[Mik325tds]

DWR and I have been doing a lot of thinking about how to improve the shift behavior of our transmission without dropping a kilo of $ to an UK tuner.
We have developed a game plan that might work but will require a lot of time and effort, so I was wondering if we could make this an open source type of project. Here's the game plan and some background information:
Our tranny GA6HP26Z is build by ZF (Zahnradfabrik Friedrichshafen) and is used in many carlines across multiple OEMs. Attached is a detailed doc of how it works.
Reading the "User information field" job in EDIABAS, mine returns the following information:
apiJob("GS19D","aif_aktuell_lesen","","")

Satz : 0
OBJECT = gs19d
SAETZE = 1
JOBNAME = aif_aktuell_lesen
VARIANTE = GS19D
JOBSTATUS =
UBATTCURRENT = -1
UBATTHISTORY = -1
IGNITIONCURRENT = -1
IGNITIONHISTORY = -1
Satz : 1
JOB_STATUS = OKAY
BMW_VEHICLE_IDENTIFICATION_NR = xxxxxxx
BMW_PROGRAMMING_DATE = 20100712
BMW_ASSEMBLY_NR = 000007606273
BMW_CALIBRATION_DATASET_NR = 000007606274
BMW_EXHAUST_REGULATION_OR_TYPE_APPROVAL_NR = 000000000000
WERKSCODE_HAENDLERNUMMER = 011111
TESTER_SERIENNUMMER = 12345
KM_STAND_PROGRAMMIERUNG = 0.000000E+000
PROGRAMMSTAND = 0479S90T641Z
_TEL_ANTWORT = 71 Bytes

The interesting number here is the Calibration dataset number 7606274 which corresponds to a file that is found in the SP-Daten updates for ECU reflashs. The file is called A7606274.0da and is completely unencrypted and in ASCII. The file is also attached (Password is the username of our banned friend Jess).
Looking through it with a txt editor reveals many repeating patterns which most likely are maps (3d tables with parameters depending on two input values eg. speed, torque. Best tool to discover and view those maps is WINols.
[to be continued...]

[Persian Whisperjet]

Mik when you said you did the alphabet soup thing on your back using jackstands, in Detroit in the winter, I suspected you were operating at a high level than the rest of us. now i know you are.

[Mik325tds]

From what I've read in other threads, there is supposed to be an updated file around that slightly improves the shift behavior. Unfortunately, I haven't found that one yet. It is supposed to be in the SP-daten set of 2.48.1 - Istufe E89X-12-07-508. If someone finds it, please post it here.
The plan is to take some data of our current shift behavior, then flash the new calibration, take data, correllate the changes in behavior to the changes in the file and thus identify the maps in the calibration file and learn what they do. Then we'll be able to manipulate them ourselves, calculate a new checksum and flash it back into our tranny.
The tool to flash the tranny is WinKFP which is part of the BMW Toolset that Ediabas comes with. I'll post a howto when I've tried it.

Another idea is to just flash the calibration file of the European 335d for instance this one here:
;;K_Stand: 19.08.2008
;;K_File-Name: A7595729.0da
;
;;Fahrzeugidentifikation
;;K_F1 Datenstand fuer
;;K_F2 E90 M57D30TU2TOP
... and see what happens. stevers314 had made a few of those attempts but as far as I understand he has used complete sets of SW and calibration files. These sets are described in the file GKE195.DAT which WinKFP uses to determine which SW updates are available for the ECU to be reflashed. The one I found looks like this:
$ PS10INIT N00000000000000000000000015000000000000 R
$ VERSIONKFCONF: kfconf10.dat
;Zusbauvorschrift vom 19.11.2009 15:28
;SG-TYP: GKE195
;ZB-NR TYP-NR HW-NR IX SW-NR AM PIN S CS
7587867,0000000,7573100,A,7587868DA,0FFFFFFFFFD,00 0,1 C
7610590,0000000,7591972,A,7610591DA,0FFFFFFFFFD,00 0,1 H
7606273,0000000,7591972,A,7606274DA,0FFFFFFFFFD,00 0,1 J
7609646,0000000,7591972,A,7609647DA,0FFFFFFFFFD,00 0,1 P
7569996,0000000,7566895,A,7569997DA,0FFFFFFFFFD,00 0,1 J
7569994,0000000,7566895,A,7569995DA,0FFFFFFFFFD,00 0,1 7
7592089,0000000,7591972,A,7592090DA,0FFFFFFFFFD,00 0,1 O
7595728,0000000,7591972,A,7595729DA,0FFFFFFFFFD,00 0,1 B

First number is the assembly number, then Type, then hardware number, Index, Software number, ...
Ours is underlined.

