2006 MS6 being a PITA with fueling

VTMongoose

VTMongoose

John/MD1032
Greenie N00B Member
So, I'm working on @Chef's car right now that keeps running lean at low RPM's and rich at high RPM's. I have attached two maps below which are the most recent revisions. I've attached a pair of trimmed WOT logs for each one (in a single file), one of which I took the liberty of graphing so you can see the especially puzzling response in the top end.

The only difference between these two maps is that in the second map labeled "5", I:
  • Changed the AFR target from 11.5 to 11.0 (to rule out "table hopping")
  • Used the calculated AFR/target ratio at each MAF voltage to correct the MAF curve
The result was the exact same type of thing. It goes up to about 7-8% above target midway through the log and then when MAF voltage levels off in the higher RPM's, AFR plummets and ends around 4-5% below target. It's not possible to correct for this since it's actually registering completely different AFR's at the same voltage.

Information on the vehicle:
  • 2006 Mazdaspeed6 with generic bolt-on modifications, boost-leak tested recently @ 25 psi
  • CP-E "nano" intake
  • K04
  • E33 E85 mix
  • Most tables are at the Cobb OTS Stage 3 values, except the fuel pressure ones (I use the Stage 1 tables for those)
Tuning history: The car was on a stratified OTS+ tune and showed the same exact AFR trend during WOT logs. I loaded the car with a base spring pressure map with Cobb's SF intake MAF curve and it immediately displayed the same behavior. The only thing I have noticed out of the ordinary for this car is that it seems to have a lot of boost creep for having a catted downpipe. Right now in both of the maps I've attached here, I have the "Closed Loop Max Load Limit" tables set to the "Closed Loop Max Load" values so that the car is operating in true open loop mode at WOT. Setting these tables to the default values (2.50 load) seems to makes no difference other than smoothing out fueling error here and there as you'd expect.

Gonna tag some speed6 people here, who are already tired of me PM'ing them and complaining about this car. :)

@phate @nindoja
 

Attachments

Last edited:
That whore ms6, it's mine..
[doublepost=1462671897][/doublepost]This is the 6 that Lynze reached out to you about back in October. I haven't added any mods since I've bought it, just some maintenance work.

Cp-e FMIC, CBE, intake, tip, hpfp / Atp catted DP / synapse synchronic - recirculated
 
Hi Rob. I am using ATR still, and so is Chef. The maps are attached to the OP, anyone is more than welcome to take a peek and make sure I haven't forgotten to copy something over.

I have "Load Target Comp - High RPM (BETA)" set to 0.00 (default is 0.20 I think?) and "Load Target Comp - Injector Base Offset (BETA)" at 20 (default is 0). Is this what you're referring to?

edit: those shouldn't technically be relevant here though since WGDC is set to zero.

Also, can I just get a read on your opinion as to whether it's a tune/ECU issue or a sensor/hardware issue? I have not really made up my mind about this subject yet.
 
Last edited:
I think a part of the issue is that when the small blips of KR are occurring, the car is adding fuel, and that is included in your analysis and calculation of the corrected fuel curve.

In the initial part of the logs, the open loop transition can probably be explained by the closed loop exit delays, because when I cut out the reported trimming, the corrected curve fell more in line.

I use the open loop trimming to my advantage, but if you wanted to try the resulting curve from pulling out all of the kr, and the closed loop delay... I fit a curve onto both data sets and attached it here from 3.36 volts on.

Worth a try if you'd like (note the curve is going to need some massaging left of 3.36)

Code: 
93.60    99.72    106.05    112.57    119.27    124.40    131.38    138.48    145.70    153.02    160.42    167.88    175.39    182.93    190.49    198.04    203.70    211.20    218.66    226.05    233.36    240.57    247.67    254.63    261.44    268.08    274.54    279.25    285.35    291.22    296.84    302.20    307.27    312.05    316.52    320.65    324.44    327.04    330.18    332.92    335.25    337.16    338.62
 
rfinkle2 said:
I think a part of the issue is that when the small blips of KR are occurring, the car is adding fuel, and that is included in your analysis and calculation of the corrected fuel curve.

In the initial part of the logs, the open loop transition can probably be explained by the closed loop exit delays, because when I cut out the reported trimming, the corrected curve fell more in line.

I use the open loop trimming to my advantage, but if you wanted to try the resulting curve from pulling out all of the kr, and the closed loop delay... I fit a curve onto both data sets and attached it here from 3.36 volts on.

Worth a try if you'd like (note the curve is going to need some massaging left of 3.36)

Code: 
93.60    99.72    106.05    112.57    119.27    124.40    131.38    138.48    145.70    153.02    160.42    167.88    175.39    182.93    190.49    198.04    203.70    211.20    218.66    226.05    233.36    240.57    247.67    254.63    261.44    268.08    274.54    279.25    285.35    291.22    296.84    302.20    307.27    312.05    316.52    320.65    324.44    327.04    330.18    332.92    335.25    337.16    338.62
Click to expand...

Thanks. Honestly I don't think that curve will make any difference but I'll put it into a map and we'll give it a shot as soon as I figure out how to actually get those values to paste into excel, openoffice calc, or ATR... @Raider any idea how to make this work BTW?

I don't want to waste your time and I value your input @rfinkle2 so let me zero in on my primary question/concern.



You can see in this log snippet, that in the two highlighted rows, where there is no knock in the vicinity, the car is registering the exact same MAF voltage, however in the lower line (at the higher RPM) the car is adding an extra 2% IDC and the primary O2 is reading 0.80 lambda instead of 0.84. This happens every single log, and it's not isolated to one cell of the MAF curve either because we're at spring pressure, so logs in different temperatures are yielding different final MAF voltages, but with the enrichment trend (another example below):



Let me know your thoughts.
 
The car is going to change the IDC based on RPM.. although I'm not sure that is exactly what you were asking.

You can copy and paste that code directly into the map using data > text to columns.

I would turn on the open loop trimming and stop fighting the ecu.
 
Can you post up the untrimmed logs @VTMongoose? I think I agree with @rfinkle2's assessment of why the car's going super rich at the beginning.

Did you see the same trend when calibrating for just pump gas?

@rfinkle2: I think his plan is to get the fuel curve as close as possible then turn the trimming back on.
 
I should say, rather, put the code I posted into excel and use the data / text to columns function and then paste into the map.

Good luck guys.

Are you 100% sure this car is mechanically in check (aka no boost leaks)?
 
Last edited:
I'm not 100% sure about the leak testing @rfinkle2, because the car is in another state, but honestly I trust the owner when he says it's free of boost leaks.

I took the time to put your MAF curve into a map and Chef logged the results which are attached.

We're still seeing the same trend as before, albeit, the top end is not quite as rich. Still, the car displayed the same behavior with one of my maps with a radically different MAF curve (also with WOT compensation enabled) (will post if requested).

I don't really get what happened in datalog1. There is persistent, HUGE knock in the top end, the likes of which I've never seen on my own vehicle. I suspect the injector seals may be to blame. My own car has displayed some of this same behavior before, just not with numbers this large. I will be interested to see if you have any further ideas Rob because honestly I have run through this in my head a million times and just can't figure out what is the problem here...mainly why the car seems to like to run lean in the low end...

I think the next logical step here honestly is to put the car on 93 to take the scalars/fuel gravity out of the equation.
 

Attachments

You must log in or register to reply here.
Back
Top