lean on full load

[350matt]

So then chaps bit of a puzzler
as you may know my car is slightly modified with a sports cat, toyo catback, cosworth air filter and mazdaedit ECU tune ( as carried out by me) and RRP coils
now this was all working well and my last map change was back in february and the car has done about 3K miles since then

I've recently noticed my long term fuel trim figures getting greater than -10% which is bad news in my book as on my Ecu tune the LTFT would sit + / - 3% also the reported AFR is about 13.5- 13.8 on full load which is should be 12.8 at its leanest ( map AFR is 12.8 at this point)

So it knows its lean and yet its taking fuel out ?

So tonight I've tried another lambda sensor ( original Denso part) and also checked for any exhaust leaks before the O2 sensor and checked fuel pressure is OK with the engine running at idle ( 4 bar)

the test drive post lambda sensor swap 'felt' better but AFR still shows up lean on the logger

thoughts?
Fuel pressure needs checking under full load is my next attempt
and I tried cleaning the MAF but that was clean and may no odds anyway

[ChrisHolmes]

I know you did but did you calibrate the sensor before fitting it?

[350matt]

No I did not
wasn't aware that was a thing - hows that done then ?
I just unscrewed the old and fitted the new - I should point out the replacement isn't new its had one careful owner but I was assured it worked OK

not impossible I've just replaced one duff sensor with another of course

[ChrisHolmes]

Did the "new" sensor come from a 231?

Just checking as the 192 sensor is a different specification, though it looks and fits the same.

[ChrisHolmes]

I think I may be talking rubbish as it must be my Innovate gauge that I am calibrating, not the sensor. Hopefully some with more knowledge will be able to help. A “rich” signal would cause the issue I think.

[mrspiller]

I think I may be talking rubbish as it must be my Innovate gauge that I am calibrating, not the sensor. Hopefully some with more knowledge will be able to help. A “rich” signal would cause the issue I think.

Yeah that’s your wide band I think Chris


Sent from my iPhone using Tapatalk

[Milos_Balunovic]

the car is not running in closed loop above 4.5k rpm or if above 70% throttle or something.

The ECU is only doing open loop (what it is preset to do) and doesn't listen for the lambda while running, but only "learning". as most already mentioned mazda ECU is "smart" and will sometimes overwrite some numbers based on best knowledge.

Also 12.8 is very rich. why are you aiming that low stoic ?

[MadTaz]

the car is not running in closed loop above 4.5k rpm or if above 70% throttle or something.

The ECU is only doing open loop (what it is preset to do) and doesn't listen for the lambda while running, but only "learning". as most already mentioned mazda ECU is "smart" and will sometimes overwrite some numbers based on best knowledge.

Also 12.8 is very rich. why are you aiming that low stoic ?

The ECU will monitor the lambda up to 4999rpm based on load condition and up to between 60-70% based on throttle position and rpm.

The ECU does not overwrite anything it just applies a correction based on what the MAF is seeing against what the O2 is recording, it does this correction as a percentage called Long Term Fuel Trim. Long Term fuel trim is a correction factor for variances in the MAF, fuel injectors, and engine characteristics (scaling for injectors and MAF).

The issue you are having Matt is that the car is learning trim on a small proportion scaling i.e. -10% LTFT but then applies that to the complete scale when moving to Open loop as this is seen as best fit.

If this has happened recently and not through a change in the map it could several reasons why:

MAF is dirty;
Vacuum leak;
Engine is just rebuilt and bedding in so characteristics are changing
MAF scaling is out
Fuel injectors are tired, dirty or failing
coils are weak, one has failed or maybe not connected nicely. (yes even RRP coils can fail) .

I'd start by checking the easy stuff and then go for some rescaling of the MAF.

finally 12.8 is not rich for a rotary on full load, you could push it a little more lower down the rev range but I wouldn't up top.

Hope that helps

[ChrisHolmes]

the car is not running in closed loop above 4.5k rpm or if above 70% throttle or something.

The ECU is only doing open loop (what it is preset to do) and doesn't listen for the lambda while running, but only "learning". as most already mentioned mazda ECU is "smart" and will sometimes overwrite some numbers based on best knowledge.

Also 12.8 is very rich. why are you aiming that low stoic ?

The ECU will monitor the lambda up to 4999rpm based on load condition and up to between 60-70% based on throttle position and rpm.

The ECU does not overwrite anything it just applies a correction based on what the MAF is seeing against what the O2 is recording, it does this correction as a percentage called Long Term Fuel Trim. Long Term fuel trim is a correction factor for variances in the MAF, fuel injectors, and engine characteristics (scaling for injectors and MAF).

