Karter16's Silbergrau E46 M3 Journal

[karter16]

Torrential rain yesterday and today so I can't do any work on my retaining wall at home. Got round to a little job I've been wanting to do for a while. The trim around the SMG shifter boot makes an annoying plastic on plastic creak sometimes when shifting, which isn't particularly pleasant from a driving experience point of view.

Today I took the trim apart and used some tesa tape to insulate the clips and mating surfaces. I also gave everything a good clean while I had it apart. Put it back together and the squeaks are completely gone (unless I press very hard directly on the plastic trim), shifting is now a satisfying mechanical click, rather than the annoying creaking sound it could be before.

[karter16]

Had a great weekend working on the car. On Saturday morning Dad and I got the car up on stands and exhaust and driveshaft out. Unbolted the exhaust at the headers and dropped the whole thing, then the same with the driveshaft. So much faster than last time given we're just removing and reinstalling rather than taking it all apart to clean. I'm super happy with how everything under the car is holding up, it's 8 years since we did the underbody refresh and everything is looking really good under there.

I did the new gear position sensor first, access is a bit of a mission but the mutilated allen key approach does work. The old sensor was extremely notchy and rough after 20 years of service. Worth noting that we ran the first part of the SMG adaption (up to the point it wants the car started) to make sure it was happy with the new GPS.

After that we moved on to lowering the rear subframe to facilitate the installation of the Yurkan Cages 6 point brace. We took off the brake calipers and rear trailing arm pockets and lowered the entire thing about 100mm on a couple of jacks (to give us more control over forward/rearward weight distribution). That gave us enough access to unbolt the front RACP studs and get them out. We then jacked the subframe back into place and loosely bolted it, which gave us a better safety factor while working under it.

At this point we drained the diff, swapped out the rear diff cover for a new one and reinstalled with new bolts and drain plugs. I've mentioned previously that I'd long experienced some clunkiness when shifting/loading/unloading the drivetrain. This was vastly improved by adapting the SMG. I wasn't sure whether what remained was the diff getting a bit tired or whether the rear bushes needed replacing but I was pleased on checking the diff that there is no detectable lash in the diff at all, likewise with the driveshaft. I was therefore pretty hopeful that it was just the rear diff bushes that were the culprit.

The rest of the process was pretty straightforward, we unbolted one point at a time and using a threaded drill guide that dad turned on his lathe and an extra long 6mm drill bit we went up through each of the 4 points in turn. Then it was simply a case of come down from the top with a 10.5mm bit before running a tap on a long extension up from underneath to complete the thread the whole way through. With that done it was then a case of bolting in each of the 4 points and torquing to spec.

The rest of the reassembly took a couple of hours and we had the car on the ground again. The install of the brace was super quick - fitment of the brace was absolutely perfect - Marin's fabrication is spot on. Then it was just a case of running the complete adaption on the SMG and going for a test drive.


Today I drove the car on a route I'm very familiar with and I'm extremely pleased with the end result! The last of the clunkiness when shifting is now resolved. Even under heavy acceleration in S5/S6 all you get is the sound and surge of power. I'm super stoked.

And the brace. Well, I know others have reported a big improvement, but I wasn't completely sure if I'd find the same. Suffice to say the improvement is not subtle! For me the rear has always felt slightly less co-ordinated than the front (not to say it's not objectively excellent to begin with, but it's always stood out to me that the car is just a little bit more put together in the front). Now with the brace it's the other way around! The change in stiffness is hugely noticeable even just driving around town. As soon as you get to a roundabout or uneven tarmac it becomes very apparent how much more rigidity there is. The car really feels glued together in the back now.

I love it and am excited to go for a decent spirited drive.

[karter16]

Well after more than 9 years of E46 M3 ownership I finally joined the flatbed gang. The GPS died on me entirely quicker than I was anticipating and left me with a box full of neutral this morning. Car is now at Dad's ready to be worked on. bmwfnatic (thank you!!!) is picking up a new sensor this week and fedexing it to me.

Had been going to do the RACP brace about a month ago but wasn't feeling up to it at the time, glad I didn't as can do this at the same time.


Sent from my iPhone using Tapatalk

[0-60motorsports]

Well after 9 years of flawless SMG ownership the car threw SMG codes 36 and 37 on the way to work this morning and again on the way home. Not the worst timing as was planning to have the car in the air soon for the RACP brace install, so can do the gear position sensor at the same time. I look forward to spending 500 Euro on a new one…


Sent from my iPhone using Tapatalk

LOL just get a used one.

[karter16]

Well after 9 years of flawless SMG ownership the car threw SMG codes 36 and 37 on the way to work this morning and again on the way home. Not the worst timing as was planning to have the car in the air soon for the RACP brace install, so can do the gear position sensor at the same time. I look forward to spending 500 Euro on a new one…


Sent from my iPhone using Tapatalk

[karter16]

On the way to a much more complete disassembly of the 0401 software, bmwfnatic and I have been working on disassembling the 1801 software first as being an MSS54HP version it should be more closely aligned with the 0401 software and help reduce the unknowns for us to deal with when we get to 0401.

As part of this one of the things I've been working on is validating a lot of the characteristics (parameters/curves/maps) for the 1801 partial against the public XDF, disassembled code, etc. plus referring to the 0901 A2L and binary to map as many of the SMG characteristics as possible (given these are almost all wrong in the 1801 public XDF). In total I've identified 246 characteristics in the XDF that are incorrect, and an additional 226 characteristics to be added which aren't in the XDF currently. My plan is to build out an updated and corrected XDF which corrects all known characteristics, adds additional known characteristics, removes known no-longer-existing characteristics (e.g. were in the 0901 code but are gone as of 1801) and specifically identifies characteristics which exist but are unknown, along with what information is known about them (e.g. module, etc). There's about 20 of these in general, plus another 65 or so specific to the SMG module (the SMG module appears to have undergone heavy development between 0901 and 1801).

Once this is done I'll publish the updated 1801 XDF for everyone's reference.

[karter16]

Hell yeah! This reads like I felt.

It's so good and so satisfying when it comes together and you see what the car is capable of when it's tuned right.

Did you get through high load, low rpm testing? Specifically around 2k rpm? There might be a gem waiting for you to discover.

So it's almost there - the pinging is resolved, there's just a bit of a lump there still which I have to figure out - this is probably the gem you're referring to

Those were also my thoughts exactly on how to get more CAN variables transmitted. RPM and speeds (vehicle, individual wheels), as well as a whole host of other DSC data are already on the CAN bus. I’d say the most useful to add would be relative opening, lambda integrators and the engine and ambient temperature/pressure parameters that aren’t already on the CAN bus somewhere else. Honestly they might all fit in a single CAN transmission.

edit: here’s a json config file that you might enjoy perusing because it has all of the CAN values I log in it

Yeah I've been looking most closely at the lambda integrators and RO, as you say the additional temp/pressure params would be good as well. Only other thing I realised is the lambda functions at least are executed on the Slave CPU and I'm not sure yet whether they are sent to the shared bus/memory space for the Master CPU to access. If they aren't that will be another complexity to figure out as I haven't dived into exactly how that shared bus/memory space works and if there's any room left on it.

Thanks very much for the config file - that's very useful!

[Bry5on]

Hell yeah! This reads like I felt.

Did you get through high load, low rpm testing? Specifically around 2k rpm? There might be a gem waiting for you to discover.

Those were also my thoughts exactly on how to get more CAN variables transmitted. RPM and speeds (vehicle, individual wheels), as well as a whole host of other DSC data are already on the CAN bus. I’d say the most useful to add would be relative opening, lambda integrators and the engine and ambient temperature/pressure parameters that aren’t already on the CAN bus somewhere else. Honestly they might all fit in a single CAN transmission.

edit: here’s a json config file that you might enjoy perusing because it has all of the CAN values I log in it

config-11-20-24_moreCAN_clean.txt

[karter16]

I was back to work this week running v5 of my tune with 5 rounds of VE tuning on top of it. It's given me the opportunity to drive a familiar route and get some good comparisons, as most of my recent driving has been VE tuning runs.

Firstly - low rpm, round town drivability is now, believe it or not, BETTER than it was for me with stock airbox and stock program/tune! (to be fair I also cleaned the TBs, ICV, etc. at the same time so need to consider the impact this will have had also). Needless to say I'm extremely happy with how this is turning out. Drivability concerns are what put me off doing the CSL airbox for so long, so I'm stoked that this approach has turned out so well. I'm not sure if it's that just my car wasn't completely happy with the standard 1901 tune my car came with or what, but I guess I'd got used to avoiding certain RPMs at certain loads and speeds and all that is gone. I can cleanly and smoothly run up and down the range in first gear in stop start traffic, I can stay in higher gears at lower speeds than I would have and it's smooth sailing. SMG shifts and downshifts are excellent and repeatable. On the euro software (both with and without the CSL255 parameters) I couldn't be completely confident that I wouldn't get a bit of a jolt sometimes downshifting from 2nd to 1st with my foot off the accelerator. Now all that is gone completely. Not to make out that there aren't still improvements I can make, there's a couple of rpm/load spots I need to smooth out some more, but there's not many of them left at this point. Another scenario I often encountered previously was the tendency for the car to jerk under anything more than mild acceleration for the first 30 seconds or so after startup when completely cold. That's gone as well and drivability is smooth from completely cold.

When starting out with this I would have been stoked to have ended up with the drivability being no worse than it was on the Euro tune, to have ended up with something even better is magic. I'm looking forward to continuing to fine tune.


Relatedly I've been making some good progress with disassembly and have started looking at options around pushing more useful values onto the CANBUS. Looking at the address tables, functions, etc there doesn't seem to be any particular reason why it wouldn't be possible to add a function for an additional ARBID to be inserted onto the bus. If given an appropriately low priority ARBID it shouldn't get in the way of more critical signals. The bit that would be most difficult to figure out (apart from the bit where you DON'T brick your DME) is the easiest way to inject the additional function call into the flow. Looking at the 10ms task it would probably be easiest to take an existing function, move that function and use it's original address to build out a new function that calls both it and the additional function for the new ARBID. Unless I can find some spare bytes somewhere to inject an additional JSR (no doubt there's others here with more skills than me at identifying injection points for patches, so sing out if you have any suggestions). Other pre-reqs to this include tracing the variables that would be useful to send (things like lambda integrators, rpm, relative opening) and following the program flow to confirm the actual internal refresh rates for those variables (some of these are defined in the funktionsrahmen, but some aren't so I'd like to confirm for peace of mind).

[M3Garand]

This is a nice clean build

[proto.corsa]

Killer build.

[0-60motorsports]

The only one I've seen is this closeup Tom shared on cslregister: http://www.cslregister.com/forum/showpost.php?s=e577eebadd58ef42641b29ff5af9ace1&p= 202139&postcount=5

Oh yeah thats pretty bad! Thanks for the reply. I am trying to see who can locally print me your adapter in the right material. Thanks for that design!

[karter16]

Does anyone have a picture of the new part number 12528091765 installed on the CSL rail?

The only one I've seen is this closeup Tom shared on cslregister: http://www.cslregister.com/forum/showpost.php?s=e577eebadd58ef42641b29ff5af9ace1&p= 202139&postcount=5

[0-60motorsports]

Hey so just again with the caveat that I haven't done this yet, and what I'm doing is just based on my own research (I am not an expert).

Yeah that's my understanding. The original CSL MAP sensor (13627833561) is ideal but is NLA. The BMW replacement (12528091765) is a bit of a hack job and an aesthetic disaster. As you note the discussion on CSL register identifies 0261230101 as a possible replacement and I've decided to go down that path. It should be noted that in that discussion Tom notes that the nose on the sensor is approx 1.5mm too long. The sensor also has 1x mounting hole rather than 2. Once I have the parts in hand I plan to check this, and if it is indeed the case that it needs to be stood off slightly, I intend to CAD design a spacer which the sensor can be glued to to hold it in the right position. If this works out then yes, I think it will be a better option than the BMW replacement item, but that depends on me being able to make it fit properly.

Does anyone have a picture of the new part number 12528091765 installed on the CSL rail?

[wahsm]

FYI (in particular ugaexploder and wahsm) I've published the STL for the MAP sensor adapter - new post in the sticky: https://nam3forum.com/forums/forum/main-forum/e46-2001-2006/271707-e46-m3-community-designed-cad-files-master-list?p=290245#post290245

Much appreciated, thank you!