I know the gases obviously affect the element/probe on the sensor, but does this equate to a varied resistance across the circuit?

So when the engine is running, the lambda sensor puts a resistance across the circuit and the ECU interprets this resistance as either correct or incorrect?

I suppose it must vary between new and old cars, i'm guessing newer ones have the ability to adjust their fuelling and ignition but old ones just flag up a fault code that the sensor is reading something not right?

Cheers

Depends what type of sensor bud!

You have two wire, 4 wire, 5 wire etc.

A very basic one is a type of material, I'm thinking zinconium but I know that's wrong haha, which generates a voltage with exhaust temp which varies with the AF, this was adapted to be heated to get it working faster so the engine can go into closed loop fuelling faster after startup to meet the ever more demanding emissions criterias for newly built cars.

Depends what type of sensor you're referring too tbh, iv got an assignment somewhere I did on them lol I'll have a look for it!

^ not quite. But almost there, a normal 4 wire narrow band sensor acts more like a switch, it produces a very small voltage which is around 0.3 volts in a low state and around 0.8v in the high state, typically it switches at around 4.5 volts which is also the roughly the 'correct' mixture point for the engine, so in effectively all the ecu is seeing is a Boolean logic input of either 1 or 0 which will constantly switch at around 2 second intervals as the closed loop circuitry enriches or leans the fuel mixture as the exhaust gas oxygen contents fluctuates accordingly...

or to simplify it, the ecu adds a little fuel, then the sensor switches, so it takes away some fuel, the sensor switches and so on, it does this constantly in a closed loop... However the engine does not operate a closed loop operation 100% of the time.... typically the engine is only running closed loop at idle or part load throttle and cruising(motorway/constant speeds etc)... At other times it relies purely on 'mapped' settings for fueling such as when at mid to full throttle, overrun, cold engine etc.. There are also other auxiliary maps which look after temporary enrichment during warm up, sudden snap throttle opening, deceleration etc.

a wideband sensor, 5 wire type, works in a completely different way. These produce an variable output voltage between 0 and 5v, on these the output voltage is directly related to the air fuel ratio.. This adds a whole new level of complexity to the mapping of the ecu because they can vary the ideal air fuel ratio required at certain running conditions, for example they may make the engine speed which related to say 70mph deliberately run slightly lean to increase fuel economy, or they may make a turbocharged engine run slightly rich on full boost to keep the engine running slightly cooler

I hope this makes sense.. If you have any more specific questions ask away :-D

Depends what type of sensor bud!

You have two wire, 4 wire, 5 wire etc.

A very basic one is a type of material, I'm thinking zinconium but I know that's wrong haha, which generates a voltage with exhaust temp which varies with the AF, this was adapted to be heated to get it working faster so the engine can go into closed loop fuelling faster after startup to meet the ever more demanding emissions criterias for newly built cars.

Depends what type of sensor you're referring too tbh, iv got an assignment somewhere I did on them lol I'll have a look for it!

^ not quite. But almost there, a normal 4 wire narrow band sensor acts more like a switch, it produces a very small voltage which is around 0.3 volts in a low state and around 0.8v in the high state, typically it switches at around 4.5 volts which is also the roughly the 'correct' mixture point for the engine, so in effectively all the ecu is seeing is a Boolean logic input of either 1 or 0 which will constantly switch at around 2 second intervals as the closed loop circuitry enriches or leans the fuel mixture as the exhaust gas oxygen contents fluctuates accordingly...

or to simplify it, the ecu adds a little fuel, then the sensor switches, so it takes away some fuel, the sensor switches and so on, it does this constantly in a closed loop... However the engine does not operate a closed loop operation 100% of the time.... typically the engine is only running closed loop at idle or part load throttle and cruising(motorway/constant speeds etc)... At other times it relies purely on 'mapped' settings for fueling such as when at mid to full throttle, overrun, cold engine etc.. There are also other auxiliary maps which look after temporary enrichment during warm up, sudden snap throttle opening, deceleration etc.

a wideband sensor, 5 wire type, works in a completely different way. These produce an variable output voltage between 0 and 5v, on these the output voltage is directly related to the air fuel ratio.. This adds a whole new level of complexity to the mapping of the ecu because they can vary the ideal air fuel ratio required at certain running conditions, for example they may make the engine speed which related to say 70mph deliberately run slightly lean to increase fuel economy, or they may make a turbocharged engine run slightly rich on full boost to keep the engine running slightly cooler

I hope this makes sense.. If you have any more specific questions ask away :-D

Thanks a lot for the input guys :)

Well that's certainly very informative.

