Electronic Fuel injection Lambda sensors et al
Discussion
Id like to have a better understanding of this. I can follow the ECU taking readings from various engine speed & load sensors then looking up a predetermined map for the fuel setting but not why it needs to monitor the exhaust as well. If the ECU can determine fuelling requirement from just exhaust readings why have all the other stuff too? Also what is open & closed loop - what are wide and narrow band sensors - any thoughts gratefully received
You need a base point for the fueling to start with from the sensor inputs and mapping. The lambdas then simply trim this amount depending on the remaining exhaust oxygen as a reference point. You dont want to have to make large corrections, as you are always playing catch up, much better to get the mixture right in the first place from a good map, rather than correct it once its wrong. Narrow band sensors only work around the optimum point for minimum pollution, and simply switch very rapidly when this point is reached, (around about 0 to 1 volt) know as Lambda 1 or 14.7: air to fuel ratio and only give very small voltage changes outside this design range. A wide range sensor is designed to give a much more linear voltage response against a wider range of mixtures, so the ecu can then determine the required mixtures for both lean cruise running at say 15:1 air to fuel ratio (best economy) and full power at say 12.5:1 air fuel ratio. Modern emission laws means engines may never reach these optimum points as some engines are clamped at 14.7:1 to keep the emissions low, but if you are remapping you engine for the best economy and power results you need wide range sensors. The closed loop status is when you are using the lambdas to clamp the mixture to the required value, and open loop is when you simply rely on the mapping data to get the mixture right, which may be well outside the range you get from a narrow band sensor. The switching point between closed and open loop operation is tweeked so the emission laws can be met on narrow band sensors, as cars are not gas tested under full load and RPM for MOT purposes, so the engine can be allowed to run richer than lambda 1 in open loop to achieve full power. If you have wide band sensors you can run closed loop all the time, but cars running closed loop all the time on narrow band sensors will never develop maximum power.
Edited by blitzracing on Saturday 20th August 19:00
smac said:
Could the map not be accurate enough to start with that the lambda is redundant?
Depends how many sensors are being used.Some simple sytems only monitor throttle position so the map will only be correct at a certain combination of air temp, coolant temp, air pressure, manifold vacuum etc.
If lots of sensors are used then the engine could operate closer to the map.
The Lambda sensor is checking for unburnt oxygen in the exhaust gas. If it finds some then the mixture is not at 14.7:1. In closed loop the ECU will use this to correct the fuel supplied.
As a narrow band Lambda can only see the 14.7:1 switching point that is the only point it can be truely accurate. A broad band can see a much wider range so when the map says it wants a rich mix to make power the ECU can still Close Loop because the sensor can show it the 12.5:1 (say) it is looking for.
If simple and cheap is what you are looking for you can map an engine on a dyno using the dynos sensors. The map will not be responsive to changing conditions but the engine will perform in a reasonable fashion.
Steve
They can be pretty close, but the closer you can get to perfect the better. Both from the emissions point of view and engine efficiency. For example, my 20 year old Sierra doesn't have a catalyst / sensor, and passed it's MoT with emissions that were considerably below those required of a catalyst car. I couldn't believe it either, and had to ask on here to make sure what I was seeing was right!
smac said:
Could the map not be accurate enough to start with that the lambda is redundant?
ps dont know how the angry smily thing appeared - or indeed how it has now gone
There are too many variables that are not able to be compensated for e.g. fuel blend, fuel temperature (though some cars have fuel temp sensors) and as components age the calibration can change. The switching point of a narrowband sensors doesn't really change over it's life, it simply becomes slower to respond as it ages.ps dont know how the angry smily thing appeared - or indeed how it has now gone
If you use an airflow meter for measuring the volume and density of the incoming air its perfectly possible to maintain a reasonably accurate air fuel ratios without lambda sensors, as when the engine wears the airflow alters to match, unlike mapping against the throttle pot or inlet pressure. As the AFM becomes the major sensor for fuel metering, all you then have to do is trim the mixture depending on say the throttle pot movement, rpm, fuel and engine temp. Problems arise however if you modify the engine from when it was first mapped, or use a long duration cam that breaths poorly at low RPM where the AFM output can become inaccurate due to reversion pulse during the valve overlap period.
