E46 M3 highway shutdown: tach/speedo drop, DME unreachable afterward

[OldRanger]

Here where I’m landing on this:

Observation — EKP Module and Fuel Pump Relationship on High Mileage E46 M3’s

Two cars offered an interesting pattern worth sharing.

A few years ago, on my 2005 with 162k miles, I installed a new EKP module to address a fail-to-start on first crank and a fuel pump relay code. About a week later the original high mileage fuel pump failed. I didn’t even record them as separate events in my maintenance log.

On the 2002 with 182k miles, I installed a new OEM fuel pump, filter, and regulator ahead of a dyno session (to refine a base performance tune we’d been running for a bout a year). One week later the original EKP module failed (thermally) on the highway. Then failed again a second time after some misguided attempts to address a DME power supply issue I thought I was having.

It’s plausible that in both cases the new component — performing as designed — placed greater electrical demand on an aged partner than that partner had been experiencing. The degraded component simply couldn’t meet it.

I can’t prove it from two data points. But the pattern across both of my cars suggests that on high mileage examples these two components may be worth replacing together when either one shows signs of age or fault codes.

Curious whether others have observed something similar.

[sapote]

It’s plausible that in both cases the new component — performing as designed — placed greater electrical demand on an aged partner than that partner had been experiencing. The degraded component simply couldn’t meet it.

I can’t prove it from two data points. But the pattern across both of my cars suggests that on high mileage examples these two components may be worth replacing together when either one shows signs of age or fault codes.

Curious whether others have observed something similar..

New e-relay should not put more stress on the pump. Regardless new or old e-relay the DME calculated the PWM duty for the pump supplied voltage based on rpm, MAP, throttle, and so the pwm signal should still swing from 14v to 0v at the calculated pwm. If the signal is not a clean square wave from 14v to 0v then the power FET inside the e-relay would burn up really fast due to heat from too much voltage dropped across the FET.

[OldRanger]

New e-relay should not put more stress on the pump. Regardless new or old e-relay the DME calculated the PWM duty for the pump supplied voltage based on rpm, MAP, throttle, and so the pwm signal should still swing from 14v to 0v at the calculated pwm. If the signal is not a clean square wave from 14v to 0v then the power FET inside the e-relay would burn up really fast due to heat from too much voltage dropped across the FET.

Sapote - thank you for the clarification on the switching behavior. To make sure I’m describing this correctly for the US spec car — my understanding is that the DME manages fuel delivery demand through the EKP module, and the EKP module’s Power MOSFETs execute that signal to the pump.

When the MOSFET degrades it can no longer switch efficiently, generating heat internally as a result of its own inefficiency rather than any external stress from the pump or the DME signal itself.

If that’s an accurate description of the failure mode - I’ll update my mental model, and write off my personal two-car pattern as a coincidence. Two aged components on high mileage cars failing in proximity rather than changing one causing a failure in the other.

Appreciate the precision as always.

[terra]

To my recollection, there's a decent sized capacitor inside the 'relay', presumably to smooth out the power delivery to the fuel pump. I wonder if that could be the failure point.

Also came across this thread of the EKP module failing after forgetting to connect the spark plug ground. Failure mode not quite the same as yours, but might be worth double checking that if you haven't already https://nam3forum.com/forums/forum/m...ring-schematic

[OldRanger]

Pictures.

[OldRanger]

For those interested in what’s inside the failed module — I opened it up.

The power switching component is a BTS650P — an Infineon Smart High-Side Power Switch.

Also present is a 470µF electrolytic capacitor dated 02.04 — original to the module.

Visual inspection of the PCB shows no catastrophic failure — no burnt traces, no obviously failed solder joints, no visible damage to the BTS650P package externally.

Curious whether anyone with electronics experience recognizes a known failure mode in this specific component on high mileage applications.

[sapote]

The BTS high side switch (also called semiconductor relay) has an N-channel power FET to drive the pump. 14v supply is connected to pin4 and the body soldering tab (also for heat dissipation), and I bet this pin4 is connected to the 470uF for stabilize the 14v supply on the board. Since there is long wire going from 14v source to pin4, so w/o the cap acts as reservoir the BTS would be starved from 14v whenever it switches on the power FET to drive the pump.
The 3 pin device below I think it's a dual diode connected to the BTS output for the flywheel effect when driving a massive inductor (the pump motor). Without the flywheel diode to keep the inductor current continuously flow even when the FET turned off during pwm cycle, the inductor current would kill the FET fast, based on Lenz law. Can you read the part # on this part?