My thoughts were that if you wanted to decat and remove the lambda sensor, could you fool the ECU into thinking it was still there. However, from what you say, the constant variance in signal from the lambda to the ECU would mean you couldn't just send one constant 'tricked' signal to the ECU, because it relies on the variance to fuel correctly at different temperatures and throttle settings etc.

Correct?

I'm going to have to say, not exactly, to all the above in one way or another. Temperature has nothing to do with operation of the sensor, just when it produces a good signal. Oxygen is how it works and it is reading the amount in the exhaust gasses and a "reference" sample taken from outside air. It is made of a ceramic material coated with platinum and once it reaches operation temperature the material reacts to the oxygen difference between the exhaust sample and the reference sample. This reaction creates a small amount of voltage, roughly 0 - 1 volt for common sensors and 0 - 5 volts for wideband sensors, which the computer reads and adjusts the mixture accordingly.

It does not pulse or switch on/off but the voltage is a constant signal. It will vary as the fuel computer alters the fuel mixture, you can see this with a multi-meter attached, but it does not do any switching. The difference with the wires also does not change the sensor any. Old sensors were fixed in the exhaust manifold and the exhaust heat got them up to operational temperature, so one wire. Then cam three wire models with a built in heater to get them up to temperature faster and they could be mounted futher downstream, now we have three wires. The fourth wire one, even newer, is just an added ground wire. Also, wideband sensors are much more accurate and is why good mixture gauges use them.

If you decat the system you can still, and should, maintain the Lambda sensor. Just have a sensor bung welded in the exhaust, very cheap to have done or do. That way the fuel computer will still give you a good mixture when required and the system will still go "open-loop" when it needs to. The Lambda sensor really regulates the fuel mixture so you can run with a CAT, not because the system has a CAT.

I can't remember too much detail haha, I know they have ambient air as reference and flicker, and the average of that is the correct AFR

Some good info already posted.

If you want to decat then yes you need to maintain the lambda. As tuners, we would map out the lambda efficiency errors.

However the sensor itself still needs to be physically connected to the ECU, otherwise you get into the whole world of DTC DELETE's which really we would only ever do if totally necessary on some sort of a customer build where an ECU is being retrofitted to something that needs certain DTC's removing completely.

Sensors always need to maintain connectivity as the ECU logic is looking for those signals. However the mapping can tell the ECU to ignore certain signals.

I'm working on a new Mazda 6 diesel this afternoon and we're smashing the DPF through. But we have to ensure the sensors remain in tact as the ECU will expect to receive pulses from them.

Hope that makes sense.

Edited 31 July, 201214 yr by Irmscher

excellent thread, Clubpolo needs more of this, sensible questions with sensible answers.

back on topic my old lecturer used a simple analogy to 'dumb it down' for some pupils

he drew a line on the floor and then hopped from one side (rich) to the other (lean) and back and repeated.

the question was where did he spent the most time?

smack back over the line, at the perfect AFR.

the previous posts are much better.

Yeah tbh all the answers you need are here really - I learn lots of new things then forget within days haha :(

Yeah tbh all the answers you need are here really - I learn lots of new things then forget within days haha :(

where has your'e build thread gone dude?

It does not pulse or switch on/off but the voltage is a constant signal.

yes you are quite right, I used the word 'switch' quite cautiously, the output from a narrow band sensor will be a constant voltage of either 0.3 or 0.8v depending on the air fuel ratio, it doesnt switch as such, it just toggles between the 2 output voltages as the exhaust gases hovers over the stoichiometric point, apologies if this was misleading I was just trying to explain it in laymans terms.

I only make reference to a 4 wire sensor because that's what is relevant to a polo, 2 of the wires are purely for the heater element which heats the tip of the sensor, the other 2 wires relate to the actual sensor itself, the heater is not actually needed but it does give a stronger signal.. However most of the late vag group vehicles need the heater element to be working for the closed loop systems to operate.. On an early polo like a 6n the heater element part of the sensor is not wired to the ecu directly, it is fed from the fuel pump relay live (sometimes via it's own separate fuse also) which obviously means the heater is on all the time the fuel pump is alive.. but the ecu will still 'know' if the heater element is working or not by the amount of current the sensor is producing, for example a failed oxygen sensor heater will give you a fault code 'signal too low' and will force the ecu to run open loop even though the actual oxygen sensing component of the the sensor is still functioning correctly. On some of the early golfs mk3 16v etc the oxygen sensor has a relay which is controlled by the ecu specifically for switching on or off the heater in the oxygen sensor, I guess this is to prolong the life of the sensor by only heating it when necessary, but this relay is not fitted on a polo.

