Troubleshooting a no-crank issue

[E46m3zcp]

Just bumping this because there were unrelated replies, and it could make my last reply with the latest development less visible.

As of now, I have ruled out the battery and the starter, and I'm investigating an SMG-related issue that could make the clutch position not satisfactory for the EWS to unlock the starter (more specifically, it looks like it is periodically unlocking/locking it back too quickly for the engine to have time to start). Any hints from people having seen these symptoms much appreciated!

EDIT: I tried bleeding the SMG system with INPA, no progress. I am suspicious of the clutch sensor and slave cylinder, but unsure how to test that (I'd like to avoid removing the aluminum plate, as the bolts are TTY and technically need replacement).

Symptoms recap

• No crank.
• INPA “starter not powered because of …” toggles between Clutch actuation and None.
• Relay-like clicking sound.
• Bypassing EWS pin 1 & 2 makes the starter crank normally.
• Battery & starter are known good.
• Bled SMG (because of the clutch actuation symptom), but no-crank remains.
• No SMG fault codes currently stored.

SEE: https://www.youtube.com/shorts/E5rugnuuR1U

[E46m3zcp]

ok the end of the saga is a bit disappointing. It was my ignition switch that was failing intermittently. This mislead me when I was putting the fuse in the EWS pin 1/2 and the car was cranking. I noticed that once every 5 attempt on average it wouldn't crank, then the frequency started dropping until it wasn't starting anymore.

I was also mislead by apparent SMG symptoms based on observation in the EWS INPA module that the clutch signal was preventing EWS to activate the starter (see my video above). To rule out the clutch problem, I followed steps to code the EWS to ignore the clutch position:

https://www.e46fanatics.com/threads/diy-how-to-code-ews-to-disable-clutch-park-neutral-switch.1313143/

After coding this, the clicking stopped (it went to a no-crank without any clicking, typical from a bad EWS or Ignition switch). This means that the clicking was the SMG solenoid for the clutch, as the car was getting some signal from the EWS to start, but since the ignition signal didn't reach the starter the system was giving up and letting go the clutch, then cycling again (I don't really understand deeply the reasons behind that, it could be how it's programmed to operate when it's not getting a starter signal despite the OK from the EWS).

Once I realized that coding that out didn't solve the issue, I concluded that the problem wasn't related to the clutch position sensor (builtin into the slave cylinder on the SMG).

Then 2 possibilities were remaining: EWS or Ignition switch. See this video: https://www.youtube.com/watch?v=LUYgSFgc7-c
At this moment the ignition switch completely failed (it was until then intermittent, so very confusing).

Since I have 2 M3, the solution was simple, swap out the ignition switch and see if it solve the problem, and it did. I coded back the EWS to factory specs, and everything works now.

[sapote]

After your hack of bypassing the EWS, I get cranking and so my conclusion is that the problem was corrosion on the terminals.
.

What terminals? Key switch terminals? Bypassing EWS with fuse can't help with a bad switch terminal.
I think the problem is the EWS internal relay. Could also be its input signals (N, clutch, SMG pressure sensor, etc) but in this case the starter should not even clicking.

[sapote]

Well well,

After reenabling the EWS, the syptoms are back. But at least now I know it isn't the starter! (I could have just shorted EWS pin 1 and 2 and known this without removing the airbox, but now I know!)

I suspect some SMG components (like the slave cylinder) to be the cause. It looks like the clutch disengages for a brief period, but bounces back before the starter has time to spin. It sounds like the slave cylinder pushes the clutch, then lets it go immediately.
Take a look:

EWS: https://www.youtube.com/shorts/E5rugnuuR1U
SMG Digital 1: https://www.youtube.com/shorts/-_O0caQYSxM

Any reason not to leave the fuse in until I get some free time to tackle this? I guess that would mean I'd have to be more careful not mistakenly turning the key while the car is in gear (like good ol' 90's cars)

Bypassing ews can kill the key switch earlier as all 10A starter solenoid current flows thru the switch contacts. This is why the EWS has an output relay.

[sapote]

ok the end of the saga is a bit disappointing. It was my ignition switch that was failing intermittently. This mislead me when I was putting the fuse in the EWS pin 1/2 and the car was cranking. I noticed that once every 5 attempt on average it wouldn't crank, then the frequency started dropping until it wasn't starting anymore.

