|
Snow Trooper
|
 |
« Reply #195 on: November 08, 2012, 10:52:52 PM »
|
|
|
i bet they will still swing, r32 tuners had these issues for years. it wants individual bank control and readings. when it reads the same exact voltage it tries to make sure it has control and it trims the banks away from each other, keeps doing this till it gets results, with the same reading still it never gets those results and eventually swings to +25% on one and -25% on another.
system constant for stereo lambda is our issue.
i tried to fix it by offsetting my simulated narrow band data by .001 volts from bank to bank, it helped but didnt cure it.
|
|
|
|
|
Logged
|
cartoons?
6A 61 72 65 64 40 76 6C 6D 73 70 65 63
|
|
|
|
prj
|
|
« Reply #196 on: November 08, 2012, 10:57:28 PM »
|
|
|
phila_dot - cool, thanks. I would have done it tonight, had to travel due to work so was not able to earlier. Good you sorted it  |
|
|
|
|
Logged
|
|
|
|
|
jibberjive
|
|
« Reply #197 on: November 09, 2012, 12:08:51 AM »
|
|
|
i tried to fix it by offsetting my simulated narrow band data by .001 volts from bank to bank, it helped but didnt cure it.
What if you did the offset by more than .001 v? If it really is that there is some threshold that the difference between the 2 banks needs to be higher than, then you should be able to breach it some time. They may have a check for 'too close to identical', but I can't believe they would have a check for 'make sure the values aren't just offset from each other'. Wouldn't this stuff be visible somewhere in the disassembled code (I don't even know how to disassemble, but if there is a stereo check routine it has to be visible somewhere down the logic stream past the O2 voltage.) |
|
|
|
« Last Edit: November 09, 2012, 01:37:54 AM by jibberjive »
|
Logged
|
|
|
|
|
phila_dot
|
|
« Reply #198 on: November 09, 2012, 05:42:34 AM »
|
|
|
If it is that bad then don't turn long term trims back on.
We should be able to track it down and figure out a solution. I'll get some variables if y'all want to do some logging.
There are offsets in the code that we can use as a last resort.
|
|
|
|
|
Logged
|
|
|
|
|
prj
|
|
« Reply #199 on: November 09, 2012, 07:05:53 AM »
|
|
|
i bet they will still swing, r32 tuners had these issues for years. it wants individual bank control and readings. when it reads the same exact voltage it tries to make sure it has control and it trims the banks away from each other, keeps doing this till it gets results, with the same reading still it never gets those results and eventually swings to +25% on one and -25% on another. Please back up your claims with some data from the FR |
|
|
|
|
Logged
|
|
|
|
|
phila_dot
|
|
« Reply #200 on: November 09, 2012, 07:12:40 AM »
|
|
|
There is a subfunction, LR-SYNC, in LR that may be related. I also see a codeword, CWLRSYNC, that apparently allows one to sychronize fuel trims and set which bank is master.
I will sort through the code when I get the chance.
|
|
|
|
|
Logged
|
|
|
|
|
marcellus
|
|
« Reply #201 on: November 09, 2012, 08:22:24 AM »
|
|
|
If it is that bad then don't turn long term trims back on.
We should be able to track it down and figure out a solution. I'll get some variables if y'all want to do some logging.
There are offsets in the code that we can use as a last resort.
I turned off the trims because with them working I became slave to whenever I drove the car, I HAD to have the laptop and VCDS running. Maybe a quick down the street and back I would be fine, but any thing over 10-15 minutes and the LTFT would have adapted to the point where the car seemed like it would barely run. I was hoping that it was a fueling issue and once I got the fuel closer the trims would settle more, and then I would be able to dial in the trims better. I drive a good distance to and from work. Half the trip to work the trims would sit at some real nice and low numbers like +/- 3%, it would be close to the same for both banks. At some point into the drive the trims would start distancing themselves from each other. I was usually able to feel it coming because the car would feel like it was stumbling/misfiring during cruise. Since it was the LTFT the only way I was able to get them to clear out was by clearing codes in VCDS while driving. It wasnt uncommon to see -25 on one bank and +25 on the other by the end of the trip until I figured out what to do to clear them out while driving. I usually caught them when they were around 12% split since right aorund there it was really noticeable. |
|
|
|
|
Logged
|
|
|
|
|
Snow Trooper
|
|
« Reply #202 on: November 09, 2012, 09:53:37 AM »
|
|
|
Please back up your claims with some data from the FR Not a claim, just what it does. Every dual bank me7 system that has tried to go to one downpipe path has had these issues. The FR doesn't talk about it much at all. |
|
|
|
|
Logged
|
cartoons?
6A 61 72 65 64 40 76 6C 6D 73 70 65 63
|
|
|
|
prj
|
|
« Reply #203 on: November 09, 2012, 12:58:51 PM »
|
|
|
Not a claim, just what it does. Every dual bank me7 system that has tried to go to one downpipe path has had these issues. The FR doesn't talk about it much at all.