Bit off topic, but do boosted cars use the lambda for fuelling control on boost or not? As I know in "performance" type things they'll run richer than like 12:1 AFR, but only seem to have narrowband sensors, or is it just purely for an emissions control side of things at low load and idle etc?

Regards to OP question, if you have ODB2, pre and post cat lambda and decat, but leave the lambda in place is the ECU not going to throw a wobbly as it would think the cat isn't doing anything as there is no change from pre to post cat? I know that can be mapped out and such but people sometimes just decat.

Unless there is a way to fit a resistor or something, so the sensor still reads but slightly different to the pre cat sensor but will still alter to suit just marginally offset from the pre?

Thanks for all the info guys!

On my 6N I completely scrapped the ECU, so Lambda/Knock/TPS/Injectors all went too.

This all helps for future reference though, as I was under the impression on most cars if you remove the Lambda or it's knackered, the car can go into limp mode etc.

Surely this would happen if you left it connected but removed it from the exhaust stream?

I want to make this very very clear, I’m not trying to pick on you or cause you to stand out in any way, I’m just following the notice posted everywhere which is highlighted in pink. I wanted to state that up front as it important to understand and is often the case that people feel they are being attacked or made to look the fool on these forums. Then everything one party or the other says will be screened for the smallest of mistakes to pounce on. I will not do something like that unless extremely provoked which is not an easy task to do. I just talk cars and keep my ego in the closet to use someplace else.

. . . the output from a narrow band sensor will be a constant voltage of either 0.3 or 0.8v depending on the air fuel ratio, it doesnt switch as such, it just toggles between the 2 output voltages . . .

If you have ever tested the voltage output of a Lambda sensor and it reacted in the manner stated above, throw it in the trash. The voltage output should vary with the mixture throughout the entire range, 0 to 1 volt for narrowbands, and not just show only two voltage readings, constant or otherwise. Granted some test equipment might not be sensitive enough to read the constant voltage swings or can’t show the readings fast enough, but they are there. I use a digital Fluke multi-meter which is rather good quality and it still has a little trouble showing every little quick swing. That’s one reason those Christmas Tree Light air fuel ratio gauges are so bad. They can only react to major swings in voltage so they always show in the green unless the mixture goes super lean or rich. My point, in reference to the example above about the teacher, is that the voltage does act like his hopping back and forth over the center line, but with good equipment you car measure the complete path he takes during those hops and not just the high and low pause points.

. . . the heater is not actually needed but it does give a stronger signal..

The above quote and what was written after it fall into an area of new information since my last response which should be addressed (for the same reason as above). You are correct that the heater is not actually needed, but that statement is a little misleading for new aspiring mechanics. The sensor needs two things to function properly, ambient air to use as a reference and an operating temperature of about 300C. Before heated sensors came into use they were mounted in the exhaust manifold so they could reach that operational temperature quickly and stay there. With the heated version it was possible to locate the sensor away from the extreme heat and still allow it to function properly, and also later install a second to measure before and after the CAT. I tried years ago while doing some “Hot Rodding” of single wire setups, to move the non-heated sensor to a position away from the engine (down by the CAT as with newer models). It took a long time to heat the sensor and to provide a signal the ECU would accept and it would heat and cool with the different driving stages, i.e. idle or Highway or city driving. That caused the system to keep bouncing between open and closed loop operation and also giving bad mixture outputs. So yes, the heater is not required so long as the sensor can be installed in a location which gets the temperature up to where it has to be.

Just where the heating element gets its power from varies from car to car and year to year in many cases. Some get it from the fuel pump relay, some from the ECU power relay and some from the ECU itself. I ‘am sure there are other ways with some cars, like direct from the ignition switch, I can’t say as I don’t work on every car ever produced. But the where part is really not important so long as it gets the voltage and it stays hot. Yes, the ECU on most models will know or sense if the signal voltage from the Lambda sensor is wrong and most will then switch to a fixed fuel metering program (often a little rich to be safe). It does not know if the heating element is bad or if the wire has broken or if the sensor itself has gone bad, it just knows that the voltage reading is incorrect. On the sensor relay you quote, mk3 16v models, I never found one to have such a relay or show one in the wiring diagrams, but I don’t live in the UK so there might be a difference. I am somewhat new to Polo models and have only worked/owned older ones (to 94), plus I drive on the right side of the road (right as in direction not "correct") so there are always some things which are different.

Bit off topic, but do boosted cars use the lambda for fuelling control on boost or not? . . . if you have ODB2, pre and post cat lambda and decat, but leave the lambda in place is the ECU not going to throw a wobbly as it would think the cat isn't doing anything as there is no change from pre to post cat?

I don't do boost but do know that engines that do have superchargers/turbochargers also use Lambda sensors in many cases. Just how they are programed to respond I can't say, but they do use them.