The issue you are having Matt is that the car is learning trim on a small proportion scaling i.e. -10% LTFT but then applies that to the complete scale when moving to Open loop as this is seen as best fit.

If this has happened recently and not through a change in the map it could several reasons why:

MAF is dirty;
Vacuum leak;
Engine is just rebuilt and bedding in so characteristics are changing
MAF scaling is out
Fuel injectors are tired, dirty or failing
coils are weak, one has failed or maybe not connected nicely. (yes even RRP coils can fail) .

I'd start by checking the easy stuff and then go for some rescaling of the MAF.

finally 12.8 is not rich for a rotary on full load, you could push it a little more lower down the rev range but I wouldn't up top.

Hope that helps

Some great knowledge shared here, many thanks Carl I am sure lots of people will also appreciate this insight.

[Milos_Balunovic]

...
finally 12.8 is not rich for a rotary on full load, you could push it a little more lower down the rev range but I wouldn't up top.
....
Hope that helps

I oversimplified the fact about what the ecu does.

But also it tend not to listen to some of the tables in M.E. as people would anticipate ect..

Also ecu will not simply adjust the fuel trim based on the lambda as it doesn't know which of the two sensors is missreading. And than it interpolates...

[350matt]

So bit of an update I've done some continuity testing of the front O2 sensor signal wire back to the ECU and found it had a high resistance ( 350+ ohms) so peeling back the tape covers I've found the wire insulation had cracked in a couple of places, cutting out this damaged section of wire and splicing in a new bit and then heatshrinking / re-insulatiing the lot then got the wire back down to 3 ohms
taking for a drive this evening and it seemed / pulled better - annoyingly my tactrix logging was a bit intermittant for some reason so I don't have hard data as yet to show if its solved it or not

so we'll see
it could probably do with new plugs as they've been on the car for about 12K miles I think

Also Milos I don't understand your comment about lambda sensors? surely the front semi wide band is the one that does the fuel control and the post cat item just does monitoring / switching from lambda 1
and that 12.8:1 AFR isn't my part load target its full load see here

rx8 fuel map.jpg

[Milos_Balunovic]

Ecu is giving fuel based on mass of air consumed. And than double check if the target afr are reached to adjust the fuel trims. The ecu doesn't know which of the two sensors is missreading. The MAF of the Lambda (wide band)

[MadTaz]

The narrow band sensor, the one that is at the rear is only really accurate at stoich and anything away from stoich it can only give an indication of 'too rich' or 'too lean'.

As Milos has said it does affect the commanded AFR and I have witnessed this through customer logs, although I'm unsure how much authority the sensor has within the calculation and whether this actually feeds back into the fueling calculation at all (based on logs and experience) , it should do and this is how other cars are set up, perhaps someone can post up any information on this would be interesting to see.

Practically, it makes no difference if your front O2 sensor is working as it should, there is no loss of function, if the feedback as witnessed does make it back into the fuel calculation then if you are targeting anything other than stoich while in Closed loop then its probably better just to delete the rear sensor due to its limitation/inaccuracies.

Matt,

I'm assuming that table is just to help you scale the MAF?

[350matt]

Hello Karl

thanks for the info
I wonder if its worth removing the rear O2 and turning off all the DTC codes as while I'm targeting 14.7 / stoich at the moment it'd be nice to run it a bit leaner 15.5 ish on low load cruise 60mph ish

and yes I've rescaled the MAF to achieve those target lambdas I've not touched the VE table or injector linerisation

[MadTaz]

Hello Karl

and yes I've rescaled the MAF to achieve those target lambdas I've not touched the VE table or injector linerisation

That is not the ideal AFR target range for a fuel table, I'd be inclined to add a bit of fuel up top, you can remove a bit down low if you wanted.

[Milos_Balunovic]

As the revs increase, so the burn efficiency reduces, e.g. less of the fuel gets ignited.
This happens with all ic engines but the effect is more pronounced on rotarys as they rev higher and have a complex and moving combustion chamber.
This means that you have to increase the AFR to compensate.
Failure to do so reduces power and increases the risks associated with lean running.

Sent from my SM-G398FN using Tapatalk

[Milos_Balunovic]

The narrow band sensor, the one that is at the rear is only really accurate at stoich and anything away from stoich it can only give an indication of 'too rich' or 'too lean'.

As Milos has said it does affect the commanded AFR and I have witnessed this through customer logs, although I'm unsure how much authority the sensor has within the calculation and whether this actually feeds back into the fueling calculation at all (based on logs and experience) , it should do and this is how other cars are set up, perhaps someone can post up any information on this would be interesting to see.