But first things first. Here are a few traces of 1st to 3rd gear pulls with different throttle inputs. I used Test-O, a free tool of our Finish friend Pheno (donations appreciated), to capture:
STAT_ABTRIEBSDREHZAHL_WERT - Tranny output shaft RPM
STAT_FAHRPEDALWINKEL_WERT - Throttle angle
STAT_ISTGANG_WERT - current gear
STAT_MOTORDREHZAHL_WERT - Engine rpm
STAT_MOTORISTMOMENT_WERT - Engine torque
STAT_SA_WERT - ???
STAT_TURBINENDREHZAHL_WERT - Turbine rpm (torque converter)
STAT_WK_WERT - Lock up clutch

My complaint is easily visible in these plots: In first and second gear, the engine rpms shoot up to about 2000 rpm before the tranny starts accepting torque and it takes until the shift into 3rd gear before it looks up for this first time. The reason why it is calibrated like this seems obvious: Protect the tranny from the huge amounts of torque our engine is producing. That's also the reason why our engine output torque is originally limited to 500Nm in 1st and 2nd gear. But I don't want to floor it all the time, I want defined and controllable launch behavior without my engine revving up to 2000+ rpms all the time. So if we succeed in tuning this behavior it will come with a bit of responsibility of the driver in order not to damage the tranny.

Any suggestions or feedback is welcome.

[TDIwyse]

Originally Posted by Mik325tds

DWR and I have been doing a lot of thinking about how to improve the shift behavior of our transmission without dropping a kilo of $ to an UK tuner...

This is an outstanding example of what makes this a great online community. Thank you for your efforts and contributions.

[temporaptor]

the later built 335d's had the 6hp28. such as my 2011 335d m sport. I know the 6HP26Z and 6hp28 are similar yet different.(not sure what changed) I remember hearing about some failures of the 6HP26Z that the 6hp28 had fixed. Would this remapping apply to both?

[Yozh]

This is exciting. Thank you for pioneering this, hope we can all pitch in and come up with a solution. Also wonder how the 26 vs 28 difference will play out.

[dixy2k]

I don't understand a thing in this thread, but I like the idea.

[AzureHaze]

Originally Posted by dixy2k

I don't understand a thing in this thread, but I like the idea.

Same here. If OP wants a donation for this project, I'm in 100%.

[DWR]

Mik352tds and I are still sorting out the 6HP26/28 differences. The solenoids and their activations do seem different. However, at a high level, that is not going to matter. Just like in engine ECUs, certain maps have certain kinds of patterns. An EGR map, for instance, is not too hard to identify in any ECU just because of its characteristic shape. I am learning as fast as I can what 'shapes' are in a typical TCU calibration. Here's a teaser. This an easy to find map in the calibration file. The peak has a value of 4648. Hmm sounds like our WOT shift point doesn't it? Still trying to figure out the axis.

BMW info is hard to find. But if anyone has any insights into the FORD 6R60/6R80 transmission, that would help. They are ZF licensed replicates. Also, if anyone has any experience or know anyone with exerience with Powertrain Control Solutions software, that would also help. Apparently, they have recently buttoned up their website and I no long have access to their documention and software updates.

[DWR]

Originally Posted by TDIwyse

This is an outstanding example of what makes this a great online community. Thank you for your efforts and contributions.

You see the first couple of posts - think how lucky I am to be able to colaborate with a guy like Mik325tds! Very honored

[DWR]

Originally Posted by DWR

Apparently, they have recently buttoned up their website and I no long have access to their documention and software updates.

Please disregard, the panic was unwarranted. Found their developer section.

[torqueisking]

Amazing thread! lot of popcorn and beer cued up for this one

I was thinking of electrically manipulating the clutch slip with a piggyback microcontroller but this is much less intrusive solution.

[dixy2k]

Will this development address the rough shift pattern between the 2nd and 3rd gear, permanently?

[DWR]

Originally Posted by dixy2k

Will this development address the rough shift pattern between the 2nd and 3rd gear, permanently?

Don't even know what "rough shift pattern" means, let alone how to address it, lol. But seriously, Mik352tds is interested in TTC lock up and I am interested in faster shifts. It may not be obvious, but those could be opposing views of the same control issue. For drag racing, it can make more sense to have the TTC unlocked to absorb hard shifts. For mileage, direct coupling with soft engagement is more desireable.

We are a long way from being able to address anything ... and we might fall flat on our faces. Just going to try our best.

[Hoooper]

Interestingly I have a file for the N57 which is indicated as the same hardware number as my m57 files (7591972). The file is A7609647.0da. There are several other files in my folder for the m57 which appear to be older software than the A7606274 that is currently running, such as A7595729.0da. Would an older flash not be just as useful in this attempt as a newer file.

[DWR]

Originally Posted by AzureHaze

Same here. If OP wants a donation for this project, I'm in 100%.

Damn! Mik and I already agreed to do this for free! LOL

At this point, the only thing we need is the nerve to mess around with 2 perfectly good transmisions . Mik has agreed to make the first attempt at reflashing, since he is braver, smarter, and better looking than me (I hope I didn't leave anything out - oh yeah, and has the short straw). I will be following him into the quicksand shortly thereafter ...

[Mik325tds]

Originally Posted by temporaptor

the later built 335d's had the 6hp28. such as my 2011 335d m sport. I know the 6HP26Z and 6hp28 are similar yet different.(not sure what changed) I remember hearing about some failures of the 6HP26Z that the 6hp28 had fixed. Would this remapping apply to both?

