LAMBDA abnormalities

[sylvain oboeuf]

Hello, to begin a big thank you to the developers of these tools that allowed me to diagnose and solve a number of problems. I own a SPI British Open mini rover of 1995, equipped with a MEMS 1.6. All the components such as catalyst, oxygen sensor, vacuum detites, vapor receiver and hoses, air filter .... are new and functional. This vehicle works very well, it does not smoke does not consume too much. By cons I can not explain the problem that remains which is the lambda value which remains a little too high for the passage to the technical control .... I was able to bring out the lambda value and the pollution by reducing the fuel trim richness to -128 but that is not enough. Also during the diag with MEMS-scan, I realize that on the tab errors the LED "Lambda Signal error" comes on after a few minutes. After retrieving the logs and using MemsAnalyseur, it appears a run-time error '13': Type mismatch "I wonder if this is a problem of misinterpretation by MemsAnalyseur log or bad data that come back from the ECU? I placed a multimeter on the terminals of the oxygen sensor and I see the voltage change and evolve between 000mV and 700mV. Thank you for enlightening me and give me your opinion to guide me on some tracks that I will not explore. Sincerely Sylvain O.

[Martin Rubenstein]

“the lambda value which remains a little too high for the passage to the technical control ....”


This lambda value is a calculated value (if it is the same as the British emissions test). And a value of, say, 1.05, suggests it is 5% too lean. But, if your oxygen (and CO and HC) readings are all good, then it suggests your high lambda reading is because of a leak in the exhaust wgich pulls in fresh air (and oxygen). You could try starting your car when it is cold and put your hand over the exhaust pipe and look underneath to see if water is leaking from any if the joints.

“After retrieving the logs and using MemsAnalyseur, it appears a run-time error '13': Type mismatch "I wonder if this is a problem of misinterpretation by MemsAnalyseur log or bad data that come back from the ECU?”

That error message, I believe, comes when MEMSAnalyser cannot read the logfile. Are you using the latest MEMS-scan (0.5.2) to create the logfile? Do you have the latest MEMSAnalyser and the latest program to generate the logfile? In the MEMSAnalyser.zip package you’ll find MEMSLogger.bat. You can edit that bat file so it points to the correct COM port eg COM3, COM5 ... and you’ll also see where the lofile is placed ....Result.txt. You might have to edit that as well so that it is in the correct place. It is best to stay with Al Richey’s recommendation: C:\Spare\MEMSAnalyser\Logs\Result.txt.



If you use MEMSLogger to create the ligfile, you shouldn’t have a problem. But you need to have the ignition switched on and the cable connected to the correct COM port when you open MEMSLogger.bat. If the windows opens and then closes, you have made a mistake and you are not connected.

[Sebastián González]

Hello everyone, I have kind of a similar problem. I own a 1992 Mini 1.3 SPI that i recently rebuilt as a general maintenance (silicone hoses, new radiator, changing insulation, etc) The car runs fine, the mpg is good and in appearance nothing is faulty. But last month, when I went to pass the technical inspection that here, in Spain, we have to pass every year it didn't pass. They perform two CO tests in order to check the correct numbers of the engine. The first of both is at idle (engine was warm enough), and the results were OK; the second one is keeping the accelerator pushed around 3500 rpm. That was the one that failed; the usual result for CO when accelerating should be around 0.3 and it measured 1.3 (!?). This was when I discovered this fantastic tool, and the diagnostics that I performed didn't clarify at all if the Lambda sensor is malfunctioning at all. I captured some screenshots and a log file from cold to warm to check if someone has a similar issue (I've read the entire topic to clarify how the tools works and typical issues on the engine). On the other side, I'm not sure if some of the errors (coolant temp, intake temp) are real because in the rest of tests didn't appear again. Cheers!

[Martin Rubenstein]

Hi Sebastian

I’ve not looked at your logfile yet but I have one question to start with: is the catalytic converter original or is it an aftermarket one? It appears that aftermarket cats generally last 12-24 months. There’s a good reason why they are so much cheaper than OEM catalytic converters and it’s because they put less precious metal in, or even leave one of them eg platinum out.

So it’s possible that a person could have a perfect engine but the catalytic converter is not good enough. So always buy the manufacturer’s cat if possible. That’s just something to bear in mind. (I’m not saying that your cat should always be changed. In fact, the cat should be the last thing to be changed, especially if it’s an OEM cat, because if a good quality cat has failed, it’s generally because of an upstream fault which has poisoned it, and replacing it might only hide the symptoms for a few months,)

I hope to look at your logfile maybe tomorrow.

[Martin Rubenstein]

Sebastian

I think you should clear all the fault codes including the lambda signal error and start again.     I have noticed that you can get such errors stored if you sometimes disconnect a sensor and then reconnect.    So it's best to see which faults are genuine and return.

Does your battery need  replacing?    Less than 12V at idle and then jumping up to 14V at higher rpm.      It looks as if it's not acting as a buffer/reservoir for the alternator.

Are you permitted to take the car for a drive on the road?       It's better to look at some of the parameters under load.     To me, I think your lambda signal looks good, or let's say I can't see anything bad about it, but it would be better to see it under load.     Also, the sampling rate that MEMS gives us in MEMSAnalyser eg 2.33Hz, is very good but it's not like an oscilloscope, so it's not possible to measure the time taken to switch from low to high or high to low (ideally about 0.1 sec).    Nevertheless, we can get a good idea from the number of times per second that lambda changes and the waveform itself, especially under load when driving eg driving uphill or at say 60 mph, whether or not the oxygen sensor is getting old or lazy or has some other fault.    But from your logfiles, I think it's looking ok.  

[Sebastián González]

Hello Martin! Thank you for the fast answer! Let me explain some of the points:

• The catalytic converter is the OEM one, but it's the original that came with the car, so it's 26 years old. The car hasta an aftermarket exhaust from the end of the catalytic converter to the final exit. The exhaust manifold has been replaced when I did the rebuild (the flex joint was quite destroyed).
• On the second time that I logged data I cleared all the errors and the only one that remains (not fixed) is the one that says "abnormal lamdba value" when i press the accelerator and let it off very quickly.
• I guess, by what you say about the errors that could be produced when you disconnect sensors, that could be the first issue that I found when I connected the app to the car's ECU. Now I haven't seen again the error code.
• Last week when I was checking with the aplication I saw that the battery was at 11.9V so I connected the CTEK to charge it. I'ts quite new so I think that the battery is OK
• I have a month to perform the test again, and allowed to drive it, so I'll generate another log with the conditions that you are telling in order to check the numbers.

Thank you so much for the help! I'll report with news :)

[Sylvain O.]

 

Hello Martin and thank you for your return.

 

In France for a vehicle in clean-up mode (with catalytic converter) the CO and Lambda values ​​which must not be exceeded:

CO slowed down = 0.5%

Accelerated CO = 0.3%

0.97 ≤ lambda ≤ 1.03%

 

For lambda values ​​measured on my vehicle, it looks a lot like what Sebastian describes on his vehicle ..... but I did not keep the exact values.