A problem will also arise if you try and run a catalyst with a fuel map that's tuned to the engines best requirements, the amounts and types of pollutants exceed the catalysts ability to clean things up, and in some cases will cause the catalyst to overheat, hence the need to clamp the mixture with lambda feedback when emissions are the major consideration.
As a matter of interest even narrow band sensors give a voltage change well outside the sharp switching range at lambda 1, but its only a few hundred millivolts for quite large mixture changes. Its possible to enhance these small voltage changes to give an approximate display of mixture, and from my experiments it seems to be possible measure between about 12:1 to about 15.5:1 AF ratio. The big downside to this is you have to calibrate probes individually as you are using the probes outside their design range, so the results are not consistent across new and old probes.
A problem will also arise if you try and run a catalyst with a fuel map that's tuned to the engines best requirements, the amounts and types of pollutants exceed the catalysts ability to clean things up, and in some cases will cause the catalyst to overheat, hence the need to clamp the mixture with lambda feedback when emissions are the major consideration.
As a matter of interest even narrow band sensors give a voltage change well outside the sharp switching range at lambda 1, but its only a few hundred millivolts for quite large mixture changes. Its possible to enhance these small voltage changes to give an approximate display of mixture, and from my experiments it seems to be possible measure between about 12:1 to about 15.5:1 AF ratio. The big downside to this is you have to calibrate probes individually as you are using the probes outside their design range, so the results are not consistent across new and old probes.
You really need to split "disturbances" to the mean that lead to fuelling varriations into 4 main sections:
1) Component tollerances - important if you are going to build a high volume product. Typically EMS sensor components will be specified to be within approx 2% when new.
2) Component ageing - Different parts age differently, depending on the useage and cycles, a MAF meter may drift up to 5% over it's life
3) Build tollerances - Everytime you build something, it will be a little bit different, as you have to have some production tollerance (otherwise you'd never build anything in volume)
4) Environmental conditions beyond those that the manufacturer can realistically calibrate for. i.e extreme altitude, hot/cold, and cruically fuel type and oxygen content.
These factors add up, and the problem becomes "tollerance stack up". For example, say your MAF sensor reads 2% high, and your fuel pressure regulator is also 2% high, and maybe your engines airflow is 2% down (due to say slightly incorrect cam timing (but within production tollerances) Now you are looking at a 6% fuelling error, which is MASSIVE. (in reality, not all the components will have a tollerance on the same side of the mean, but you have to protect for the worst case)
In order to meet the tight modern emissions and fuel economy targets, engine AFR will need to be within <0.1% of target to maintain the certified performance. (and conformity of production tests are carried out by the certification authorities to check the OEM's are building cars that continue to meet the stds). So trying to control AFR with sensor tollerances possibly in magnitudes of several % is impossible. Hence all modern cars use a Lambda sensor to "feedback" the actual delivered exhaust AFR, and then "adapt" the fuel delivery quantites to meet the necessary tight targets. (a second Lambda probe is also used to monitor the catalysts health, also crucial to emissions performance). This "adaption" is a critical part of the engine managements strategy, and the reason that cars are run over a "preconditioning" cycle prior to any exhaust emissions testing, to allow the adaption to occur and the system to get to the exact AFR target.
Also, for a 3way catalyst to operate at maximum conversion efficiency, the AFR must oscillate either side of stoichiometry (Lambda 1), this is achieved by using a "jump and ramp" strategy that imposes small extra fuel mass additions and subtractions onto the steady state base fuelling to perturbate the exhaust AFR. On a "switching type" narrow band sensor, the sudden voltage jump as the exhaust AFR crosses stoichiometry is used to trigger these oscillations.
An OEM powertrain calibration program will spend up to 3 years and £25M testing the calibration under a huge number of climatic conditions and with the widest possible range of "limit tollerance" parts (parts deliberately manufactured to sit at the end of their likely tollerance band). The aim is always the same, to ensure that the base "open loop" calibration is a good as possible, because all closed loop control systems will be more accurate and stable when the closed loop control has the minimum to do.