I was also mislead by apparent SMG symptoms based on observation in the EWS INPA module that the clutch signal was preventing EWS to activate the starter (see my video above). To rule out the clutch problem, I followed steps to code the EWS to ignore the clutch position:

https://www.e46fanatics.com/threads/diy-how-to-code-ews-to-disable-clutch-park-neutral-switch.1313143/

After coding this, the clicking stopped (it went to a no-crank without any clicking, typical from a bad EWS or Ignition switch). This means that the clicking was the SMG solenoid for the clutch, as the car was getting some signal from the EWS to start, but since the ignition signal didn't reach the starter the system was giving up and letting go the clutch, then cycling again (I don't really understand deeply the reasons behind that, it could be how it's programmed to operate when it's not getting a starter signal despite the OK from the EWS).

Once I realized that coding that out didn't solve the issue, I concluded that the problem wasn't related to the clutch position sensor (builtin into the slave cylinder on the SMG).

Then 2 possibilities were remaining: EWS or Ignition switch. See this video: https://www.youtube.com/watch?v=LUYgSFgc7-c
At this moment the ignition switch completely failed (it was until then intermittent, so very confusing).

Since I have 2 M3, the solution was simple, swap out the ignition switch and see if it solve the problem, and it did. I coded back the EWS to factory specs, and everything works now.

SMG clutch sensor has nothing to do with EWS enable/disable.
If the key switch is the root cause, then why connecting switch signal via the bypass fuse caused the starter to run?

[E46m3zcp]

Sorry, I made this thread a bit confusing. The car currently cranks and starts fine after cleaning up the ignition switch's contacts (opened it and cleaned inside). The issue appeared to be the ignition switch, but it was intermittent at first, and this confused me.

What terminals? Key switch terminals? Bypassing EWS with fuse can't help with a bad switch terminal.

When I wrote that, I had just cleaned the starter terminals and bypassed the EWS. The car cranked (because the issue was intermittent). I concluded the starter terminals must have been dirty, I was wrong. I reverted the EWS hack and the car no longer cranked. I was very confused.

If the key switch is the root cause, then why connecting switch signal via the bypass fuse caused the starter to run?

This is true and confusing, but it can be explained by the fact that the issue was intermittent. The ignition switch was failing, but it would sometimes work. It happened to work right after I had put the fuse, but putting the fuse didn't make it work, it was just a coincidence (to confuse me even more, it cranked 4-5 times in a row).

SMG clutch sensor has nothing to do with EWS enable/disable.

Then why is the clutch being reported as a reason for the EWS to disable the starter in INPA? See https://www.youtube.com/shorts/E5rugnuuR1U "P,N/US-clutch-switch"

Something that really still confuses me is the SMG clutch being reported as the problem preventing the EWS from starting, and the behaviour shown in the YouTube video above, where it is oscillating between clutch disengaged and clutch engaged states. I wonder what is going on: why is the clutch seemingly cycling between engaged/disengaged? What in a bad ignition switch would make the SMG system behave like this? If current isn't flowing through the ignition switch because of a bad contact, then why is the car even "reacting" when I turn the key?

Here is a theory: the ignition switch had corrosion, causing a voltage drop across one of its internal terminals. This made the voltage picked up downstream closer to a threshold that would trigger the starter and SMG system to begin the ignition sequence (clutch depressed, starter powered). However, powering the starter drew enough power that the voltage dropped below the control threshold, causing the sequence to abort. The starter no longer being powered, the voltage would raise again above the threshold and begin a new cycle.

[sapote]

Then why is the clutch being reported as a reason for the EWS to disable the starter in INPA? See https://www.youtube.com/shorts/E5rugnuuR1U "P,N/US-clutch-switch"

P,N/US-clutch-switch: a generic term that the EWS emited to indicate the reason it disabled the starter due to one of these reasons: P, N, clutch switch. In this case clutch switch means the 3rd pedal switch, not the SMG clutch sensor.

[E46m3zcp]

I think you may be wrong here. This reason (P, N, clutch) in an SMG is initially "YES", then when you crank and the SMG system depresses the clutch, it toggles to "NO". I think this means the SMG system does take into consideration clutch position just like the manual (although there isn't a position sensor, my best guess is the slave cylinder has a signal that mimics it). Of course, I could be the one getting this wrong, but this is my current mental model at least.

There is some coding you can do to tell the EWS to ignore P, N and clutch and when you do that, the EWS digital reading in INPA shows a constant "NO" for the P, N, clutch thing.