That's good heads up on the pre/post CAT issue. I believe it would cause an issue if the readings did not vary between the two. I remember reading somewhere about this problem and how it was addressed. It also discussed problems with fault codes if the sensors were just deleted. As I don't work on cars with these setups I can't say just where or what was discussed, but it was about the dual sensor delete issue.

The post cat sensor diagnoses that the catalyst is working.

If the two sensors are seeing the same gases (because there isn't a cat brick) then you will get a fault code indicating a failed catalyst.

For decat this either has to be coded out or some extra components are required to fool the ECU.

Edited 1 August, 201214 yr by Ben S

Most high performance cars will run 1 wideband/bank and 1 narrowband (mid or post cat) / bank. You'll also find most recent sensors are now pressure compensating too.

Nick- more and more OEMs are aiming to run near to lambda 1 at peak power as they can- this has several obvious problems, namely component protection and thermal management. Most engine management systems will stiwch from closed loop to open loop at wide open throttle, and run of fixed values, with corrections for component portection coming from other sensors (e.g. Knock, EGT).

If you have ever tested the voltage output of a Lambda sensor and it reacted in the manner stated above, throw it in the trash. The voltage output should vary with the mixture throughout the entire range, 0 to 1 volt for narrowbands, and not just show only two voltage readings, constant or otherwise. Granted some test equipment might not be sensitive enough to read the constant voltage swings or can’t show the readings fast enough, but they are there.

sorry to labour this point but you are wrong

I use an oscilloscope for readings of this nature, a narrow band sensor has 2 distinct logic states, there is a certain amount of crossover from rich to lean at around the 4.5 volt crossing point but the output is non-linear so cannot be relied upon to give an accurate air fuel ratio reading... The only important bit as far as your ecu are concern is the toggling from state to another.

Narrowband lambda voltage output (0-800mV approx) is proportional to the oxygen content of the exhaust, measured with reference to ambient air. There's some variation with temp and pressure too, but older systems ignore this.

However, this proportionality is distinctly non-linear such that the voltage either side of stoichiometric air-fuel ratio is near full-scale or zero. It's the non-linearity that makes a narrowband sensor no use for tuning or for implementing closed-loop fuelling at any operating points where anything other than stoich/lambda 1 is required. So anyway, whilst it gives the appearance of a device with a decidedly digital output it is very much analogue!

This is a random diagam I found on the net to illustrate the point, it's very similar to what's in the Bosch Automotive Engineer's handbook.

The exact voltages and waveform you measure on a vehicle will depend on how the ECU is controlling the fuelling, how the lambda's wired up and where you're measuring it (e.g. which ground reference you're using...).

I like this thread, :)

Andy, yes you are right it's not a digital signal its all fuzzy but circuitry in the ecu smoothes this out and applies a certian amount of hysteresis to it...and it is non linear... This is exactly my point, the only way you can accurately keep the engine in it's right afr band is by constantly cycling between rich/lean every second or so because the voltage from the sensor wanders about when it's around the correct afr point... Pretty much all multipoint injection systems use this method when a narrow band sensor is used, motronic, magnetti-marelli, simos etc etc.

So long as things remain civil I have no issues with discussing things. I do have to say that your post of the oscilloscope printout and the old graph Andy posted show exactly what I was saying. My original post in this string, if you read back, was to sort of correct things that were said which were inaccurate. This is the explaination you first did above.

. . . so in effectively all the ecu is seeing is a Boolean logic input of either 1 or 0 which will constantly switch at around 2 second intervals as the closed loop circuitry enriches or leans the fuel mixture as the exhaust gas oxygen contents fluctuates accordingly...or to simplify it, the ecu adds a little fuel, then the sensor switches, so it takes away some fuel, the sensor switches and so on,

The oscilloscope and the Lambda graph both show a curved line, analog as has be stated, and not an on - off digital line which would be as first explained. I am sure it was not intended to look like it was a digital signal but whit the 2 second thing thrown in it really did look to be said that way. It is also somewhat obvious that identifying the steady change in voltage would be hard to read with most backyard equipment. It was only important to me that people understood that the Lambda output was not as you first said it was. You yourself even now indicate it is not as you first stated, so for that part “case closed” as far as I am concerned.

Now about that Lambda heater relay. Have to say I was wrong and I did find one listed in a diagram. Thought it was the Motronic relay at first but just for kicks I crossed the number yep, it lists it as the Lambda heater relay. The ECU does do the grounding side but I really doubt it cycles in on and off as that would wear the sensor out faster I think.

Andy, yes you are right

I'd hope so - Powertrain Electronics is my job!