M42 Head Gasket Blown, Project Thread

[ForcedFirebird]

LTW FW generally will have a falling idle at a stop/clutch in when "too" light. And yes, any Getrag is a joy to drive with a LTFW, bit noisy as all get out. Not just BMW, you can Google "Getrag Rattle" and it's all makes/models - Dodge, Chevy, International...all of them....

[bmwman91]

OK, the issue is mostly resolved. The car runs, and I had a good time driving it for 30 miles, running it about as hard as hard as I could while keeping it under 5000RPM (I am going to put more like 100 miles on it before really revving it just to be sure the rings are worn in). It turns out that it was a software issue after all. I have a corrected high performance Sssquid tune in the car, which is what he has developed across something like 8 other 2.1L M42's from MM. After driving a turbo diesel for 7 months, I have to remind myself that punching the throttle at 2000RPM isn't going to do much haha.

With that said, the idle still has issues, but they are relatively minor compared to what I was dealing with. They seem to be heavily dependent on engine temperature. I have outright removed the LLCO wire from the Motronic connector, so I know it isn't that lol. Anyway here's where the idle seems to be at now, in the order that I encountered the issues:

1) The car did the thing where it stalled after running for a minute as I pushed the clutch in coming to a stop sign. It started right back up and did not do it again (for a while).
2) Once it was warmed up, it would stumble and catch itself as I came off the throttle approaching most stops, and it would intermittently exhibit a 800~900RPM hunting when sitting at idle.
3) I drove it to visit my parents, and it sat for maybe 30 minutes. When I got in to start it, it did not want to start. As it cranked it just barely gave an indication of a couple of attempts to fire, and a little throttle was needed to get it to actually start. The temp needle was a little shy of the 1/4 mark at this start.
4) As I slowed to turn in to my driveway ~20 minutes after #3, it stalled. I again needed to apply a little throttle to get it to start, and it stalled again when I let off the gas to park. I again needed to give it a little throttle to start it.

I still have the stock AFM in there, and it has the little extension harness / low-pass filter that BMW issued to most M42's at the time. I am replacing the CLT sensor when the new one comes in later this week, although the current one seems to Ohm-out just fine. Also, I assume that the super light RHD flywheel is not helping anything. I will still be re-checking for vacuum leaks and messing with bypassing the ICV as suggested since the above issues are a lot more similar to ones people have with more "common" faults in these cars.


For the record, the 7.4lb flywheel is fun as shit. Effortless rev matching, no weird noises at start (some people said that the RHD flywheel sounded raspy, but I do not notice anything) and crazy good response. To hell with the noisy rattling from the transmission, BMW should have shipped these cars with light flywheels!

[Victell]

The issue may be mechanical within the engine. Something may be happening internally with clearances and whatnot when warmed up. I would spin the engine by hand cold, then try again when hot.

[bmwman91]

A thing I have done in the past is use an Arduino to tap directly into the UART lines for the microcontroller and communicate directly that way. It is then all simple TTL signaling with a total bypass on the one-wire K-line interface. You probably saw my threads where I totally reverse engineered M1.7 and M1.7.3, so it was fairly easy to determine how to do it. The same deal applied to the L-line, which was only used for a one-time 100-baud handshake to make the ECU enter diagnostic mode. No need for any level shifters or one-wire logic that way. That is how I dumped the 8K ROM on the MCU, allowing me to form the full 40K binary (8K MCU ROM + 32K UV-EPROM) and get IDA to disassemble it properly. The 8K on the MCU is all of the interrupt vectors, which are needed to be able to trace anything since the program counter needs to be directed correctly from the start, and I found a very old HTML site with a list of commands and the protocol for telling the ECU to dump blocks of ROM and RAM.

Anyway, you probably knew all that already. The point is that I could also solder some wires from the various TTL-side lines of the L- and K-lines to unused pins on the Motronic connector, and route those wires out to an Arduino. It is my understanding that all of the axis descriptors for the maps (3B, 4A, etc) are actually RAM addresses for those very parameters. So, many of the runtime variables that I'd want to watch are already pretty well known in terms of location and scaling into physical quantities. The stock K-line interface is slow, but it would be easy enough to use VB.NET to make a basic interface to poll the values and display them.


The other test I am going to do, along with you and digger's recommendations above, is to pop the hose for the ICV from the intake boot and see what happens if I apply ~5PSI. I expect a little leakage from the TB, but if there is some sort of major gap elsewhere I should be able to hear it.

[ForcedFirebird]

I haven't experimented with logging Motronic since I use an emulator, there's data tracing (kinda like "follow mode" in aftermarket ECU's), so you can see what cell the ECU is using on the various maps, and make corrections.

[bmwman91]

i agree, but you need to make sure there arent multiple issues so id verify when the ICV is totally isolated from the engine that there are zero other issues

If you are logging, then remote is fine. I was tuning an m20 on a Dyno in Denver last Sat using the Dyno plot and logs.

Digger is onto something. At work, we often will disconnect the ICV and plug the hoses for diag. A 3mm vac hose going from the intake boot to manifold will let the car idle, and keep the air metered via AFM. Just don't expect to rev it much until it warms up, the ICV opens quite a bit on tip in, when in closed loop. It will want to stall with throttle input during warm up.

Sounds good, I will keep with the stock tune, stock AFM and experiment with removing the ICV from the loop. Once I get the engine warmed up, I will try pinching-off the ICV feed line to make sure that air is not getting in elsewhere. If that checks out, I will replace the ICV with some spare 3.5mm vacuum hose I have left from replacing the FPR vac line.




Firebird, do you happen to know if the INPA/EDIABAS software will be of any use for reading various runtime parameters in real time? I have the "Factory Interactive" service manual, and it has a decent amount of troubleshooting guidance for the various signals, if one has the diagnostic software. It looks like the INPA software is fairly easy to find around the web, and K-line interface adapters also seem to be reasonably cheap.

I also found the PN for the 88 pin breakout box for testing all of the signals in/out of the Motronic, but if anyone actually sold them they would not be worth the cost. Someone in Eastern Europe has to have old BMW dealer stuff...if only I could find them!

[ForcedFirebird]

If you are logging, then remote is fine. I was tuning an m20 on a Dyno in Denver last Sat using the Dyno plot and logs.

Digger is onto something. At work, we often will disconnect the ICV and plug the hoses for diag. A 3mm vac hose going from the intake boot to manifold will let the car idle, and keep the air metered via AFM. Just don't expect to rev it much until it warms up, the ICV opens quite a bit on tip in, when in closed loop. It will want to stall with throttle input during warm up.

[digger]

It seems that there is some erroneous signal somewhere that is confusing the Motronic and making it operate the ICV incorrectly. Evidence I have so far indicates that the ICV's operation is a symptom rather than the illness itself.

i agree, but you need to make sure there arent multiple issues so id verify when the ICV is totally isolated from the engine that there are zero other issues

[bmwman91]

There were probably a few seconds here and there of MBT when driving it around since I was aiming to do at least a little break-in. It was all 2000-4000RPM with one or two little bursts to almost 5000 which did not feel great, but I did not hear any indication of detonation. But yeah, no more of that until it idles properly.

I had some pretty wonky tunes provided remotely very early on and the car ran fairly well with those compared to what it is doing now. I just hope that all of my re-wrenching and probing will turn up something conclusive. Tunes that Sssquid came up with for engines built to the same spec might not be perfect for my individual engine (which is why I was doing tons of data logging and sending it to him), but it would at least run "well" I think.

Maybe I just need to forget about this for a week and spend time doing other stuff. I think I am burning out a bit lol.

[ForcedFirebird]

A modified engine will start, run, and idle with a stock ECU. Just don't get into MBT, more than likely, the stock table will be too advanced.

That being said, return to stock, or go stand alone for easy diag.

Just an FYI, the Dyno plot for the squid stock upgrade is my Dyno sheet, sent it to him when testing the second m20 ITB car we did back in 14(?) when doing the base runs....

https://www.r3vlimited.com/board/forum/e30-technical-forums/engine-drivetrain/m20/300937-200whp-with-stock-n-a-rotating-assy?p=5997267#post5997267

I prefer NOT to sell chips on the interwebs, though, live/remote and/or local only. Never was a believer in "stage 1,234,567”. Every engine has specific demands IMO.

[bmwman91]

I have experimented a little with the ICV. In the cases where the idle was too high (stable or oscillating), unplugging the ICV would make the idle return to normal. As mentioned, the AFM output was indicating that the ECU was purposely opening it that much for some reason since it looked to be the case that the extra incoming air was all metered.

I have not tried unplugging the ICV when the engine wants to stall, but it makes sense to try. If the ECU sometimes decides to open it too far, maybe the stalls are because it is not opening it enough?

I can try pinching off the hose to the ICV when it is idling too high, just to rule out any other source of incoming air. At this point, I am 99% sure that the ICV itself is not the culprit. It operates smoothly on the bench with a variable power supply and does not stick at all. I have tried 2 different ICVs to no avail. I also tried swapping in a spare ECU and it also did not improve things, so the mid-current driver stage that drives the ICV is also not the issue.

It seems that there is some erroneous signal somewhere that is confusing the Motronic and making it operate the ICV incorrectly. Evidence I have so far indicates that the ICV's operation is a symptom rather than the illness itself.

The engine runs too smoothly at all throttle positions from 1000-4000RPM for me to suspect a physical issue with the intake or fuel supply system.

[digger]

I’d get it warmed up and block the icv to see if there’s any change. Presumably it’ll run ok without it. That will isolate the issue to the icv or something else

[roguetoaster]

Well, until I read to the bottom I was thinking "where's the vacuum leak," having arrived at the bottom I nominate TPS connector intermittent strain based wiring fault as the likely culprit. How's operation without the TPS plugged in?

Also, is your evap purge valve still present?

[bmwman91]

As crafty as you are, surprised you don't just get an emulator and tune it yourself.

Nah, after working with Sssquid for a while now, I know I will get much better results from someone who has thousands of hours of experience with tuning. And, Motronic is enough of a pain in the ass that a stand-alone solution is where I'd go if I really wanted to have proper control of things.


Speaking of having control of things, the damn car still does not run properly. It is perhaps not a software thing after all. I am going to lay out what all I did today and hope that the good folks on here can help me figure this one out, because I am thinking that it is going to be something pretty obscure. Everything is on the table, including faults in the wire harness since I had it totally apart for cleaning & restoration.


- Sssquid sent me a bone stock tune with the only change being what was needed to properly run 28# injectors. The car started and idled properly from a cold start, and I drove it a few miles. About a minute after that cold start, it would stall coming off the throttle (such as at a stop sign) unless I have it a little throttle. The idle would sometimes stabilize if I was careful to really ease off the throttle.
- I swapped out the MAF conversion for the old stock AFM (conversion is plug-n-play) and gave tat a shot with the stock tune. The engine was still fully warmed up. It idled better, did not stall at all, and the only hiccup was a small stumble here and there coming off the throttle.
- After the engine fully cooled off, I started it with the stock tune and AFM to pull it into the garage. It would stall without me giving it a little throttle.
- When the engine was fully warmed up still, I popped in some of the other high performance tunes from Sssquid. They are basically the exact ones that he has developed and successfully run on other 2.1L M42 engines from Metric Mechanic. The car would start and the idle would oscillate between 1100-1600RPM. I did not try driving it, just idling.
- With the engine cold and the performance tunes in, it will idle steadily at ~2100RPM for a minute or two, before starting to oscillate.

I checked the resistance between terminals 43 & 78 on the Motornic plug (the two lines for the coolant temp sensor) and it came out to 400 Ohms when the engine was still luke-warm. Wiggling the wires at the sensor plug did not change the value, so there is no broken wire. I am replacing it anyway since my problems seem to have some link to engine temperature, but it seems unlikely that the sensor is bad since it is only about a year old.

I have replaced or swapped the following items while troubleshooting the oscillating idle to no avail:
- ICV
- throttle body
- TPS
- ECU
- AFM (swapped with MAF conversion)

Sensors & stuff that have been replaced within the last year include:
- Crank position
- Cam position
- Coolant temperature
- Crank damper wheel (60-2 toothed pulley)

Before the HG blew late last year, I was having an issue where the car would stumble and sometimes stall about 1 minute after a cold start, and it would go away after another minute as it continued to warm up. That particular behavior probably started around the beginning of 2019. I assumed it was the tune, but maybe not. I am unsure if it is related to my current issue since the entire wire harness is now a different one, damn near everything has been replaced and/or swapped and I can't think of anything that does not check out properly on my freshly rebuilt engine.


Prior electrical measurements indicate that this is NOT a vacuum leak. First, I have had vacuum leaks before, and they do not make the idle enter a constant oscillation pattern as far as I have ever seen on an E30. Stalling and a stumbling idle, sure, but not the oscillation. Anyway, the air flow signal at idle when it is idling either too high or oscillating is around 2.25V, which means that the ICV is being opened too much. Since I have replaced the ICV, cleaned it very well, checked the wiring connections, and verified that the signal looks about correct with an oscilloscope, I think that there is some issue with one of the inputs to the Motronic which is causing it to purposely open the ICV too much. As far as I know, the only things that are going to have an effect like this on the idle control are the coolant temp sensor, AFM, air temp thermistor and TPS.

At this point, I think that I am just going to have to yank the intake and fuel rail for some inspecting, take apart the big Motronic plug to verify that nothing got yanked when I rewrapped the harness, and probe through every connection from the Motronic plug to each endpoint. I am at a total loss otherwise. The engine runs well enough when not idling that I do not think it can be some sort of major cam timing issue or other mechanical fault.

Any and all thoughts and insight are appreciated.

[ForcedFirebird]

As crafty as you are, surprised you don't just get an emulator and tune it yourself.