AP1 S2000 Lambda error, Air Per Cylinder question

[mrdjeezy]

Hi everyone,
So I am switching from an eManage to a Link G4x.  I installed the Link, got the idling sorted out.  Target lambda is 1, actual ecu lambda is right there.  Sub 1% correction - all that is fine.

My questions:
The lambda on the gauge is slightly off.  The ECU is reading roughly .126 higher than the gauge.  The gauge is an AEM UEGO 30-4110.  What could be causing that slight difference?

Second question, the Air Per Cylinder Estimate is weird.  If I adjust the settings to get it to .10-.12g/cyl then the Lambda is overcorrecting and the car is idling weirdly.  However, if I just forget about that value and I focus on getting the lambda and the ignition angle right, the car idles great but the Air per Cylinder Estimate hovers at around .06-.07.  Not sure where to go from here.  It seems like, if I adjust to get the Idle RPM+Lambda down to reasonable numbers, the APCE is off.  When i get the APCE right, the idle and lambda is off.   What can I do to address this, or should I not even worry about it if the car is idling fine?

My setup - everything stock except: 255lph pump, 550cc injectors, Turbo@9psi, test pipe

Thanks in advance

[Adamw]

Those old AEM widebands are pretty crap you will always have a difference between the gauge and the ecu. Two problems with them - 1) there is no analog reference ground, so any difference in potential between the power source of the wideband and the ECU reference ground will offset the voltage.  Connect the AEM black wire to the engine block will usually give the least offset and variation but wont cure it.  When electrical load on the battery cables changes (fans, head lights turning on etc) will change the Lambda.  2) even if they had a proper analog ground, the analog output on those early one seems quite variable - even with the sensor unplugged one day you will get say 2.5V coming out of the signal wire, then the next time you test it it will be 2.35V...

For the air per cyl est, are you using modelled fuel equation?

 

[mrdjeezy]

Good to know, I had no idea.  I'll have to switch that up then for my own sake.

I'm using a modeled fuel equation and for the equation load source I'm using MAP.

[Adamw]

Assuming MAP and charge temp are reading correct, if the air per cyl estimate is smaller than you expect it should be then that generally means the engine is receiving more fuel mass than the ecu believes it is injecting based on the metrics it has been given to calculate fuel mass. - so the net effect is to get the air/fuel ratio correct you have had to tell the engine there is less air going in (ie you have smaller VE numbers in the fuel table than realistic).   

There are many possibilities where that error is coming from, fuel density settings, stoich ratio setting, injector flow rate or dead times, fuel pressure etc. If the air mass estimated is reading half what you expect it suggests there is a fairly significant error coming from somewhere.   

It would generally not cause too much of a tuning issue but some of the fuel model possibly wont work as well as it should if the estimates are a long way off.   

If you attach a copy of the tune and short PC log I will take a look.  

[Confused]

Add to that, if the value the ECU is thinking it's seeing from the rubbish old AEM sensor is different to what is on its own display, then the engine isn't actually running at the lambda it thinks it is.

I've just swapped one of these out on a car here for the exact same reason - the Link CAN Lambda was a breeze to install (I added the 22uF capacitor as per the alternative wiring diagram) and it's all substantially better now! I'm no longer fighting the readings on a daily basis!

[mrdjeezy]

On 12/31/2022 at 6:22 PM, Adamw said:

Assuming MAP and charge temp are reading correct, if the air per cyl estimate is smaller than you expect it should be then that generally means the engine is receiving more fuel mass than the ecu believes it is injecting based on the metrics it has been given to calculate fuel mass. - so the net effect is to get the air/fuel ratio correct you have had to tell the engine there is less air going in (ie you have smaller VE numbers in the fuel table than realistic).   

There are many possibilities where that error is coming from, fuel density settings, stoich ratio setting, injector flow rate or dead times, fuel pressure etc. If the air mass estimated is reading half what you expect it suggests there is a fairly significant error coming from somewhere.   

It would generally not cause too much of a tuning issue but some of the fuel model possibly wont work as well as it should if the estimates are a long way off.   

If you attach a copy of the tune and short PC log I will take a look.  

Here is my tune, only log I have on my pc is of me adjusting the idle and it's 38minutes long - not worth sending.  I'll get you a short log tomorrow.

s2000turboitle.pclx

17 hours ago, Confused said:

Add to that, if the value the ECU is thinking it's seeing from the rubbish old AEM sensor is different to what is on its own display, then the engine isn't actually running at the lambda it thinks it is.

I've just swapped one of these out on a car here for the exact same reason - the Link CAN Lambda was a breeze to install (I added the 22uF capacitor as per the alternative wiring diagram) and it's all substantially better now! I'm no longer fighting the readings on a daily basis!

Yeah, I wanted to switch from eManage to Link without changing anything else.  Should have went the extra step and got the CAN Lambda.

[Adamw]

I had a quick look over your map, there are definitely odd VE numbers around idle and your load abs looks off too.  You have ~38% VE at idle, typical would be more like 50-65% and your load abs is at 10% where typical would be more like 17-20%.  

So something is off with one of the measurements.  Will need to see a log to confirm if there is anything obvious like over sensitive accel enrichment dumping exta fuel in at idle.  

Be aware that you have a duel fuel table set up that enables table 2 when vtec engages, that 2nd fuel table isnt set up so it is going to go dead lean if vtec ever engages (currently set to 9000rpm so unlikely, but still a bit risky).  I would either turn that off for now or cope table 1 into table 2 position.  

[mrdjeezy]

7 hours ago, Adamw said:

I had a quick look over your map, there are definitely odd VE numbers around idle and your load abs looks off too.  You have ~38% VE at idle, typical would be more like 50-65% and your load abs is at 10% where typical would be more like 17-20%.  

So something is off with one of the measurements.  Will need to see a log to confirm if there is anything obvious like over sensitive accel enrichment dumping exta fuel in at idle.  

Be aware that you have a duel fuel table set up that enables table 2 when vtec engages, that 2nd fuel table isnt set up so it is going to go dead lean if vtec ever engages (currently set to 9000rpm so unlikely, but still a bit risky).  I would either turn that off for now or cope table 1 into table 2 position.  

Here it is.  Thank you for the heads up on the duel fuel table.  I didn't get that yet.  After I get this sorted out I'll work on getting that second fuel table up and running.

PC Datalog - 2023-01-2 8;50;56 pm.llgx

[Adamw]

I dont see much wrong in there.  I suspect either your injectors flow more than you believe, the dead times are too long, or the fuel press is higher than what you have set.  More realistic values in the charge temp approximation table will help a little in the right direction but not by much I think.   

As a test I would turn off Async injection and wall wetting comp just in case they are adding extra fuel due to unstable MAP or something but I tested your map on the bench simulating similar conditions as your log and got pretty much the same PW.  

[mrdjeezy]

On 1/3/2023 at 5:51 AM, Adamw said:

I dont see much wrong in there.  I suspect either your injectors flow more than you believe, the dead times are too long, or the fuel press is higher than what you have set.  More realistic values in the charge temp approximation table will help a little in the right direction but not by much I think.   

As a test I would turn off Async injection and wall wetting comp just in case they are adding extra fuel due to unstable MAP or something but I tested your map on the bench simulating similar conditions as your log and got pretty much the same PW.  

The dead times were too long.  I had RC 750cc settings in there instead of RC 550cc settings.

Adjusted that, and multiplied all the values in fuel table 1 to account for the dead times and the CC size (by 1.26 approx) and now the air per cylinder read is pretty steady at .12 and lambda isn't overcorrecting.

Thank you for pointing me in the right direction.