lambda sensor status

[leopold]

Hello,
who knows and can tell me which parameters have to be fullfilled that ecu says lambda sensor status ok?

[rubenste...@gmail.com]

Hello Leopold.

I don’t know for sure, this is only what I have observed, but I believe 2 conditions have to be met.

With the ignition key on and engine NOT running (KOEO), if the ECU sees the bias voltage of 0.45V on the lambda sensor’s sensor circuit, the status is One.    Then, with the key on and engine running, if that bias voltage falls, or starts to move, within a certain time, the status remains at One, otherwise, if it stays at 0.45V, the status drops to Zero.    By ‘a certain time’, I don’t think the measure is in time, as such, because when I looked at this a few years ago, I noticed that it could drop to zero (when I introduced a fault in the sensor circuit) after, let’s say, 30 seconds, and yet, when the circuit was working correctly, sometimes it might take, say, 45 seconds for the bias voltage to start to drop, but the status still remained at One.     And it wouldn’t make sense to use time as the measure: it takes much longer for the exhaust gases to heat the sensor if the engine is left at idle than if the car is driven off soon after start up.     (I know the lambda heater helps, but the hot gases make a big difference to speed of warm-up of the sensor.)

So my guess is that rather than using time as the measure, the ECU counts the number of engine revolutions after the engine starts, and if the bias voltage has not moved from 0.45V by that number of rotations, the status then drops to Zero.

But that’s all supposition.     I never got round to it, but I had thought I could test that theory with the PicoScope by counting the number of engine revolutions it takes for the status to drop to Zero with an induced fault under different rpm eg leaving the engine at idle compared with, say, 1500 and 2000 rpm.     

As I say, that’s just based on my observations; I haven’t seen anything in writing.

[rubenste...@gmail.com]

I forgot to mention: in my post above, I was referring to MEMS 1.6, as fitted to my 1996 Rover 100.    I’ve no experience with other MEMS variants; Leopold has since told me, on MEMS 1.3 there is a coolant temperature threshold of 88C before the fuelling can go into closed loop.

[Andrew Jackson]

Recently fitted a new lambda sensor (bosch) and coolant sensor. Engine is warm (93C) and at idle the lambda readings appear to oscillate as expected.
However, the ECU has a lambda status of 0

Does this mean the ECU will never go into closed loop?
What would cause the ECU to set a status of 0 when the readings appear to be good? 
See attached CSV file.

[rubenste...@gmail.com]

Hi Andrew,,

Remind us: which car is this on?

Dean in the Shetlands has just had a problem which turned out to be a faulty ECU.     Whilst a DVM showed that his 02 sensor was switching correctly, the lambda voltage reported by the ECU (via PSCAN) stayed at the bias voltage (0.49V - MEMS 2J).        You, on the other hand, can certainly be confident that the lambda-voltage-monitoring circuit of the ECU is good because your csv file shows the sensor switching correctly within the normal voltage range.     Your coolant temp is also being read and, at 93C, is clearly a true value and not a substituted, 60C limp-mode value.      The inlet air temperature is also a realistic value, quite compatible with the coolant temp.

Your ambient air temp reads as 200, but I presume that’s merely because like my car, there is no such sensor fitted, and is therefore normal and irrelevant.

One question I have for you, and it’s not something I have previously paid any attention to, concerns air-fuel ratio.  (I usually use Al Richey’s memslogger and MEMSAnalyser, and, if it has an air-fuel ratio data output, it’s not one I normally look at.)   Yours reads 14.6 in the csv file.     Now that strikes me as the in-practice stoichiometric value for CLOSED LOOP running.     My gut instinct is to expect to see 13.6 or similar for OPEN LOOP running, and 14.6 for CLOSED LOOP.       Am I correct?     If so, I’m wondering why, if your car is stuck in OPEN LOOP, is there an air-fuel ratio data output more akin to running in CLOSED LOOP? 

I think I will have to try this on my car to see what is normal, but, when you start up and, for the first 30 secs or so, when you are definitely in OPEN LOOP, do you see 13.6, changing to 14.6 after a minute or so?

I expect you can probably explain if I’m mistaken on this.

Martin

[rubenste...@gmail.com]

One thing I might try, given that this happened when you changed the 02 sensor, would be to refit the old one and try it.    But, given the ECU is reading the new one, I’d be surprised if the car then behaved normally.

Similarly, if there was a fault in the heater circuit, and the sensor wasn’t warming up quickly enough, I would have expected the ECU not to report an oscillating 02 voltage as your is correctly doing, but, instead, to output flat bias voltage (eg 0.45V etc) as the reading.   So this all suggests your new sensor is good.

It is the correct Bosch sensor for your vehicle?