[sapote]

I think you may be wrong here. This reason (P, N, clutch) in an SMG is initially "YES", then when you crank and the SMG system depresses the clutch, it toggles to "NO". I think this means the SMG system does take into consideration clutch position just like the manual (although there isn't a position sensor, my best guess is the slave cylinder has a signal that mimics it). Of course, I could be the one getting this wrong, but this is my current mental model at least.
.

I'm curious of this too, so take a look at the 13pins EWS pinput and find no clutch sensor signal being used for SMG cars:

Note: pin8 is for manual box clutch pedal sensor.

For SMG, I think the ews doesn't need to know the clutch position status in order to enable the starter, but the SMG does disengage the clutch before cranking and this is why you see its state was changing as noted.

[E46m3zcp]

Could it be pin6? That would be a combination of the two enabling factors: car in N, and clutch disengaged. SMG tcu may be responsible for combining these factors

When I had the problem I could clearly see in INPA, that condition becoming true/false as I heard the clutch engage/disengage (and car remained in neutral)

[sapote]

Could it be pin6? That would be a combination of the two enabling factors: car in N, and clutch disengaged. SMG tcu may be responsible for combining these factors

When I had the problem I could clearly see in INPA, that condition becoming true/false as I heard the clutch engage/disengage (and car remained in neutral)

EWS pin6 is an output (A is out and E is in) to SMG, not from SMG to ews.

[E46m3zcp]

Are you positive that pin 8 is not connected on an SMG? The reason I'm still doubting we're missing something is that Clutch is listed as a reason to prevent the starter from spinning in the EWS digital values in INPA. Why would that be there if there was no clutch criteria? Could pin 8 be connected to something coming from the SMG slave cylinder? Or maybe to something coming from the SMG TCU?

Also, the SMG system does communicate with the EWS (from https://ia800902.us.archive.org/26/i...20SMG%20II.pdf).

Responsibility for starter operation remains with the EWS module. As the key is turned to the KL50 position, the EWS module waits for confirmation from the SMG II control unit that the shift lever position is 0 and that the brake pedal is depressed.

Then

Seeing shift lever position 0 and the brake pedal depressed, the SMG II momentarily supplies a switched ground signal to the EWS module to serve as a release to allow starter engagement. Upon receiving these confirmations the EWS module activates the starter.

Oh but then

The clutch is open (not engaged) during start-up operation.

So that suggests you are right, and that the clutch does not prevent the starter from cranking. The SMG system, once in N and the brake is pressed, will release the clutch, but the EWS has already sent the signal to the starter at this point.

But, I'm still very confused about that INPA "reason for not powering the starter" stating the clutch. The technical documentation makes no mention of the clutch being a prevention criterion. Could it be something that was only applied to the US cars and did not make its way into the documentation? (this theory comes from the observation that the INPA menu states "US clutch").

Also, once I coded that criteria off in the EWS using NCS Expert, I wasn't hearing the clutch actuator anymore.

[E46m3zcp]

Hey look at this diagram. "Clutch position Closed No start" seems to be in blatant contradiction with the paragraphs I quoted above, and it sounds like the clutch IS a prevention criterion:



I bet the document was edited when the clutch was added to the sequence for the US market, but the writer didn't read the text and just added a one line "The clutch is open (not engaged) during start-up operation." to account for the change.
Haha! Just speculating.

[E46m3zcp]

Also relevant to the discussion, although probably obvious, the slave cylinder has a PLCD to determine the clutch position:



It's interesting to note that other parts of the document allude to a potentiometer at the clutch, which contradicts this PLCD design, unless the two exist in combination to assess plausibility? I didn't notice a clutch position sensor when I worked on mine, though...

[sapote]

This adds more confusion (from Tis):

To start the vehicle, the gearbox must be in neutral.

When the vehicle is parked, the following situations are possible:

• Gear is engaged (1, 2 or R), selector lever is in position ”0”. If a vehicle has been parked in gear (1, 2 or R) and the selector lever is at position ”0”, the brake must be operated first and the selector lever set to position ”N” in order to engage the gear.
• Gear is engaged (1, 2 or R), selector lever is in position ”N”. If a gear is engaged when the vehicle is parked and the selector lever is at ”N”, the shiftlock must first be deactivated by pressing the brake. The selector lever can then be moved to position ”0”. The display in the instrument cluster then once again corresponds to the gearbox position. Thereafter, a shift can be made back to the neutral position.

If ”N” is engaged, there is no need to press the brake to start the vehicle. For the start, the SMG control unit automatically opens the clutch.

The document treats the selector at N and 0 as two separate cases. True or false? What about no need to press the brake to start?