That is a good point. As DWR already responded, we are still working that one out. Mine is a 2011 m-sport as well, the transmission is built on Jun/23/2010 and the cars B-day was shortly after (06/2010), so I should have the 6HP28 but I was pretty certain I had the 6HP26 when I looked it up for the transmission fluid kit.
The BMW part number read by Ediabas is 7591972, but Realoem doesn't find anything under that. It shows part number 24007590129 for my VIN which is a GA6HP26Z.
Very strange. More research is needed.

[Mik325tds]

Originally Posted by DWR

You see the first couple of posts - think how lucky I am to be able to colaborate with a guy like Mik325tds! Very honored

Originally Posted by TDIwyse

This is an outstanding example of what makes this a great online community. Thank you for your efforts and contributions.

Thanks guys - comming from you that means a lot to me. I'm humbled.

[Mik325tds]

Originally Posted by DWR

Mik352tds and I are still sorting out the 6HP26/28 differences. The solenoids and their activations do seem different. However, at a high level, that is not going to matter. Just like in engine ECUs, certain maps have certain kinds of patterns. An EGR map, for instance, is not too hard to identify in any ECU just because of its characteristic shape. I am learning as fast as I can what 'shapes' are in a typical TCU calibration. Here's a teaser. This an easy to find map in the calibration file. The peak has a value of 4648. Hmm sounds like our WOT shift point doesn't it? Still trying to figure out the axis.

Wow, Derek you are great! I have yet to install WinOLS... . We'll be looking at these type of maps a lot over the next weeks and months. I was a bit surprised to find out that the transmission controller is actually build by Bosch! The fact that WinOLS doesn't show a checksum for this map is a good sign. It seems there's just one CRC at the end of the file then which we'll still have to figure out. We might need to purchase the CRC builder for WinOLS. But we'll deal with that later.

[Mik325tds]

Originally Posted by dixy2k

Will this development address the rough shift pattern between the 2nd and 3rd gear, permanently?

Originally Posted by DWR

Don't even know what "rough shift pattern" means, let alone how to address it, lol. But seriously, Mik352tds is interested in TTC lock up and I am interested in faster shifts. It may not be obvious, but those could be opposing views of the same control issue. For drag racing, it can make more sense to have the TTC unlocked to absorb hard shifts. For mileage, direct coupling with soft engagement is more desireable.

We are a long way from being able to address anything ... and we might fall flat on our faces. Just going to try our best.

+1 on that. Our desires for shift behavior are very different - I'm looking for fast shifts at low rpms and early lock ups so the car runs most efficient, others might look for optimal launch for drag racing (wonder who that might be ).
If we succeed in finding the relevant maps we (or you) can change the behavior for anything you like, but make no mistake - there is a looong and steep learning curve ahead for all of us. There is a good chance that we don't make it but that shouldn't keep us from trying.

[Mik325tds]

Originally Posted by Hoooper

Interestingly I have a file for the N57 which is indicated as the same hardware number as my m57 files (7591972). The file is A7609647.0da. There are several other files in my folder for the m57 which appear to be older software than the A7606274 that is currently running, such as A7595729.0da. Would an older flash not be just as useful in this attempt as a newer file.

You got it! They all belong to the same ECU family GKE195 and should all flash as long as the HW is the same. Based on the assembly instruction file we should be able to flash at least 5 different combinations:
$ PS10INIT N00000000000000000000000015000000000000 R
$ VERSIONKFCONF: kfconf10.dat
;Zusbauvorschrift vom 19.11.2009 15:28
;SG-TYP: GKE195
;ZB-NR TYP-NR HW-NR IX SW-NR AM PIN S CS
7587867,0000000,7573100,A,7587868DA,0FFFFFFFFFD,00 0,1 C
7610590,0000000,7591972,A,7610591DA,0FFFFFFFFFD,00 0,1 H
7606273,0000000,7591972,A,7606274DA,0FFFFFFFFFD,00 0,1 J
7609646,0000000,7591972,A,7609647DA,0FFFFFFFFFD,00 0,1 P
7569996,0000000,7566895,A,7569997DA,0FFFFFFFFFD,00 0,1 J
7569994,0000000,7566895,A,7569995DA,0FFFFFFFFFD,00 0,1 7
7592089,0000000,7591972,A,7592090DA,0FFFFFFFFFD,00 0,1 O
7595728,0000000,7591972,A,7595729DA,0FFFFFFFFFD,00 0,1 B

Today, I successfully flashed the ZB 7595728 with cal 7595729 with WinKFP . There were a bunch of DTCs afterwards but cleared fine. However, the driving experience wasn't that much different. If it all, the shift points are now later (around 2400 rpms). I took some data today which I'll post soon but intend to let the cleared adaptions settle down a bit and then take more data.

[SouthDiesel190]

How does one perform the transmission adaptation reset? I've seen mentions of a manual method and I assume there is some sort of electronic/software method. How are you all doing it after transmission services or flashes?