The square wave PWM voltage will be filtered by the pump motor inductor coil, so if you can display the current on the scope then it's DC current with some ripples but not the whole pwm signal.

If the BTS is partially damaged as in your case, put a finger on the device while it's driving the pump and feel the temperature. Cool is normal and hot is bad device.

[sapote]

From the pics it's fairly easy to repair a damaged e-relay, so people should save them for repair if they are NLA, or too expensive. Parts are standard and availlable from Digitkey or others.

[OldRanger]

sapote The 3-pin component is a 12CWQ04FN — International Rectifier dual Schottky rectifier, common cathode, 40V 6A per leg. Confirmed freewheeling diode exactly as you predicted.

[OldRanger]

From the pics it's fairly easy to repair a damaged e-relay, so people should save them for repair if they are NLA, or too expensive. Parts are standard and availlable from Digitkey or others.

Worth mentioning: both replacement modules I’ve sourced recently were dated 2016, suggesting production may have stopped approximately a decade ago. Remaining new stock is aging NOS inventory, which makes the repair you’re describing increasingly important.

From what I can document on the failed module from my 2005:
∙ BTS650P — Infineon Smart High-Side Power Switch
∙ 12CWQ04FN — IR/Vishay dual Schottky rectifier, 40V 6A, DPAK
∙ EPCOS B41683, 470µF electrolytic — voltage rating not visible, dated 02/2004

Would you be willing to confirm the complete component list including the capacitor voltage rating? That information documented here would be genuinely valuable for the community long term.

[terra]

Seems like the capacitors come in a lot of different voltage ratings. Could probably confirm with the physical measurements in the datasheet (attached) for the various variants. Guestimating with photoshop and the dimensions of known components, I think it may be the 63v variant, but that's definitely a guess on my part

Does seem to be NLA in general so would need to find an alternative, but that shouldn't be a tremendous issue

[sapote]

Worth mentioning: both replacement modules I’ve sourced recently were dated 2016, suggesting production may have stopped approximately a decade ago. Remaining new stock is aging NOS inventory, which makes the repair you’re describing increasingly important.

From what I can document on the failed module from my 2005:
∙ BTS650P — Infineon Smart High-Side Power Switch
∙ 12CWQ04FN — IR/Vishay dual Schottky rectifier, 40V 6A, DPAK
∙ EPCOS B41683, 470µF electrolytic — voltage rating not visible, dated 02/2004

Would you be willing to confirm the complete component list including the capacitor voltage rating? That information documented here would be genuinely valuable for the community long term.

BTS650P is NLA but this part here should work for $5 https://www.digikey.com/en/products/...AATMA1/2080778
The rating of the new part is lower than the original part which I think it's over spec anyway. 38V max is perfectly fine for 14v system, and with the 470uF cap it should keep the 14v supply ripple low.

The 470uF should be easy to find replacement. Give me the length and dia of the stock part and I will find it. The flywheel diode is easy to find.

[sapote]

This is a better part but with a few pins need to be swapped around, but the good news is that the 14v power tab and pin4 are the same as the stock part, so it's not difficult to use, and cheap:

https://www.digikey.com/en/products/detail/infineon-technologies/BTS500251TADATMA2/9693062

[OldRanger]

I ‘disassembled’ mine so you guys could see more.

Measurements checked using (my cheap) digital calipers — average measured diameter 14.2mm, length 29.4mm. Nominal 14 × 30mm. Voltage confirmed 63V visible on reverse side. Matches B41683 470µF 63V exactly per the datasheet Terra shared.

[terra]

How far apart are the holes on the PCB would you say? TDK / Epcos' closest replacement would be B41693, but seems like the 470 µF variants are 39mm long (and 18-20mm diameter).

Otherwise there are alternatives such as https://www.mouser.com/ProductDetail...MAL213828471E3 -- though most of the alternatives I'm finding have somewhat higher ESR. How much that matters for this, I'm not sure.

For a lot more money some of the Tantalum options are interesting. But I think I'd just swap to a mechanical relay before going down that path.