 

The exhaust line was completely repaired a short time ago, reinforced manifold gasket at the cylinder head, decent manifold gasket and catalyst gasket. The exhaust RC40 with relaxation is new and the catalyst.

 

In fact I had already noticed a cold leak at the exit of the catalyst after changing it, drops of water escaped and I felt the air out and I tightened everything, no problems at this.

 

Regarding tools, I use MEMSAnalyser and mems-scan 0.5.2, I used the first version before. My configuration of MEMLogger.bat is good such that:

CD C: \ Spare \ MEMSAnalyser \ bin \

readmems com10 read-raw inf> C: \ Spare \ MEMSAnalyser \ logs \ Result.txt

 

But with MEMSAnalyser old version and new version, I have the following problem, I do not see the graph of the voltage lambda evolve above and below 0.43V, the value remains flat ??? I allow myself to transfer the log file "result old 1".

 

For accuracy, with mems-scan-0.5.x-Windows version 1 and 2, if I generate a log file, the log file is read by MEMSAnalyser except the Lambda data that causes run-time error "13".

 

Regards,

 

Sylvain

[Martin Rubenstein]

Hi Sylvian,

Worth checking the exhaust joints with a legth of plastic pipe (1.5m long?) with the engine idling, lay on the ground and moev the pipe along the exhaust concentrating on the joints.   You will soon know if you have successfully stopped the leaks!     By the way, you probably know this, but in case not, you should never use exhaust paste upstream of the catalytic converter.      In fact, having used this stuff myself and later looked at the joints, I really don't like it.   I have just fitted a new catalytic converter and the instructions that came with it stated that exhaust paste must never be use upstream, but it did say that silicone rubber could be used instead...   I had thought silicone was also bad for catalytic onverters (and oxygen sensors).    So I used some high temperature silicone rubber on the gaskets but a very thin layer and left a few mm without silicone on the inside edges so it wouldn't squeeze out onto the inside.when the joint was tightenend, and I used a sanding tool the clean off all rust on the exhaust joints and to try and get them flat.

Now, your logfile.     You are correct: you have a problem with the oxygen sensor circuit.       It's staying at the bias voltage of 0.45V all the time and so is permanently in open loop.      No wonder CO is high!      And it never moves off 0.45V.     If your oxygen sensor heater wasn't working, you would still see movement in the voltage but yours is totally dead.     Can you check first to see if, after the exhaust repair, the oxygen sensor connector was reconnected?        Your circuit is so dead that you have an open circuit.     If it is connected, can you disconnect it and check to see that the pins have not been damaged or corroded?       (If you had a short circuit, you would see 0V not 0.45V, so that's why I think you have an open circuit somewhere.)        The 0.45V suggests the MEMS ECU is good, but that you have an open circuit in the sensor signal wire or the return (sensor ground).       When you disconnect the oxygen sensor connector (assuming it is indeed connected) to look at the pins/connections, perhaps you could measure the voltage across the signal wire and the signal ground wire?     You would need to put 2 very small diameter pins in the 2 female connectors and turn the ignition on (engine NOT running).    Do not open up the female connectors with the pins.   If the circuit is good up to that point, you will measure less than 0.45V, because the bias voltage is so weak that even a high impedance voltmeter will drop it.     With a cheap voltmeter you might even see only 0.2V; that doesn't matter: you just want proof the wiring is good up to that point.   ( If you measure zero volts then you have a break somewhere between that connector and the ECU.)   

You said you measured a voltage changing between 0 and 0.7V across the terminals of the oxygen sensor.    So that shows the oxygen sensor is good, but the signal just isn't getting to the ECU.

[Sylvain O.]

Hi Martin and thank you for the last return,
I'm going to look at the exhaust leaks, but from the start of the manifold to the catalyst it was done by a professional. After his intervention all the exhaust noises had disappeared. He placed a reinforced joint and changed all the dowels, as for the exhaust downcomer and the bolts, he did not use a joint paste. He changed the lambda probe and the water temperature sensor at my request. After I changed the catalyst myself and I tightened the exit screws that leaked a little.
I did not understand the test of joints with a pipe, can you explain to me?
For the oxygen sensor and the values ​​raised, it's been a few months that I go round in circles after recovering various software (memgauge, memsrosco, memsanalyzer, memsscan, great job from developers) the sensor connector is well connected and I tested the continuity to the ecu terminals (pin 7 oxygen sensor signal and 18 oxygen sensor return) I placed a multimeter and I see the electric current pass and oscillate (000mV to 800mv). Unfortunately I have already done all the continuity tests that you suggest.
I also think that the ECU does not receive the signal, but it would appear to be more serious, internally at the ecu card level, eprom or other .... but I do not know how to test that. I thought to send the ECU in exchange standard but it costs more than 300 €, without the guarantee that it will work in reassembly
Would not there be a relay that controls the probe that would fail?
Sincerely, Sylvain

[Martin Rubenstein]

Very sorry, Sylvain: I didn’t explain it well. Use the plastic pipe like a stethoscope. Put one end to your ear and move the free end to anywhere where you have a joint. You will soon hear any leaks.

You have now produced a logfile and all the other parameters are showing so your interrogation of MEMS is good and it would be safe to assume the failure to display lambda sensor information is real and not an artefact of the logfile.

You’ve proved the lambda sensor is switching between high and low, so the oxygen sensor is good.

Now, can you be really sure that the signal across pins 7 and 18 is getting all the way to the MEMS ECU? I don’t like back-probing (pushing a pin into the back of the ECU connector (to pins 7 and 18)); it prefer to pierce the insulation with a piercing probe (and sealing with ladies nail varnish afterwards - and I’m not too happy with that, either!) but such methods would allow you to know if the signal from the oxygen sensor is arriving at the ECU coonector. (Assuming also the signal then passes into the ECU). Could you test that?

Can you buy the same Mini MEMS ECU on ebay cheap? Would it work or is there an immobiliser that would stop it working? I have bought 3 or 4 MEMS ECUs for my Rover 100 on ebay and defeated the immobiliser part so I can use these ECUs. That eould be cheaper than sending the ECU to be tested.

[Sylvain O.]

Hello Martin,

No problem, I will try with the pipe stethoscope p to check for leaks.

Yes indeed, all the log captures show the defect of lambda values ​​that do not go back to the ecu, I am forced to question the proper functioning of this ecu. But I wanted to be sure and your experience confirms it too.

Yes the oxygen sensor is good and I had the same problem with the old probe.

The old probe allowed me to test to see if the resistance was heating up.

I will try again continuity tests up to the ecu, maybe also make a weld inside the ecu on the connecting rods to the card (7, 18), or drill the cable with a Needle as you suggest ... I will work on the mini this weekend, I also have some operations to do, the front and rear lubrication and the bleeding of the brake fluid.

It's been a long time since I question the proper functioning of the ecu, yes it would be necessary that I find a used ecu ... the model of ecu which is mounted on my car is a AJ MNE101150 0865 4044, only it is necessary to be Confident about the condition of a used case that could be defective or blocked ... .. I will look at this, but I did not find it on ebay.

At my dealer DATCH parts in the Paris region the standard exchange costs € 390, it should be noted the problem, they are repaired in Britain.