Modern EMS systems now have multiple degrees of freedom (think variable cam timing x2, electronic throttle, EGR, varriable fuel pressure, ignition angle, etc etc) so to achieve a base calibration within a sensible time and with a sensible budget, OEMs now use "Design Of Experiments" (DOE) techniques to mathematically project a multidimensional model surface across all the interdependant parameters. Combined with the latest "physics based" EMS strategies, this delivers fuel economy and tailpipe emissions at levels that were consitently thought to be "impossible" as little as 5 years ago
1) Component tollerances - important if you are going to build a high volume product. Typically EMS sensor components will be specified to be within approx 2% when new.
2) Component ageing - Different parts age differently, depending on the useage and cycles, a MAF meter may drift up to 5% over it's life
3) Build tollerances - Everytime you build something, it will be a little bit different, as you have to have some production tollerance (otherwise you'd never build anything in volume)
4) Environmental conditions beyond those that the manufacturer can realistically calibrate for. i.e extreme altitude, hot/cold, and cruically fuel type and oxygen content.
These factors add up, and the problem becomes "tollerance stack up". For example, say your MAF sensor reads 2% high, and your fuel pressure regulator is also 2% high, and maybe your engines airflow is 2% down (due to say slightly incorrect cam timing (but within production tollerances) Now you are looking at a 6% fuelling error, which is MASSIVE. (in reality, not all the components will have a tollerance on the same side of the mean, but you have to protect for the worst case)
In order to meet the tight modern emissions and fuel economy targets, engine AFR will need to be within <0.1% of target to maintain the certified performance. (and conformity of production tests are carried out by the certification authorities to check the OEM's are building cars that continue to meet the stds). So trying to control AFR with sensor tollerances possibly in magnitudes of several % is impossible. Hence all modern cars use a Lambda sensor to "feedback" the actual delivered exhaust AFR, and then "adapt" the fuel delivery quantites to meet the necessary tight targets. (a second Lambda probe is also used to monitor the catalysts health, also crucial to emissions performance). This "adaption" is a critical part of the engine managements strategy, and the reason that cars are run over a "preconditioning" cycle prior to any exhaust emissions testing, to allow the adaption to occur and the system to get to the exact AFR target.
Also, for a 3way catalyst to operate at maximum conversion efficiency, the AFR must oscillate either side of stoichiometry (Lambda 1), this is achieved by using a "jump and ramp" strategy that imposes small extra fuel mass additions and subtractions onto the steady state base fuelling to perturbate the exhaust AFR. On a "switching type" narrow band sensor, the sudden voltage jump as the exhaust AFR crosses stoichiometry is used to trigger these oscillations.
An OEM powertrain calibration program will spend up to 3 years and £25M testing the calibration under a huge number of climatic conditions and with the widest possible range of "limit tollerance" parts (parts deliberately manufactured to sit at the end of their likely tollerance band). The aim is always the same, to ensure that the base "open loop" calibration is a good as possible, because all closed loop control systems will be more accurate and stable when the closed loop control has the minimum to do.
Modern EMS systems now have multiple degrees of freedom (think variable cam timing x2, electronic throttle, EGR, varriable fuel pressure, ignition angle, etc etc) so to achieve a base calibration within a sensible time and with a sensible budget, OEMs now use "Design Of Experiments" (DOE) techniques to mathematically project a multidimensional model surface across all the interdependant parameters. Combined with the latest "physics based" EMS strategies, this delivers fuel economy and tailpipe emissions at levels that were consitently thought to be "impossible" as little as 5 years ago
Edited by anonymous-user on Tuesday 23 August 21:34
blitzracing said:
If you use an airflow meter for measuring the volume and density of the incoming air its perfectly possible to maintain a reasonably accurate air fuel ratios without lambda sensors, as when the engine wears the airflow alters to match,
How would that compensate for e.g. injectors gumming up a bit, or the FPR spring relaxing with time and dropping the fuel rail pressure a little bit?
Gassing Station | Engines & Drivetrain | Top of Page | What's New | My Stuff