Emulated narrowband signal from Innovate does not work with ME7. Even on non-stereo exhaust systems. And the FR talks in *great detail* about almost everything, since this is the *specification* for the system. |
|
|
|
|
Logged
|
|
|
|
|
Snow Trooper
|
|
« Reply #204 on: November 09, 2012, 01:34:51 PM »
|
|
|
and if the system is setup with a system constant they go into basically zero detail about setting things up, because you cant with system constants, they just are what they are. stereo lambda is a system constant.
posts on this very issue go back over a year, trust me I would love to find a true solution but up until now I do not know of anyone, even supposed professionals who have an answer other than to run two sensors pre turbo, bank specific.
I have spent more time going through the various documents available to me on just this issue than most have probably spent reading the fr in general. the answer is not in there. if it is in some better version that you or others have then they should probably share that version before arguing about not backing up info that comes from real world experience, over and over and over again, not reading a document once again for info that isnt there. i have tried two narrows in one downpipe (didnt work, ecu still tried to trim), i have tried both reading from one sensor (which looks like this new approach is repeating the effects of, still tried to trim), I have tried the emulation route to have two unique signal inputs (sorta of works but still eventually trims very slowly) I have tried two sensors pre turbo/divided banks (only thing that worked) and now I am going to go back to pre turbo sensors and only run them when dialing in new hardware as they get killed after lots of time in boost.
if someone can figure out how to make the banks not try to trim then awesome, i have always said i am foggy with me7 due to lack of access to certain info or logic. i treat every new approach as possible and with encouragement. i have always theorized that the individual bank control was bosch just being smart, fuel pressure can greatly vary bank to bank, ve could vary due to exhaust issues, 3 injectors to trim for compared to 6 is more precise, i could go on and on. it makes sense why they did it on various engines, just makes changing things hard.
emulated data does work with me7. bosch has a voltage line offset of usually about .007 volt, you have to compensate with any emulator to make it think its a proper sensor and not an emulated signal, also the other factory settings with innovate are wrong. i trial and error ed over thousands of miles as well as measured sensors to get it right because no one else had or wanted to share the info, not even bosch on what the offsets are for their narrow bands.
|
|
|
|
|
Logged
|
cartoons?
6A 61 72 65 64 40 76 6C 6D 73 70 65 63
|
|
|
|
prj
|
|
« Reply #205 on: November 09, 2012, 01:57:27 PM »
|
|
|
Look, if there is an algorithm that compares the banks it will be described in the FR.
Whether you have found it or not is a different story.
The FR /is/ the spec for the implemented logic, and it must conform to said spec. Simple as that.
I don't have time right now to look through it, but all of this can be modified in code, instead of trying to fool the ecu into doing this or that.
There is SY_STERSY and SY_STERVK.
Searching the FR for SY_STERVK should yield the answer, or a clue.
As for what FR I am using - the german one of course. If you use the alfa FR, loads of stuff is missing or translated wrong, even from factory.
The lambda stuff is probably better used from the alfa FR though, as it should be narrowband and it is not translated.
P.S.
I think the problem is just due to how the algorithm works, so the trims start creeping away.
The right way to solve this, is to merge the fuel trims in ASM code, so that all fuel trims are used from one bank. I don't think it tries to move the banks away from each other, it just happens, because it has two trims and one sensor.
|
|
|
|
« Last Edit: November 09, 2012, 02:05:08 PM by prj »
|
Logged
|
|
|
|
|
Snow Trooper
|
|
« Reply #206 on: November 09, 2012, 02:04:24 PM »
|
|
|
dude, im not trying to be an ass but i have read all of that 100 times over. others have too. no one has found anything that has lead to a solution.
|
|
|
|
|
Logged
|
cartoons?
6A 61 72 65 64 40 76 6C 6D 73 70 65 63
|
|
|
|
prj
|
|
« Reply #207 on: November 09, 2012, 02:05:33 PM »
|
|
|
dude, im not trying to be an ass but i have read all of that 100 times over. others have too. no one has found anything that has lead to a solution.
Edited the post, and described the solution. |
|
|
|
|
Logged
|
|
|
|
|
phila_dot
|
|
« Reply #208 on: November 09, 2012, 02:10:18 PM »
|
|
|
This really isn't all that hard. We just have to isolate where it is happening and adjust.
I can provide whatever variables are needed. I will take a quick look through the code for anything that might apply.
The trims originate in fr(2)_w correct?
|
|
|
|
|
Logged
|
|
|
|
|
prj
|
|
« Reply #209 on: November 09, 2012, 02:15:38 PM »
|
|
|
I agree that it's not that hard, just most people can't do anything with assembly  To spell it out - I think the problem is because of a Snowball (pun intended :p) effect. Basically since you have two separate trims, and amplitude adjustment, the trims will eventually start creeping opposite ways. The right thing is instead of simply copying the lambda voltage, to actually adjust the logic and take the mixture adjustment of bank 1 and write it into the mixture adjustment of bank 2. This can be either done by copying trims over, or the trims can be left as is, and all mixture adjustments that actually use those trims can be re-written to use bank 1. The correct way is probably to see what is the easiest to do and requires the least re-writing. |
|
|
|
|
Logged
|
|
|
|
|