In my area there are some professionals who call themselves specialists in minis, but they have never managed to help me out, in general they do not know minis with injection.

Anecdote, this year I started using memsdiag and analyze it allowed me to see that the throttle potentiometer was incorrectly mounted on the body of the injector, the butterfly did not train the potentiometer ....

Sincerely Sylvain

[Martin Rubenstein]

I’ve attached 2 photos of the bias voltage at the female (ECU) end of the oxygen sensor connector i.e. the sensor wire and the sensor return/ground wire. I see 647 mV with the ignition on, showing a good circuit. I don’t understand why it’s 647mV: I expected a maximum of 450mV - doesn’t matter because all I want to know is, is the wiri g good between the ECU and the 02 connector, which it proves it is.

[Martin Rubenstein]

Just for interest: the bias voltage from the ECU is 650mV with the 02 sensor disconnected. When it’s connected, the bias voltage measures 400mV (measured with a Fluke voltmeter) and 200mV (measured with an oscilloscope), which shows that even a 1MOhm impedance ‘scope is enough to significantly drop the measured bias voltage, because it’s so weak and exists only as a diagnostic to test the circuit.

[Sylvain O.]

Indeed it's interresting

[Martin Rubenstein]

A follow-up to the bias voltage measrements I took yesterday (~600mV with the 02 sensor disconnected and ~400mV with it connected using a very high impedance voltmeter).

I wondered if the information from MEMS could be used to detect an open circuit and therefore save further testing. So I used Haro’s and Leopold’s programs (as well as Al’s MEMSLogger to double check). With the 02 sensor connected and the ignition on. (Engine not running), the bias voltage registers as 435mV. Disconnecting the 02 sensor makes no difference: MEMS still registers 435mV (i.e. it does not increase).

So it’s still necessary for Sylvain to make continuity measurements from the 02 sensor to the MEMS ECU connector.

[Martin Rubenstein]

Example of piercing probes which could be used to check that the lambda sensor oscillating signal is being conducted to the MEMS ECU connector. Alternatively, sharp, very fine pins could be used. The breaks in the insulation must be sealed with liquid tape; I use nail varnish instead.

https://rover.ebay.com/rover/0/0/0?mpre=https%3A%2F%2Fwww.ebay.co.uk%2Fulk%2Fitm%2F181179386348

[Martin Rubenstein]

Sylvain,

In your first post you wrote, “I was able to bring out the lambda value and the pollution by reducing the fuel trim richness to -128 but that is not enough.”

Are you sure that helped? You see, in closed loop, whatever adjustment you make to the fuel trim should be countered by the short-term fuel trim, which is taking instructions based on what the oxygen sensor is telling the ECU at that time and not on a static adjustment previously made by the user.

That’s the theory, but if you’ve found a way to help us all get tgrough the emissions test, we’ll all be interested. You might even get a job offer from VW.

So do you know if it really helped?

[Sylvain O.]

Martin,

Some time ago, I passed my vehicle to a mechanic who does pollution checks, the first time the gasoline richness was 0 and the CO2 and lambda values ​​were much higher than the second time with a wealth at -128, shame I did not keep the records ... ..

Also with a wealth of 0 the car goes much faster in laps at full load while at -128 it is much more sluggish and climbs less quickly in the towers.

I could also see the difference with the multimeter mounted on the probe at idle and at full load, the voltage is much higher with more wealth sign that the car produces more CO2 .... By decreasing the wealth the voltage decreases, but not enough to be in the norm for the French CT.

I saw my car mechanic today, he is a mechanical specialist of minis with carburettor, he is on vacation at the moment, when he comes back at the beginning of september, I will do the tests again and I will read the CO2 and Lambda values, besides he told me he had an ecu and I can test it on my car, I would like him to sell it to me hoping it is compatible ......

Tonight I did some tests, I raised the voltage from the pins 7 and 18 at the socket of the oxygen sensor (the probe is disconnected), the readings are:

000mV contact OFF,

619mV contact ON engine not started,

619mV engine started,

This proves that there is continuity between the socket of the probe and the ecu.

I tested the exhaust leaks with the hose and there is none.

About the water temperature, I let the temperature rise to 90 ° C, but I suspect that the carlostat does not open, I did not produce a log file.

Sincerely Sylvain

[Sylvain O.]

Sorry my English is very bad, google translation helps me a lot, but reading again I see a lot of mistakes. I hope you understand what I write .....

[Martin Rubenstein]

Fascinating, Sylvain.

Indeed, your voltage tests prove tou have continuity from the 02 connectir to the ECU. And it proves the ECU connector is good. So it looks as if your ECU is faulty; however, the one thing thats not been tested is whether you have continuity across the connector that the 02 sensor plugs into. You could do a visual inspection, I suppose. Just unplug the connector and lift out the yellow plastic piece in the female end and check that all 4 females are closed so they are a good tight fit around the male pins i.e. no-one has pushed a pin into them and opened them up. And check the 4 male pins are clean and straight. If that connector is good then it really does look like a faulty ECU.


What you call the “wealth”, is that the richness of the mixture? Is it what we’ve called tge Long Term Fuel Trim (LTFT) in the software? You mentioned the “wealth” setting of 0 and also -128; Did you change the LTFT setting from 128 to zero using one of the MEMS software programs?

[Martin Rubenstein]

Sylvain,

If you reduced the LTFT.from 128 to zero in the software, then I think I know what the problem is.

I once reduced my LTFT from around 128 to 60 to see what happens. MEMS went into open loop and stayed there until I put it back. 60 was too low for MEMS to adjust to using fuel trim. So it gave up and stayed in open loop. That may well explain why you don’t see a lambda trace in MEMSAnalyser and why your car runs better. In open loop I think the mixture is a rich 13:1, whereas in closed loop it’s about 14.7:1.


If you put the “wealth” back to 128 and made a logfile, I think - if I’ve understood things correctly - your lambda graph will come back in the logfile. Possibly.......

[Sylvain O.]

Martin,
I checked the connector, before the male and female connectors and through.
Yes I used the Mems-scan software to change the value "Fuel Trim LT.

[Sylvain O.]

Martin,
I will retry log records, but this had no effect on the lambda log from switching LTFT from 0 to -128 or from 0 to 128.
As I said at -128 the lambda voltage is reduced so less CO2 and I found that I consumed much less gas .....

[Martin Rubenstein]

Sylvain

Which software are you using to change the LTFT? Perhaps it is better if you use the software to reset the ECU to Factory Default settings and then make a logfile instead of changing LTFT. I don’t recognise the -128. Seting to Factory Default will put every setting back to normal. A good starting point.

[Sylvain O.]

Hello Martin,

You wrote,

 “Are you sure that helped?      You see, in closed loop, whatever adjustment you make to the fuel trim should be countered by the short-term fuel trim, which is taking instructions based on what the oxygen sensor is telling the ECU at that time and not on a static adjustment previously made by the user.”

n my case, as the ECU does not read the values ​​that come from the O2 probe, it is the value forced by the user that is prioritized.

Also, I was able to note according to the use of the vehicle and certain conditions, the value "Hot idle position" evolved automatically. But honestly I did not understand this setting, I could see that the setting evolved in cases in the air filter box or in the body of the injector with the pressure pipe ???

Sincerely, Sylvain

[Sylvain O.]

Martin,

I am using mems-scan-0.5.2-Windows

I will reset to the factory value and I will do a log sampling starting from 0 and gradually decreasing to -128 and increasing to +128.

If you have an ecu that works well, if you change the values ​​manually, the ecu must readjust the setting automatically. Do you know the time to readjust the settings ecu.

[Martin Rubenstein]

I will look at mems-scan again; I have forgotten what the LTFT adjustment looks like in it.

However, there is no need to adjust LTFT. I suggest you leave it at the Factory Reset setting. You will find - on a working closed-loop system - that, at idle, if you make a small change to LTFT, the STFT immediately changes in the opposite direction to balance it and make sure that the fuel mixture is correct. That’s in closed loop of course.

So I’d say do a Factory Reset and don’t make any adjustments yet.

The LTFT does adapt (change on its own) to the “best” value. So a Factory Reset will set it at 128 (not sure if that’s the setting in mems-scan, but it is 128 in the other software). And then after 100 miles or more you might see it come down a little eg 124.... I’ve seen 116. But it is of no consequence. All it does is come down to a value that helps STFT get as close to zero as possible. So it is best not to make any adjustments to it. In closed loop, manual adjustments at best will do nothing and at worst will stop the ECU controlling the mixture properly.

As for re-learning, other than LTFT adapting, the only other setting I know of is the idle speed. Depending on the type of journey you are doing eg in town or on motorway, the correct idle speed is learned quite quickly I’ve found. In fact, mostly I’ve not been aware of any learning because the idle seems pretty good after a reset anyway.

[Sylvain O.]

I will try what you suggest ;p

[Martin Rubenstein]

I’ve refreshed my memory and looked at Haro’s program (mems-scan 0.5.2). His fuel trim numbers indicate the deviation from the default point. So when you Reset ECU, your Fuel Trim will be zero. (At the moment mine is showing -12, which shows as 116 in the other software. It started at 0 (or 128) and has adapted on its own to the present value. It doesn’t mean anything and I leave it alone. It settles at whatever point it wants to.). What is of use though, is that if ever I see that number has changed by a significant amount, it would indicate a possible fault. For example, if instead of -12 (in mems-scan) I saw it had changed to, say, -2, I would immediatelt start looking for a vacuum leak and the other reasons why the ECU might give more fuel eg a faulty 02 sensor.

So reset the ECU; then turn off the ignition for 30 secs (I don’t think you need to but play safe) and then start logging and go for a short drive. If you still don’t have a lambda ‘waveform’, then you can be nearly certain your problem’s inside the ECU.

Again, I wouldn’t alter the Hot Idle Position (unless it’s for experimentation and temporary). It’s a learned parameter, so MEMS should change it back anyway to whatever it was after some miles driving around town.

[Sylvain O.]

Hi Martin,
I followed your last instructions.
Reset of adjustments and ECU, OFF of the contact then start of the engine, I let the engine warm a moment. The LFTS remained at 0 after 30 minutes of roads and only the "hot idle position" is
automatically adjusted to 12, yes the ecu has a problem. After recipe adjustments and the start of the car I could see that the voltage of the O2 sensor had increased to more than 900mV without
ever falling below 700mV, which is equivalent to a mixture too rich causing too much CO2. Unfortunately in the middle of my circuit the temperature sensor slammed ...... 
I send you the log as an attachment.
I noted your indications of your last email concerning the start of
another ecu.
Regards,
Sylvain

[Martin Rubenstein]

Not looked at your logfile yet, but you’ve done all the right things to prove that the ECU is at fault. The only thing you haven’t totally ruled out is the oxygen sensor connector, but you visibly inspected the pins and sockets and they are good.. So you’re as certain as you can be that the fault is in the ECU.

And the oxygen sensor is correctly reporting a rich mixture, because the car is in open loop and running at ~13:1. So hopefully when you run with your mechanic’s ECU, you can take a logfile at the same time and we’ll hopefully see the car go into closed loop and the lambda signal appear.

Perhaps the temperature measurement will be more realistic?

What happens if you drive the car on a hot day for 30 mins or more? What temperature do you see then? Even if you removed the thermostat, under such conditions, after 30 mins or so, the temperature should be hot.

I just wonder if the ECU is under-reporting the true temperature, perhaps through the same fault that’s affecting the lambda signal?

[Martin Rubenstein]

Whilst you wait to see your mechanic next week, there is another test you might do. Is it easy to get to the temperature sensor connector? That should be a 2-pin connector that sends its signal to the ECU, NOT a one-pin connector that feeds the temperature gauge on the dashboard.

If you disconnect the temperature sensor, what voltage do you measure across the connector coming from the ECU?

[Sylvain O.]

Martin,

The car warms up normally, I talked with my DATCH part seller, they used to put together minis to sell them, 80 ° C is the temperature for a mini that works well. Yes it happened that the car heats more in traffic jams and in hot weather.

Fortunately it is not the temperature sensor that was down, the temperature sensor is disconnected by the vibrations ... .. My mechanic did not work very well ... ...

I replaced the value LTFT at -128, the multimeter tells me 045mV idle, while after the default value recipe the measured value is over 900mV, it has a real effect on pollution.

The measured value of the O2 probe while the temperature probe was disconnected was the same and the forced change of the LTFT I used to do -128 had no effect on the CO2 the meter was indicating 900mV.

Good evening, Sylvain

[Martin Rubenstein]

Hi Sylvain,

Got it.    Just looked at your logfile and your post about the temperature sensor being disconnected.    Sorry: maybe you already said it and I missed it?     Your temperature shows 60C because the sensor is disconnected.     MEMS has substituted a value of 60C because it realises there's a fault.    MEMS is in LOS - limp home mode - and I would expect that in LOS it goes into open loop even if it knows the oxygen sensor is working.

Here's what Chapter 14 says (http://workshopmanual.com/doc/d3167-rovermems-pdf/read?page=3)

"LOS (limp-home mode) MEMS has a limited operating strategy (LOS) or limp-home facility, and in the event of a serious fault in one or more of the sensors, the EMS will substitute a fixed default value in place of the defective sensor. For example, in limp-home mode the coolant temperature sensor (CTS) value is set to 60°C, the ATS is set to 35°C, and engine load is based on rpm. The engine may actually run quite well with failure of one or more minor sensors. However, since the substituted values are those of a hot engine, cold starting and running during the warm-up period are likely to be less than satisfactory. Also, failure of a  major sensor, ie the MAP sensor, will lead to a considerable reduction in performance."

Fix the temperature sensor and you should go into closed loop and everything will be as it should be.     Reset the ECU (or set LTFT back to zero) once you've fixed the temperature sensor.       How soon can you fix the temperature sensor?       (You could put a resistor across the connector to fool MEMS if you have some resistors or a potentiometer.    You could try 1500 Ohms with a cold engine and maybe 500 Ohms with a hot one.    You only want to see that this theory is correct and there is nothing wrong with the ECU.     

See how easy it is to blame the ECU and to go and buy a new one only to find the problem is still there!    

[Martin Rubenstein]

Hat's off, once again, to the astonishing work done by Colin, Al, Leopold and Haro: without their invaluable software we'd all be stumbling about in the dark, changing parts until we either run out of money or hit upon the faulty part.

[leopold....@t-online.de]

 I wonder if i should redraw the possibilities for changing ecu settings and forward the software only to authoriced mechanics!

LeopoldG

[Sebastián González]

Hello everyone! Sorry for the delay, but didn't have time until this weekend to perform the test that Martin asked for. I've attached a log since the car its started including all the driving time. We did a run to the nearest highway in order to test the values. The car runs very smooth (and fast!). I have a doubt concerning the Open Loop indicator: what's the function? I saw that during the recording of data, it changed from on to off numerous times, ¿is that normal? On the other side, all the other values seem normal (battery voltage, water temperature, etc) Regarding lambda sensor you can seen on the log that under load it keeps high values permanently. Thanks in advance!

[Martin Rubenstein]

Unfortunately, I cannot read your logfile in MEMSAnalyser - I get the message that it's not a valid MEMS data file. Nevertheless, using Leopold's Readlog tab in his mems-rosco-logs, I can see that your lambda sensor readings are now recorded. At last! So your ECU is good, But I can't see in Leopold's program when it's going into open loop: it doesn't work as clearly as in MEMSAnalyser, However, looking at your lambda graph, I can tell where I would expect it to go into open loop.

Did you record your logfile with a different software? Yesterday's logfile was easily read by MEMSAnalyser but today's isn't. It's a huge disappointment that compatibilty with MEMSAnalyser is being compromised.

Closed loop is for steady-state driving, constant speed or let's say sedate driving, when the emissions can be controlled. If you need to accelerate hard, youe are demanding power, and you won't get that if the mixture is held at 14.7:1: the car would hesitate when you open the throttle So you can't have emissions control and hard acceleration, so MEMS has to come out of closed loop and give you a richer mixture. When the load has reduced, it drops back into closed loop. So it's working exactly as designed. You will see if you drive carefully, with gentle accelerations and loads, you will stay in closed loop nearly all the time.

So any ideas why I can't read your logfile with MEMSAnalyser?

[Martin Rubenstein]

I now see from the title and frame layout that yesterday you made your logfile using Al Richey’s MEMSLogger.

And today’s logfile - which I can’t read - must have been made with either mems-scan or mems-rosco-logs.

Any chance you could make another logfile now that your car is working but with MEMSLogger again?

By comparing yesterday’s (when you were in “limp-home” mode) with a logfile where everything is good, we might see if one of the unknown parameters relates to limp home.

Plus, with being able to read the logfile with MEMSAnalyser we are at a disadvantage.

[Martin Rubenstein]

I forgot to mention: you asked what is the function of the open-loop indicator.

Haro and Leopold included this indicator so that we can have a simple, quick real-time indicator of the emission-control system. With your laptop on the passenger seat, you want to see - in real time -that light turn green within a minute or so of starting the engine. Before that, you would have made a ligfile and then looked at it afterwards to find out if you had been in closed loop. Now you know instantly. And if it doesn’t turn green, you straightaway know that there is a problem somewhere. Another very useful, if not invaluable, feature.

[Sebastián González]

Hello Martin! Thanks for the fast answer! I used the same method as the other times, laptop with memsanalyser and connected to the car. I'm checking the .txt file and seems to be in the same format as the oher one. With the previous one I also couldn't read it, it says that is a not valid MEMS data file :( I understand now the function of the closed loop, as we were driving on the road it did exactly what you described: when load is applied it turns off and when you coast or press the accelerator slightly it enables. My ECU was perfmorming OK, was Sylvain's who had different problems than me. This week we are going to test with a mechanic the CO emissions again and if the open/closed loop makes a change on that numbers, we'll see. Thank you all for all the help, I will continue reporting!

[Sylvain O.]

Good evening Martin and Leopold,

The cable of the temperature sensor is off this Friday 17/08 evening during a test drive, I reconnected the sensor Saturday 18/08, I have the chance to have a garage for my operations of mecanic. As a reminder the mechanics is a hobby and the vehicle that I restore makes me very miseries ..... But it is a very pretty car. When the pollution problems are solved I can go inside the vehicle.

I think you know, if the sensor is disconnected or the sensor is off the engine does not start, then we must use the "pilot start" that is sprayed in the air box, it simulates a primer of gasoline that ignites in the cylinders and it starts. It seems that the ECU does not inject gasoline when it does not see the presence of the temperature sensor. When the engine is hot no problem with a faulty temperature sensor, it starts.

Of course I redid a test (even two) and I still have the same problem the loop opens after a few moments.

Attached is a log file with LTFT defaults, and a log file with my LTFT setting at -128.

Still no views of Lambda voltage values ​​...

Do you seem to doubt my measures?

With the default setting of the LTFT I get between 750 and 950mVa on the multimeter the measurement never goes below, the oscillation is very limited and stays high. High fuel wealth setting = pollution. >>> In attachment «Result 19082018 default value LFTS.txt»

With LTFT forced tuning at -128 I get idle values ​​of 035mV and charging it can go up no higher than 850mV. Low Fuel Wealth Setting = less pollution. >>> In attachment «Result 19082018 value LTFT minus 128»

Too bad that it is not verifiable with the log file produced with MemsLoger (the ecu of my vehicle is a very sick).

I'll have to take a resolution soon, have the ECU repaired if it turns out that the ECU is faulty, Replace the ECU, but it seems there is not a lot of ECU on the market and prices soar, migrate the injection system to the benefit of a carburetor HIF44, the latter choice would be a shame because my goal is to keep my vehicle its original design.

Regards,

Sylvain

[Martin Rubenstein]

When MEMS detects infinite resistance in the coolant sensor circuit, it goes into limp-home mode (see earlier post). It substitutes a temperature of 60C. So when you try to start a cold engine in limp-home mode, it’s only getting enough fuel for an engine at 60C. So it’s difficult to start. If the engine’s hot then it’s getting the correct amount of fuel (or nearly) so it’s much easier to start a hot engine in limp-home mode. Hence the name, limp home: designed to get you home or to your mechanic’s garage for one journey only.

[Martin Rubenstein]

And in limp-home mode, your oxygen sensor signals are ignored; the car stays in open loop.      So even if your oxygen sensor circuit is good, the signal won't be displayed.

If you look at your 2 logfiles from yesterday, Result 19082018 default value LFTS.txt and Result 19082018 value LTFT minus 128.txt, the second one has a fixed temperature of 60C and the Lambda graph shows a zero status all the time.    The first logfile where the temperature sensor is working, shows a lambda status at One for a minute or so and then drops to zero (stays in open loop) afterwards.

Did you disconnect the temperature sensor for Result 19082018 value LTFT minus 128.txt?       If not, then the car has gone into limp-home mode because you reduced LTFT to an impossibly low value -128.

You should set LTFT to zero and leave it there.    By altering it, you are introducing a second fault that is making it difficult if not impossible to solve your main fault.

And if you were to fit a carburrettor, you would still have to pass your emissions test every year; would that not be nearly impossible with a carburrettor?

[Sylvain O.]

Hello,

I was wrong log file and I send you the good "Result 19082018 value LTFT minus 128.txt" I'm sorry.

I started the MEMESLogger for this log file while the engine was running, that's why we do not see the beginning of the sequence with lambda status on One for a minute and fall back to zero.

For information in France the technical control is done every two years.

Even if it's not what I want, but with a carburetor it is possible to manually adjust the richness of the gasoline mixture.

Regards, Sylvain

[Sebastián González]

I think that we found the origin of the CO emission. Yesterday during the test a rattle started around the catalytic converter and today se disassembled it. It's in pieces in the inside! We took it to antes specialist Andrés he is going to rebuild it. Will write when a I get It back!

[Martin Rubenstein]

Sylvain

Can you carefully back pin Pins 7 and 18 - on mine a grey wire and a green one behind it, and, with the engine running, can you prove the lambda signal is reaching the ECU?

[Martin Rubenstein]

Sylvain

I think it would be a bad idea to revert to a carburrettor and especially for a fault that should be quite easy to find and fix, even if, at worst, it meant an ECU repair. And you say, “... but with a carburetor it is possible to manually adjust the richness of the gasoline mixture.”; I know the mixture can be adjusted at idle on a carburrettor, but would you be able to also adjust the mixture independently (of the idle setting) for the 2000RPM emissions? (Is that what you called the accelerated CO?)

[Sylvain O.]

Martin,
OK I will repeat the continuity test in front of the ECU connector tomorrow, pins 7 and 18. For the change in carburetor, I'm not there yet. I will monitor if ECUs are on sale to have a spare.
good evening, Sylvain

[Sylvain O.]

Hi Martin I did the continuity test on the back of the ECU as you requested, the voltage values ​​of the ECU and lambda sensor are clearly visible on the meter's display. 629mV after switching on the ignition, then oscillation of the voltage coming from the Lambda sensor after starting the engine. Sincerely, Sylvain

[Martin Rubenstein]

Then that confirms the previous tests you did. It does seem the ECU is faulty but the one thing that keeps nagging at me is that the ECU is still capable of measuring the bias voltage (~450mV), which make one wonder why it’s not measuring the incoming signal?

Are you going to try your mechanic’s ECU?

[Sylvain O.]

Yes, I will try the box of my mechanic on his return from vacation in early September.

What I have read about the Lambda sensor is that 450mV is the average value for getting closer to the best fuel injection dosage, too much above the air / fuel mixture = is too rich, too much below the mixture = is too poor. The values returned to the ECU allow it to adjust the gasoline dose with respect to the opening of the throttle.

[Martin Rubenstein]

Sylvain

Leopold has made me aware that I don't understand the green Status line in the lambda graph in MEMSAnalyser, and it is clearly a very useful parameter.  

Normally, the Status line is always equal to one.      It should not drop to zero.

Even when I disconnected my coolant temperature sensor to make the car stay in open loop and go into limp-home mode, it started at a value of one and stayed at one.     And when I set the LTFT at the impossibly low value of -128 (default value is zero) to copy what you did, I still saw the Status start and stay at a value of one.

On my car, the only time I have seen the Status line start at a value of one and then after about half a minute (or more) drop to zero is:

1.   When I deliberately crossed the lambda signal wires so the signal went to ground and the signal return (signal groaund) became the signal.   But, despite the Status dropping to zero, I did see activity on the lambda signal ( moving between 0.45V and zero).

2.    With the heater relay removed -  the Status dropped to zero after any time between half a minute and 1.5 mins and the lambda signal stayed flat at 0.435V.   However, that was on a cold engine.      When I tried it (heater relay removed) on a hot engine (coolant over 60C), the green Status stayed at One, and the lambda voltage signal oscillated between 0.1 and 0.8, (but it wasn't quite as good as when the heater was working).

So, in summary, the Status is something independent of Limp Home or Open Loop.     It seems to be related only to the oxygen sensor circuit.     And even the presence of the 0.435V bias signal is not enough to stop it dropping to zero if the ECU sees a fault.

I don't know how it works because I've seen the Status drop to zero (heater disconnected) quicker than the time it usualy takes for the lambda sensor to warm up.   So it's something other than the time to warm up of the sensor that's telling the ECU there is a fault.

Bottom line is: if your Status drops to zero after the first minute or so, you have a problem.     And somethimes the lambda signal is not displayed.

As Leopold pointed out (not sure if it was on the forum), you must check you have a supply to your lambda heater.     If you have a current clamp, that would be even better: it would show if the heater was working, whereas a voltage measurement only shows if you have a power supply; it doesn't prove power is being drawn.

CHECK YOUR HEATER CIRCUIT THOROUGHLY!        And I must try and understand what Status is all about, and what causes it to drop from one to zero.

Again, it suggests there may well be nothing wrond with your ECU: it may well be doing wverything it was designed to do.

This might be very friustrating for you, but it's helping us to understand things a bit better!

[Sylvain O.]

Martin,

Yes indeed from the start of the engine until it stops, the loop should remain at 1 and in my case whatever the setting the loop goes to 0 after a few minutes, indicating a sensor signal problem.

1. I also did the test, I crossed the polarity of the son of the sensor (gray and black) and I had the same result as you.

2. I also tested with the relay by connecting and disconnecting it.

I checked his voltage while the engine was running. I did this test also with another probe that I plugged in outside the collector to check that the voltage was present all the time and if it heated up all the time, the loop also fell 0.

All the tests I did were done with a multimeter connected to the probe and there was always a signal on the multimeter. I think you're talking about an ampermeteric current clamp? no, I do not have one. But it is possible to measure the amperage of the heating circuit.

Indeed it is very frustrating because I have already done a lot of tests without understanding what is happening and where could the problem come from.

I had thought for a moment to build a MEMS cable in Y in order to be able to measure all the circuits ...... it would be necessary for that two taken MEMS females and a plug MEMS male ..... but where to obtain his Fiches ?

Sincerely, Sylvain

[Sylvain O.]

Hi Leopold et Martin

OK, I will do what you ask, but I will not be able to drive as I wish, because the technical control of my vehicle is exceeded.

Sylvain

Cordialement,

[Sylvain O.]

Waiting to provide you with new log files, with a reset of the ecu and default settings. In PJ old log files that I had saved with mems-scan-0.5.1-windows.

[Martin Rubenstein]

A few days ago I ran some tests on the lambda Status parameter as displayed in Al’s MEMSAnalyser on the Lambda page. It seems to be a monitor of both the heater circuit and the signal circuit. Status should be 1 all the time; if it starts at 1 and falls to zero within say a minute of the engne starting, there is a fault in the heater circuit, the sensor circuit or both.

1. Heater circuit working, lambda signal circuit
open: Status dropped to zero within 30 sec of engine start. Bias voltage 0.435V all the time.

2. Lambda signal circuit shorted to ground (heater working): Status drops to zero 50 sec afer engine start. Bias voltage zero all the time.

3. Sensor connector disconnected. Status drops to zero 25 secs after engine start. Bias voltage is 0.435V all the time.

4. Coolant temperature 60C, and heater relay removed to disable the lambda heater. Status drops to zero after 20 secs with high engine revs. Lambda voltage 0.435V all the time with no lambda activity, although in the past I have seen lambda activitty with a hot engine and no heater. (Perhaps it depends on coolant temperature and the exhaust gas temperature quickly heating the sensor before Status falls to zero?)

I don’t know if there is any significance to rhe time after engine start that Status drops to zero. Perhaps it depends on the number of engine revolutions? Subject for a future test.

How ever it works, Status seems to be a very clever and valuable indicator of the total oxygen-sensing circuit.

[leopold....@t-online.de]

Lambda Status is the important signal on a working system.
 Cosed loop is the second important signal, but this changes in conjunction to acceleration.Lambda voltage is stable between 400 and 500mV if there is a fault on the lambda circuit.If it is shorted,here can't be a current.
 I will work this into mems-rosco and give lambda status the same priority as closed loop.
Thank you  Martin for your investigations.
Leopold

[leopold....@t-online.de]

@Sylvain

what to do with old log -files?
Is this thread a fake?

[Martin Rubenstein]

Sorry Sylvain; are you talking about those 2 logfiles 2018-06-02?

Until yesterday I’ve had Internet problems because my router got corrupted and t forgot about those 2 files.

As with the other logfiles, the lambda grapg together with the Status line shows the oxygen-sensing system is faulty.

Have you checked the heater circuit yet? A current measurement would be ideal, because you would still have 12V at the heater even if the heater was open circuit. But you could test for voltage at the connector and then test for resistance across the heater terminals.

[Sylvain O.]

Hello,

Yes I measured the resistance, it seemed good, but I do not remember the value.

I will test again. To answer Leopold, I provided you with these files thinking they might interest you.

I did not understand the meaning of "Is this thread a fake?

Cordialement Sylvain

[Martin Rubenstein]

It draws about 1 Amp when hot, whick would equate to 12Ihms. So I’d expect your resistance cold to be less than that. So was it less than 12 Ohms?

And voltage? Do you have 12 V on the heater contacts in the connector? (Possibly, the engine might need to be running, but try it with ignition on first.)

[Martin Rubenstein]

Sylvain

When I subtract the resistance of my test leads (0.1 Ohms) I get a heater resistance (cold) of 3.6 Ohms across the 2 white wires. So you should be seeing that sort of meadurement: sometthing around 4 Ohms, depending on the resistance of your leads and the accuracy of your meter.

[Sylvain O.]

Hi Martin,

At the connector terminals (white wires), I measure 6.8 ohms on the cold spare sensor,

I measure 8.5 ohms (the engine is slightly warm) and the cords are longer on the sensor connected to the car.

The value of the resistance rises when it is hot and decreases when it cools.

The measured voltage is 13.63 v on both sensors.

I received a spare ecu not too expensive and tested it by following your instructions.

A few seconds the car is put in startup, but did not start at the end. I placed the command D3 (0xD3) but it did not have any effect.

Sincerely Sylvain

[Martin Rubenstein]

Sykvain

The resistance of the heater is good, and you previously measured the switching of the sensor, so your oxygen sensor is good.

But please explain what you mean by:

“The measured voltage is 13.63 v on both sensors.”

On “both sensors”? I don’t understand.

Are you talking about the voltage at the heater terminals?

Do you mean you have 13V on each of the heater terminals? If so, that is where the problem is. PLEASE EXPLAIN EXACTLY - I appreciate the difficulties of writing in a foreign language.

Thank you for the screenshot showing that you sent D3 and the ECU returned D3; that is correct. It seems, then, that perhaps that command D3 only works on the system I have, and you have a different immobiliser system. I can only suggest you contact the mini forum or search on google and find out how to disable the immobiliser section in the new ECU.

[Martin Rubenstein]

If by “The measured voltage is 13.63 v on both sensors.” you meant both heater terminals, could you make 3 voltage measurements with the engine running:

Heater pin 1 to battery ground

Heater pin 2 to battery ground

Heater pin 1 to Heater pin 2

[Sylvain O.]

Hi Martin,

sorry I mis explained. I have a spare O2 sensor that allows me to compare the measured values ​​to that sensor that is mounted on my car.

On each of the sensors I measured the same value of voltage = 13.63 v when they are connected.

I measured the voltage between + and - sensor. I also measured between the sensor + and the mass metal of the car.

[Martin Rubenstein]

Thanks, Sylvain.

Then you seem to have proved both sides of your oxygen sensing circuit are good.

I believe there are people on the mini forum who know how to get a spare ECU working. They don’t make the knowledge public for obvious reasons, but once you make contact you’ll be able to get the information by private messages.

Keep in contact with us and tell us what’s happening.

[Sylvain O.]

Merci Martin de votre patience et expertise je vous tiendrai informé sur l'avancé de mes réparations.  

[Sylvain O.]

Thank you Martin for your patience and expertise I will keep you informed on the progress of my repairs.

[Martin Rubenstein]

Not a problem. Sorry we’re still trying to find the problem.

You don’t need the ignition on for the earth/ground-wire integrity check. The heater circuit is power-side switched via the relay. The earth/ground should be unswitched. So you can test it with ignition off.

[Sylvain O.]

yes I do this, contact on and contact off

do you still need these measurements below?

Broche chauffante 1 à la masse de la batterie 

Broche chauffante 2 à la masse de la batterie 

Goupille chauffante 1 à goupille chauffante 2

[Martin Rubenstein]

No, Sylvain, we don’t need them. I had misunderstood. I hadn’t realised you had 2 sensors, and I thought that, possibly, you meant you found 13V on each heater terminal. But you corrected me. So there’s no need for those measurements.

But I think all we need now is to know about your heater earth (ground). You showed you have 13V to the heater with the ignition on (and engine running?), but we just need to know that the other terminal is connected to ground; hence, the voltage check between battery positive and that heater terminal (the ground).

I know you think the ECU is faulty, and you’ve dine all the testing on your car, but I will be a little bit surprised if the fault really is in the ECU. But we’ll see.

[mrgry...@gmail.com]

I hope I understand the measure you want ;p

I measured the voltage between the + battery (cable from the battery to the starter motor) and the sensor ground terminal = 0.0 volt.

I did the same measurement with the spare sensor, I did touch the body mass of the sensor mass of the car = 0.0 volt.

[Martin Rubenstein]

mrgryfiron

The voltage measurement I wanted Sylvain to make is so he could check his heater ground/earth/return is good.

He said previously he’d checked the oxygen sensing circuit with a voltmeter with the engine running and saw that it switched from 0.1 to 0.9V.

He also checked the power supply to the heater in the oxygen sensor and that was 13V. But I wasn’t sure he had checked the heater return i.e. heater ground/earth part of the circuit.

The heater wires on the sensor are the 2 wires (often white) that are the same colour, One will be supplied with 12V and the other should connect to ground/earth/battery negative. One way to check the integrity of that wire, and be sure it really does go to ground would be to take a 12V test lamp and connect one end to battery positive and the other to the heater-return terminal in the female connector (not the connector on the sensor). Instead of a test light, a volmeter could be used.

