lambda on my 98 Mazda 626 GF 2L because it was showing a signal that was constantly pegged lean and I suspected it was defective.

After replacing the sensor I took the car for a 15 minute spin which I logged. One thing of interest to me was the relation between O2 voltage and STFT ( I logged this at the end of the drive ):

enter image description here

I was reading a page about diagnosing lambda operation and a few questions came up in my mind.

The first question was due to this sentence:

6) The O2 sensor must cycle at least once per second, which would show 3 cross counts on the scan tool PID.

It also says:

The O2 sensor not only has to cycle, it also has to cycle fast enough (proper frequency) and wide enough (proper amplitude). At least one cycle per second ( 1 Hz ) must be seen at the signal wire in order for the O2 to be considered good (not lazy). A one cycle per second will make the scope trace go across the 0.450 volts mark approximately 3 times, which the ECM recognizes as 3 cross counts.

As you can see on this zoomed in graph the sensor is only crossing 0.45 volts about once every three seconds instead of once per second.

enter image description here

The STFT is also spending about two seconds in negative territory for every one second in positive territory.

Now according to the same site:

10) Contrary to what many people think, an O2 sensor WILL NOT cycle by itself. The O2 sensor cycle is a direct result of the ECM response to the changes in the mixture.

So the question arises, why is my O2 signal cycling so much slower than what this site says is normal, and why is the STFT spending twice the time negative as positive? This is in a car suffering from a rough idle, with occasional stumbles during idle, and an initial lack of power on acceleration.

I can think of a few possibilities:

  • This is just the normal algorithm for this car's ECU
  • A defective Lambda
  • A defective ECU not responding as quickly as it should
  • Air / Fuel mixture not getting completely burnt due to insufficient air

I'm kind of leaning towards the last possibility since I noticed that the lambda is ranging between .18 to .89 volts with it staying mostly above .20 volts and the fact that the STFT is going negative twice as much as it's going positive.

I can think of a few possible causes:

  • Injector's sticking open or excess fuel pressure
  • An intake valve not opening sufficiently
  • MAF biased slightly high making ECU think there is more air than there actually is

I measured cam to lifter clearances. Spec is 0.23mm to 0.30mm

| Cyl #   |  1  |  2  |  3  |  4  |
| Intake  | .24 | .29 | .24 | .24 |
| Exhaust | .29 | .24 | .29 | .24 |

They're all within spec, but maybe the wear on I2 relative to the other intake valve lifter shims is causing the problem?

Anyways, I'd like to hear what all y'all think.

  • You show the STFT and O2 traces, but make no mention of the LTFT. What is its value?
    – Zaid
    Commented Dec 19, 2015 at 19:17
  • If I could trouble you with a request for the O2 sensor 2, bank 1 output as well please...
    – Zaid
    Commented Dec 19, 2015 at 19:20
  • @Zaid This is a European model and only has one O2 sensor. LTFT is zero, but that maybe because I've only driven the car once since changing the sensor. Commented Dec 19, 2015 at 19:22
  • So just to confirm, there is no O2 sensor after the catalytic converter?
    – Zaid
    Commented Dec 19, 2015 at 19:23
  • @Zaid Nope. Visually confirmed. European cars didn't become fully OBDII till around 2001 I think. Commented Dec 19, 2015 at 19:28

2 Answers 2


This is a normal fuel control pattern especially for idle. O2 switch patterns are not specified to look a certain way but vary greatly. They are the result of a competing set of requirements including; combustion efficiency, catalyst requirements, driver demand and regulatory requirements.

The hold lean strategy seen here is a way to hold the mixture slightly lean (Lambda 1.001). It is often used at idle because it betters mileage a bit. It can be used at idle because engine NOx output is low. NOx is poorly reduced by a three way catalyst in lean mixtures. A similar rich mixture strategy where the mixture is slightly kept rich (lambda .99), high 02 voltage for a longer time than low voltage time, is used less frequently. It provides some power and catalyst strategy help at the expense of fuel mileage.

Test at 2500 rpm, the pattern is more likely, but not always like your source described. Testing 02 sensors with scan data is not conclusive due to PCM manipulation of the signal. An oscilloscope and gas analyzer are required to definitively test this sensor. Take care when generalizing fuel control strategy and what is found describing how it is done in publicly available sources. Much of what is found is inaccurate or not detailed enough.There is also some industrial subterfuge that goes on between vehicle manufactures in this area of technology.

The state of the art has advanced far beyond this, and the games between the vehicle makers and regulators continue.

  • I assume you're referring to the emissions thing with VW? Commented Dec 20, 2015 at 20:24
  • 2
    Yes and... It has been a 25 year battle between the CA air resources board and the design engineers. The OEM's find loopholes and "misinterpret" the regs and then the Board tightens the regs and the cycle continues. The regs bear little resemblance to the 1996 version. This is one of the reasons much of what is out there on amateur sites is incorrect. Someone posts what they learn about a particular vehicle and post it as if all are designed that way, when in reality the answer to how does this work is "it depends". Commented Dec 20, 2015 at 21:45

Your lambda control looks absolutely fine.

And consequently I don't think it is the root cause for your rough idle. Here's my reasoning:

  1. The web page linked in your question says

    The O2 sensor must cycle at least once per second

    I wouldn't take this as a hard limit for a few reasons:

    • the Bosch Fuel Injection & Engine Management book indicates a ballpark figure of 1-2 seconds at idle

      At idle, the cycle from lean to rich and back again may take 1 to 2 seconds. At cruising seed, the cycle may happen several times a second.

    • lambda sensors degrade with use. They would have to be changed far more frequently than is the case if a "slow lambda response" code was set by the engine computer on the basis of a hard 1-second limit.

      Have a look at this ScannerDanner case study to see an example of "lazy" vs. regular lambda sensor output.

  2. LTFT and STFT are hovering close to 0%

    If the LTFT has already stabilized after replacing the lambda sensor, this is good news as it indicates your car's air-fuel metering is measuring at the exhaust what it believes the AFR should be.

    Don't expect them to be perfectly zero. They won't be.

  3. Narrowband lambda sensors are based on qualitative sensing of rich/lean conditions

    In other words, I'm not bothered by the fact that the sensor outputs 0.89 V on the rich side, but only 0.2 V rather than the theoretical 0.1 V on the lean side. To the fuel injection management control system, 0.1 V = 0.2 V = lean.

    Now if this was a wideband sensor the story would be quite different, but it isn't.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .