Fuel map tuning 3 methods?

[Releeuw45]

Hi,

Currently I'm working to improve my fuel map. In case of using a dyno I can use the CL correction map on the PC. Without a dyno I can use the Lambda correction % in the log file or work with the logged Lambda value I'm not allowed to use a laptop at the track so that option doesn't work.

Question here; how accurate can I use logfile data to create a correction map, knowing that the logfile created on the track contains transition correction.
Is it enough to gather all lambda values around a specific TPS / RPM and use the average as input for the correction to the main fuel map?

Rene

[stevieturbo]

I guess a lot will depend on what you're actually trying to correct ?

Transients will only last a split second and are obviously unavoidable during normal conditions.

But just how far off is whatever you are trying to correct ? and does it really need corrected in the first place ?

Is there a problem with how the bike runs ?

Sometimes you can get too anal about making changes, when there really isnt a problem in the first place.

Performance from an engine isnt about targeting specific AFR's, it's about performance.

[Releeuw45]

After doing 2 trackdays last week I logged all the sessions and saw corrections from -11 up +18% with trasitions ON. I guess that's not good enough yet.
Engine run fairly well with only a slight holding back at 4000RPM and 30% throttle.
After that is run ok up to 12500 rpm (Suzuki GSXR 1000) but is difficult to notice any too lean or rich running doing 160 mph on a twisting racetrack....

[Roverdose]

I don't use logs, but I can imagine if you can hold a throttle site for a few seconds you will be able to see that on a log, the transients will have run out by the end of that, so adjusting from there would be acceptable.

Drew

[stevieturbo]

After doing 2 trackdays last week I logged all the sessions and saw corrections from -11 up +18% with trasitions ON. I guess that's not good enough yet.
Engine run fairly well with only a slight holding back at 4000RPM and 30% throttle.
After that is run ok up to 12500 rpm (Suzuki GSXR 1000) but is difficult to notice any too lean or rich running doing 160 mph on a twisting racetrack....

You're assuming those -11 or +18 are correct though. Often the corrections it saves can be very wrong. I wouldnt put too much faith in them, especially when you have given it such a huge amount of correction like 18%

When I do a pull on the street, half the time the fuel correction map doesnt even pick up a lot of the sites Ive just passed through. It really does prefer to have the sites passed through very slowly for it to decide on any corrections, and I'll guess a bike doesnt pass through the rpm range slowly very often.

IMO if you want to use the FCM, you should reduce the +/- level of control you give closed loop. Maybe 8% either side at maximum. Even 8% either way is a huge mixture change.

As you have a distinct problem, Id look at the logs and concentrate on that area and get that right. The hesitation could be too much fuel or too little.

And as I said....if you cant feel a problem...maybe there isnt a problem. But a large +/- swing like that can cause problems, as the ecu tries to pull fuel when it really shouldnt be.

-11% would easily be enough to cause a huge flat spot.

[Jon K]

Stevie - lets say my target is 12.5:1 AFR and my car is running 11.4:1 AFR. Could one not go back over the log and see what cell the engine was in for that point, and subtract 9.6% from the pulsewidth? Does that not work?

It seems to have worked for me and I think it doesn't matter what corrections are. If you look at the injector pulse size, that should be the size with any corrections applied. If its off by X% wouldn't adding/subtracting X% hone it in?

[katana]

First off - i'm not a tuner so take what I say with some latitude! I believe relying on a Wideband for exact numbers is fool hardy. They are not in the cylinder during the burn and are some way downstream of the cylinder when they get to sample whatever comes at them! The engine could have gone through several / 10 revolutions / ignition events when the lambda throws up a number and the map location is unlikely to be in the same place. Lambda is good for trends not exact numbers - most lambda won't update faster than 10hz when the motor is exceeding 100hz rate of change otherwise we all be doing WOT / max power closed loop tuning with one.......IMHO of course!

[stevieturbo]

I'd agree with the above.

If you're at 11.4 and want 12.5, I wouldnt try and achieve that in one fell swoop.

Go 2% at a time for example. But even 2% is a fairly noticeable change. 10% is massive.

I have mine to allow +8% and -4% via closed loop. Even in logs rarely does closed loop respond fast enough or correctly to adjust the mixture in real time when accelerating. In the FCM I do see corrections of around +5% or -3.9% )
Giving it a large scope for adjustment can often cause flatspots as it tries to lean things when it really shouldnt be.
Steady state and cruise, yes closed loop does work ok. Accelerating....no, unless it's very slow. And then when traction is an issue, again wideband readings cannot be trusted once you break traction. Generally they will read lean. Likewise if TCS is active and cutting fuel/sparks, wideband readings cannot be trusted.

I generally ignore these 5% or so numbers in the FCM because I know from looking at a log on average my mixtures are good. You could be forever chasing small changes up/down, but it's a waste of time really.

There will always be small flactuations whether due to load, temperature, road surface etc etc.

Just dont give closed loop so much control it actually works against you. Even my -4% is large. Keep it less than that, and at least you know the ecu wants to reduce fuel, but isnt capable of reducing fuel so much that it will actually cause a flatspot.
Then you can reduce fuel in those areas manually if so desired.

[Releeuw45]

Hi Guys,

Thanks for the usefull replies. I already discovered lags about 0,5 second (12Hz, LC-1) but the data in the logs is still usefull to correct areas in the fuelmap with half or one third of the suggested correction to prevent correcting up and down next time.

I will drop down the CL correction range within +-8% to get a better feel for the flat spots in case I log a lot of +8% corrections. If I can get everything within 4% it should be OK.

Also no need to disable transition because of the limited influence on the trend found in the log files.

regards,
Rene

[stevieturbo]

Drop the -% lower.

The -% can cause flat spots because the ecu sometimes tries to pull fuel when it shouldnt.

Drop it to say -2% and see it it performs. That will remove any chance CL is actually causing the flat spot.

I'd even say go something like -2% and +4%

Keep the trims small until you have a good map built. Otherwise you'll be going up/down forever.