An incandescent test light draws about 200mA, a lot more than a voltmeter, so if there is a high resistance in a wire, a test light is much more likely to show it.

[Sylvain O.]

Hello Martin, 

Sorry it's me who sent the last post with the google account of my son (mrgryfiron). Is this the measure you want?

Sincerely Sylvain

[Sylvain O.]

je referai le test tel que demandé

[Martin Rubenstein]

No, Sylvain. Not to sensor ground, to heater ground.

4 wires:

sensor signal
sensor ground
heater power (you measured 13V on it)
heater ground

We need to check that the heater ground wire is connected to grond (earth, battery negative).

So, your connection to sensor ground is of no help.

I will diconnect my connector later and send you a photo of the connection to test.

[Martin Rubenstein]

Sylvain

Use a very small diameter pin to avoid damage to the connector. The contact shown, top right, is heater ground. If your heater ground is good, you will measure 12V between battery positive and this connector. You don’t need ignition: it connects to earth all the time.

[Sylvain O.]

 Hello Martin, I did the test that you ask. The pictures speak for themselves .... I measure 0.00v on the connector at the top right. I measure 12.58v on the connector at the top left. Looks like the opposite of you ?? Would there be a bad connection to the heating mass? The color of the threads are: Blue / red top right. Black top left. sylvain

[Martin Rubenstein]

Interesting!

Of course, my car is different to yours, so I am making assumptions!!!!

Do you have a wiring diagram?

If so, please check the colour of the wires at top right and top left.

Does top right go to ground/earth/battery negative?

Does top left go to the heater relay (and, through it, to battery positive)?

[Martin Rubenstein]

You said, in an earlier post, you measured 13V to the heater with the engine running (I think).

Was that top right or top left (or somewhere else :). ?)

[Sylvain O.]

Martin,
Yes the two wires of the male connector of the sensor O2 are of the same white color at the top right and left.
And the blue / red wire arrives at the relay, see the picture.

[Sylvain O.]

Top Left (TL) is black earth "1" on the diagram

Top Right (TR) is blue and red 12v "2" power supply on the diagram.

You must have a short circuit?

OK I will measure ;p

[Martin Rubenstein]

Not a short circuit - there doesn’t appear to be anything wrong with my circuit so I assume it’s working as designed. I’ll try and find an accurate wiring diagram or just check through the heater circuit and work it out - it’s only a couple of wires and a relay. The important thing is: is there 13V across TR and TL with the engine running. But only make this meaurement if you are sure it is not possible to short circuit the pins or leads.

[Sylvain O.]

Hello Martin, The motor is on and the O2 sensor heater is 14v. I drove the car after a reset of the default values ​​and a reset of the ecu. From the first minutes after starting the engine the exhaust smells of rotten egg. I made a capture of a log file throughout the driving. In attachment at your @mail the log file.

Sincerely Sylvain

[Martin Rubenstein]

The logfile has not arrived, Sylvain.

Could you attach it it the forum so the others can look at it and, if necessary, give their advice?

[Sylvain O.]

Martin,

the excel file is too big, I can only to publish txt version.

[leopold....@t-online.de]

Hi,
aic is too high at all temperatures- set aic down to zero!
lambda status is switching to zero after 3 minutes of run
lambdasensor not working- change sensor, check wiring
You should have seen this already on main page of mems-scan.

[Martin Rubenstein]

Sylvain

It’s strange you haven’t noticed the rotten eggs smell before. But it is not entirelly surprising because your car is running in open loop.

You never smelt it before?

You can read on google some reasons why you get this smell, for example, https://carfromjapan.com/article/car-maintenance/reasons-car-smells-like-rotten-eggs/


But I assume your lambda graph is still not working so you will be hoping the spare ECU will fix it.

One thing worth thinking about is: suppose the new ECU doesn’t fix it; what would I test then? And do it now rather than wait unitl then. I woukd want to check if current is flowing to the heater, just in case there is a problem at the connector and I’d check the lambda signal was reaching the ECU connector, but it would require a current clamp and pushing pins into the back of the ECU connector and I don’t recommend that.

So really, we are hoping the new ECU will fix the problem.


There is someone on the Mini Forum cally FlyingScot I think, and he knows how to get around the immobilisor block in the ECU, but you will need to send him a private message.

[Sylvain O.]

I lightened the file, I had created a graph for each of the values

[Martin Rubenstein]

Sylvain,

Leopold very kindly sent me a version of his software that allwed me to read your logfile.

I see what Leopold means. Your Inlet Air Temperature appears impossibly high Also it fluctuates in an impossible way. Impossible if it was working correctly, that is.

When I look at it closely, these fluctuations mirror RPM! So I don’t know whether your sensor is bad or if something is bad in the software, ot the ECU....

Can you make a logfile using Al Richey’s MEMSLogger, which comes with his MEMSAnalyser software. Then, with a different logfile, we can see if your Inlet Air Sensor circuit really is faulty. Also, then we will all be able to read your logfile!


I don’t know if a faulty Inlet Air Temperature would send tge car into open loop. Possibly. Certainly, it must be corrected - if it is faulty.

[Sylvain O.]

aic is too high at all temperatures- set aic down to zero! >> OK
lambda status is switching to zero after 3 minutes of run >> YES, with all sensor changed
lambda sensor not working- change sensor, check wiring >> I have a sensor collection, I will be able to open a store soon p; The wiring has been checked several times
You should have seen this already on main page of mems-scan. >> No comment ;p

[Sylvain O.]

For the last log file I used mems-scan-0.5.2-windows.
OK Martin I will do it again with MEMSLogger.
Yes in default mode this has always felt rotten egg.
Thank you for all your information.+
Between us I would prefer to understand the problem in its entirety and thus to make all the necessary investigations.
I am going to equip myself with an ampermetric clamp.

[leopold....@t-online.de]

"aic is too high at all temperatures- set aic down to zero!"

sorry this was a mistake in writing.
i wanted to write  IAC ( intake air control valve)

[Sylvain O.]

It's noted Leopold, thank you. For an operation I had dismantled the air box, I would have badly reconnected something? I will check, it seems that I did not have that before, I will compare with the old jounal files.

[Martin Rubenstein]

Sylvain

And I was looking at iac not iat!

Maybe there’s no need to make a new logfile, but next time you do make one, please use MEMSLogger so everyone can read it using MEMSAnalyser.

[Martin Rubenstein]

Your intake air temperature looks ok to me. It starts at ~22 and rises with time. Leopold may have aeen something I’ve missed.

As for a current clamp. The ones I use have a BNC connection to plug into an oscilloscope and I’m not familiar with the compkete units. But if you do get one, you want a low range one. You’ll want to be able to measure mA if you ever use it to look for small currents that drain the battery. Something like at least 40mA to 40A would be good. There’s a very useful trick you can do to magnify low currents: if, for example, you wrap the wire 5 times around the jaw, you anplify the current 5 times. So 30mA would appear as 150mA.

You put the jaw over only one wire. You wouldn’t be able to measure the lambda current: that would probably be in micro-amps. But at1A, the heater current would easily show.

A handy tool for anyone.

[Martin Rubenstein]