Practically, it makes no difference if your front O2 sensor is working as it should, there is no loss of function, if the feedback as witnessed does make it back into the fuel calculation then if you are targeting anything other than stoich while in Closed loop then its probably better just to delete the rear sensor due to its limitation/inaccuracies.

Matt,

I'm assuming that table is just to help you scale the MAF?

I was actually talking about the wide band primary lambda sensor. but you chimed in on what could be the feedback / correction sensor.

what I wanted to say is that the ecu cannot know why it is seeing for example 13:1 and target is 12:1 as it could be:
-injector have less flow, - Fuel trim correction strategy
-Lower pressure on fuel pump - Fuel trim correction strategy
-maf is reading less air than there actually is
-lamba is misreading and the actual value is 12.0


for having a reading of example 13:1 and target is 14:1
-failing ignition (not all fuel is ignited/burned) - Fuel trim correction strategy
-leaking injectors (oil and or fuel) - Fuel trim correction strategy
-worn oil seals - burning oil - Fuel trim correction strategy
-sensor is reading richer than it actually is
-maf is reading more than it actually is

because there are so many parameters the ecu cannot know how to correct the fuel and which sensor is misreading. is it maf, or lambda (wide band) or is it both of them. this is why I believe the correction is not as simple (in open loop) as:
1. afr to low or high
2. apply short therm fuel trim
3. apply long therm fuel trim
4. (if ECU has that function) small changes to the base map and "auto tune"

it could use he secondary lambda sensor to double check which of the two is misreading by having a stoich at both sensors and use that as "calibration"

I haven't dived in to the base code of the rx8 ecu so I really don't know how they behave. for that I would need at least a .bin file from unmodified, ECU dump file and WinOLS. but I'm to tired of doing mapping by creating all the tables my self. Mazda edit / versa tuner and such will not show all the switches and possibilities and how they are correlated.

[Milos_Balunovic]

I believe the best would be to have an independent AFR gauge with a good quality lambda (wide band) sensor on it and not rely on reported data through OBD protocol by ECU !

[warpc0il]

I'm not sure what you're trying to achieve with your remapping but some of the physics can be counter-intuitive.

Most people would assume that the less fuel you input, the better the fuel consumption.
This has had an impact on the way that tests are conducted on new cars, run in very controlled artificial conditions.

This results is nearly all cars being produced with a dip in their torque curve around mid-range revs, which happens to align with the critical tests.

Most good remaps aim to remove this dip, as well as generally increase the area under the curve, with or without increasing maximum power.

The result is making the car so much more drivable and responsive in the mid-range and, in the hands of an experienced driver, this then results in better fuel economy under real-world conditions.

Higher AFR, in the right places, actually improves fuel economy while having the benefit of protecting the engine - which is all counter-intuitive.

Obviously, you can take this too far and the effects of throwing too much fuel in can create its own problems, which is why engine mapping is part science and part artistry.

Sent from my SM-G398FN using Tapatalk

[Milos_Balunovic]

I don't know if the comment was aimed at my comments or 350matt's ref lean on highway speed.

But I can further add. the most critical is 14.7 above it the engine starts to cool of again from lack of fuel. engineering explained did a great video on it. I also Agree with saying lean is higher than 14.7 and less than that is rich. ofc richer and leaner is relative


Also most rx8 tuners say that the stock map is to rich in most situations. I am not an expert on this topic.

[350matt]

OK so got some good data this morning and the d@mned thing is still lean on full load.......
this is despite fixing the sensor wiring
so plugs next ? or just rescale the MAF ....

as Milos says I'm tempted to fit my standalone wideband

[warpc0il]

Can you confirm that "lean running" with a plug chop?

It's a bit old skool in these days engine management but it gives a true indication of what's actually going on

Sent from my SM-G398FN using Tapatalk

[Milos_Balunovic]

Matt, try to verify your AFR readings are in fact correct. ie use an external AFR gauge that you stick in the back silencer box if you cant find a better solution.

WarpcOil that would be an expensive plug shop, as he has to look in to the very bottom of the ceramic (cut them open)

[350matt]

bit of a misnomer Milos
a plug chop just means you accelerate full pelt and the chop / cut the ignition and coast the car to a layby and then pull the plugs out

I think I'll try rescaling the MAF for the drive home and then look into fitting my wideband

one other thing does pulling the 'room' fuse reset the fuel adaptions ?

[Milos_Balunovic]

no, I don't agree, its not a misnomer, as you are interested in the very bottom part of the ceramic, which you will hardly see with an uncut plug espec on an rx8.