An excellent discussion point was raised in the ECUFlash Maps thread that I've decided to break out into a new thread here. This is both to keep the map thread on target and to highlight this topic for discussion. Does the Evo X ECU actually learn to boost? If so how?

The question was put nicely by Bryan from GST:


So how does it learn boost? It overshoots with the "tuned" WGDC, error correction kicks in and lowers boost, then is there stored information to run lower WGDC then what is in the WGDC tables after a certain learning period?

Discuss!

:shades:

If the ECU truly learns, then does it modify your values, or store those updated/saved values in some other chunk of memory?

And could we read those values?

If it's done like the Subarus, then the learned value is a "correction" that gets saved elsewhere. The base maps shouldn't change, and the learning should be lost anytime you reset the ECU by cutting power to it. What I've described is based on Subaru ecu's - it would be nice if someone could provide more info on how Mitsu does it.

With my old 350Z, it supposedly learned by how you drive. So If I had been driving like a granny with fuel economy in mind for awhile I would reset my ecu. In the case of the Z there is a ecu reset procedure that consists of timed pressing of the gas pedal in and out. It always seemed peppyer after a reset, maybe all in my head though. It is common practice among Z drivers.

lol in my G... i always mashed around... once i reset it felt a LOT slower lol... definitely not just you man haha

It learns but very quickly that you cant even notice. For older models, it might take 2 weeks but for Evo X itz just a matters of milliseconds. Fuel correction and boost correction is almost instantly.

as long as there's no long term gimp tables like subies have (if you knock in a certain area, it remembers and pulls timing in that area for like,,, ever)

The Mitsubishi logic is different from the subarus ..

Evo runs the initial map on the most aggressive mode.. we have a octane counter that starts at max value and slowly decrements everytime it experiences knock .. this value is then used to interpolate an intermediate value between the high octane and low octane maps .. once that 'octane' value reaches 0, you're running on the Low octane maps .. THAT is one of the reasons MOST people don't flatten the High octane and Low octane maps ..

So guess what .. there's no learning unless you have problems or a bad tank of gas on the mitsubishis :bowlol: .. its always best power ..

On the Subarus, its the otherway round.. :D

As for boost .. it takes a long time to settle down .. I won't call that fast .. its the same for the older series .. the codes on the models are almost if not similiar .. so far .. I haven't seen much changes in codes .. There's some differences in the newer boost codes though .. and that's what I'm referring to and what Bryan is alluding to ..with a twist :D

so... i just installed my cobb AP with the AZN 91 map. Car feels a lot better - much more smooth up top. However, should I have re-set my ecu after the flash? or does flashing it essentially do that for me?

There is no need to reset your ECU, you just flashed it. =)

The Mitsubishi logic is different from the subarus ..

Evo runs the initial map on the most aggressive mode.. we have a octane counter that starts at max value and slowly decrements everytime it experiences knock .. this value is then used to interpolate an intermediate value between the high octane and low octane maps .. once that 'octane' value reaches 0, you're running on the Low octane maps .. THAT is one of the reasons MOST people don't flatten the High octane and Low octane maps ..

So guess what .. there's no learning unless you have problems or a bad tank of gas on the mitsubishis :bowlol: .. its always best power ..

On the Subarus, its the otherway round.. :D

As for boost .. it takes a long time to settle down .. I won't call that fast .. its the same for the older series .. the codes on the models are almost if not similiar .. so far .. I haven't seen much changes in codes .. There's some differences in the newer boost codes though .. and that's what I'm referring to and what Bryan is alluding to ..with a twist :D

This is correct and I have discussed this with the tuners from Cobb.
There is a catch though will hard files on your ECU. When you throw a CEL it has base settings it goes to so you can limp home or to a shop. Thats why when your MAF goes out it idles @ 2k rpm. You can't just write or flash over these, you have to reset your ECU by battery disconnect.

This is correct and I have discussed this with the tuners from Cobb.
There is a catch though will hard files on your ECU. When you throw a CEL it has base settings it goes to so you can limp home or to a shop. Thats why when your MAF goes out it idles @ 2k rpm. You can't just write or flash over these, you have to reset your ECU by battery disconnect.

No you don't .. you just have to ensure the condition doesn't exist before you can clear it .. This is stored in the ETACs info memory .. that's why you can't clear it ..

The ECU will test the same condition on start up for the fault .. if it exist, the limp mode will continue .. when you have 4 consecutive starts without the problem .. the code will be released as intermittent fault and you can clear it then ..

This has nothing to do with learning .. its just fault diagnosis and hardware protection ..

So far the "learning" capability of the mitsubishi ECU has not changed from previous generation cars .. and I've worked on that for more than 4 years .. its just 'enhanced' ..

so by turning the car off and on 4 times itll reset the SES light

so by turning the car off and on 4 times itll reset the SES light

Condition is you have to start the car .. if the fault is non existent for 5s after starting 4 times .. the code is cleared ..

the CEL is cleared and the limp mode is removed .. but the code and the freeze frame remains stored until it is erased..

There is no need to reset your ECU, you just flashed it. =)

Doesn't work that way....;)

Condition is you have to start the car .. if the fault is non existent for 5s after starting 4 times .. the code is cleared ..

the CEL is cleared and the limp mode is removed .. but the code and the freeze frame remains stored until it is erased..

Gunzo, does leaving the rear stock o2 sensor (before the cat, in the stock downpipe) out of the exhaust stream cause any problems with how the car runs? I'm removing that sensor to put in a wideband temporarily while tuning so there is a CEL. Will doing that cause any issues with how the car runs while I try to log and tune? Thanks!

No you don't .. you just have to ensure the condition doesn't exist before you can clear it .. This is stored in the ETACs info memory .. that's why you can't clear it ..

The ECU will test the same condition on start up for the fault .. if it exist, the limp mode will continue .. when you have 4 consecutive starts without the problem .. the code will be released as intermittent fault and you can clear it then ..

This has nothing to do with learning .. its just fault diagnosis and hardware protection ..

So far the "learning" capability of the mitsubishi ECU has not changed from previous generation cars .. and I've worked on that for more than 4 years .. its just 'enhanced' ..


Ok, I see what your saying and the fact still stands you can't just clear a code and reflash the ECU. It still either takes the 4 starts or battery disconnect to make it all right again. Correct?