[Andrew Jackson]

Martin,

It's a Bosch Lambda Sensor 0258003193 fitted on a '93 Cooper SPI with ~87k miles.
It's plain weird - in the log file the ratio moves between 13:9 and 14:6.
I've attached a log file that I captured just after the previous one, where I idle and then hold the throttle open at 2000rpm and back to idle. AFR remains as 14:6.

[rubenste...@gmail.com]

Andrew

I checked my car: I get 14.6 as the AFR regardless of closed or open loop, so on my car it's pointless looking at it.   And I'll assume the same for yours.

However, I ran your MemsFCR so I could perhaps correlate your readings.

My Long Term Fuel Trim is -10, looking at the gauge on MemsFCR.       But your csv says 25 for LTFT.      I take it that is 25%.        If that's right - ie your 25 in the csv equates to +25% - that suggests that your car is tending towards running so lean that MEMS has had to enrich the mixture to the point that it max'd out at +25%, at which point it in limp mode.

Does that sound right?

 To test this, I would go to the adjustments (adaptive values in MemsFCR,) and wind back the LTFT to, say, zero, and run the car.       If there's a massive fault eg not enough fuel reaching the injector or massive air leak you would expect to see the short term fuel trim, once it goes into closed loop, going massively positive and LTFT going up to try and bring  STFT down to zero.       So you'd expect to get into closed loop after a minute or so, then the fuelling should keep enriching it until it hits the buffers at eg 25% at which point it should go into open loop, having realised it cannot bring the fuelling under control.

But strange it should seem there's a massive fuelling fault that appears only after you fit a new 02 sensor.

I think your LTFT of 25 may be the clue to the problem.

What do you think?

[rubenste...@gmail.com]

Andrew,

I presume the car drives perfectly well, otherwise you would have said so.   In which case, it’s safe to say you don’t have a massive air leak or a lack of sufficient fuel getting to the injectors.    So that suggests the car itself is not faulty, but there’s incorrect info being fed to MEMS or perhaps the correct data is being fed to it but MEMS is dealing with it incorrectly?

What do you think?

[Andrew Jackson]

Martin,

Interesting, I'll adjust the settings and see what happens. Also you mention the ECU dealing with it correctly, which leads me to starting to suspect the ECU.
I've attached a couple of logfiles, the first one a run from cold. Status starts as 1 as expected then goes to 0, but then goes back to 1 when I stop the engine (at the end of the run).  The logfile also show some corrupt data which I do get from the ECU from time to time.

Then the second run is when the car is nearly warm, I started the car and the lambda status stays a1, ECU goes into closed loop. The CTS reaches around 82C and then starts to go down to 78C!?

[rubenste...@gmail.com]

Thses csv files remind me why I love MEMSAnalyser so much!     A picture is worth a lot more than a thousand words!   Whilst I know nothing about coding etc, I know it would be totally impractical to convert the csv file into a form that could be displayed in MEMSAnalyser.     Whinge over,  here’s some thoughts for you:

I believe it’s perfectly normal for the status to start at one, if I remember correctly.   It’s as though MEMS is saying “innocent until proven guilty”.    When it goes to zero, I believe that’s MEMS saying ‘guilty’ ie it gave you the benefit of the doubt but it could not go into closed loop.    At switch off, it resets itself to try again next time.
 

82C dropping to 78C - I’ll bet that’s your thermostat starting to open and letting in cold water.  

Second run it goes into closed loop when the engine’s hot.    This is really a good clue: MEMS looks at the lambda voltage after startup and, if it doesn’t start to move within a certain time (I suspect it’s rotations rather than time, say 30-50 seconds) it decides there’s a fault and stays in open loop, and it stays there even if the lambda voltage starts to move later.    It had its chance but it’s too late.    So if the lambda heater doesn’t work and the engine’s cold you’ve got no chance of going closed loop unless you drive off immediately under full load, and maybe not even then.

But even if the heater is broken, and you get the lambda sensor hot with a hot engine, if you then turn the engine off and then start up immediately, the lambda voltage will start switching before that cut-off time and the car will go into closed loop.    That would explain why you got into closed loop on a hot engine.

Even if that is the answer - the 02 sensor heater circuit is faulty, I still believe that +25% LTFT still needs investigating.     25% is the maximum reading ie it represents the limp-mode, I-give-up, point.

So, at the moment, a fault in the lambda heater circuit would explain why you can get into closed loop on a hot engine if you quickly restart but not on a cold engine.    I would be sorely tempted to refit that old 02 sensor (unless you know it was faulty, or you had the same fault then).     If you did fit the new sensor because you could not get into closed loop with the old sensor, then do check the heater circuit… power and ground into the heater relay, and power out of the relay.    (I have a breakout lead for the 02 sensor - on loan to Dean at present - that would allow you to test the heater voltage at the sensor under load.)

Does that seem reasonable?