Mercedes 190e M102 UAEFI conversion

[Giobs]

Hi everybody!
I'm Gioele, an Italian almost mechanical engineer (just finishing my studies, i'm still at uni) with a strong passion for both old cars and newer technologies.
I've owned this Merc for 4 years now, bought with the payouts of my first side job and it has always been a pleasure to drive around.
2 years ago i started growing an interest in powertrain related affairs (i'm a chassis guy usually), and have been wanting to convert my car since then, initially to MS but then decided to go with rusefi since it was a far more affordable option.
So yeah, 2 years in planning, untill now, when i had 2 weeks off and the conversion finally begun.

Screenshot 2025-10-05 232513.png

this is the part list.

20250925_123451.jpg

here's the UAefi wired into it's custom case, you can find wiring info in the attached PDF file.

20250928_102557.jpg

20250928_102601.jpg

20250928_130410.jpg

Here's the lambda, CPS and toothed wheel mounted on the car; the bung for the o2 sensor has been tig welded in the nearest position to the cylinders as possible; the toothed wheel has been mounted on the stock vibration dampener after having it lowered by 5mm, so it's squeezed between damper and pulley. The CPS is placed at 0.8mm from the wheel which has ben turned on a lathe after being laser cut to achieve as little runout as possible.

Then it was the turn of the intake manifold, got it disassembled and cleaned, then modified to house the rail/injectors, MAP/IAT sensor as well as the TPS (which is giving me some issues right now, but lets go on).
here are some picture. the mounting for the throttle link and spacers have been manufactured using the infamous "eyemeter", a grinder and a lathe.

20250930_200937.jpg

IMG-20250930-WA0017.jpeg

20251001_010535.jpg

20251001_010533.jpg

20250930_200941.jpg

(i think i reached the max number of pictures on one message, keep scrolling)

[Giobs]

Ok, let's keep on rolling.
Now, everything assembled on the car:

20251001_204839.jpg

It was time to manufacture the harness and the coil support:

IMG-20251003-WA0018.jpeg

20251004_133559.jpg

20251004_023126.jpg

20251004_023120.jpg

.
The fun part sadly ends here, once i got it to start and idle well (it required a full day since i misconnected 2 spark plug wires) i was starting to taste the feel of getting the conversion done! but that's were some misterious issues started appearing.

First issue, my board is plagued by the 5v loop issue that is affecting other boards, but soldering the 10k resistance to wake up the regulator worked in solving this issue.

Second issue, while i was still troubleshooting the spark plug wires issue, i got an injector (B6 output) stuck open (output from ecu always grounded), from which i got an error in tunerstudio (which showed only once, never reappeared).

20251004_202302.jpg

and i found resistance R29 had gone from 4.7Kohm down to 0.7 Kohm for no reason that i can see (i work with some powertrain guys from the powertrain division of a formula student team, we usually use motec but the way the ecu was handled and tested was nothing out of the ordinary).
You can see here the measured voltage across the 12v and B6 pin (which wasnt active at the time of measuring).

20251004_213852.jpg

I since then switched the injector 1 output from B6 to B2 to continue to use the board, and did a 45 minutes tuning session with it, everything seemed to work fine untill, after a pause, i restarted the engine and in 10 seconds in i lost both cyl 1 (output B2) and cyl 4 (output B3) injectors output.
Now the ecu shows 5.5v across these pins, no errors and testing the outputs through software doesnt change the voltage measured.
the injectors all measure 16.something ohms.
I still have to investigate the issue, but i think i'll find other resistance shorted (maybe they were defective? idk).

anyway, attached the schematics and pinout, also for other users to maybe take inspiration if needed.

EFI_wiring.pdf

If you have any kind of suggestion on what to check / change on it to make the board reliiable i'm all ears; i would happily buy another one (hoping for more reliability) but being in italy makes it kinda hard...
Anyway will have to figure it out since i need to get the car back on the road.
Thank you for your attention

[AndreyB]

https://github.com/rusefi/uaefi/issues/95

can you take a good photo of one of the injector driver chips? just want to confirm which part number your specific unit has. those are SOIC-8 looks like those are all on the back side

[Giobs]

Ok so, mercedes M102, 4 cylinder 8v engine with a fully custom wiring and UAefi, using wasted spark and batch injection since i have only a sensor on the crankshaft (VR), you can see the complete list of parts, and the wiring scheme on the forum.
Lets go directly from the first start (3 weeks ago), on the first board (rev E) which went mostly according to plans, a few issues regarding idle and the jumpstart required for the 5v rail, a few other things but it ran completely fine to be a first install and forst start. It kept running for at least 10 minutes before i shut it off to finish other works before tuning.
Then tuning started, i started tuning the lower zone of the pressures since i didnt want to run the car on the road yet and everything went smoothly for the first tuning stage. I then killed the engine since i wanted to fix the 5v rail issue, so i soldered the 10kohm resistor on the mmu and the board started to work again without the pc.
So i then went ahead to keep working on the tune, started the car, let it get up to temp and then... disaster struck... after running fine for the past 10 minutes it started running on 3. I immediately noticed it was because of a missing injector since i could read higher afr values.

So i took the board down and started doing some benchtest on it, finding out that one of the outputs (in this case B3 IIRC, injector 4 output), started misbehaving badly, since when measuring deltaV between pos. And inj output i could measure 1.6V at non solicited (a working output measures 1.1V) and 4.7V when solicited (where a healthy output reads Vbatt).

Now that was particularly strange, it simply didnt make sense, as the driver was still working perfectly fine on one of the internal channels, while misbehaving (but sill giving signs of life) on the other one.

So thats how i worked around it: since working on a 4 cyl i still had 2 healthy channels that i could try before having to start fisical repair.

I hooked up inkector 4 to B2 and restarted the engine (after changing outputs in tunerstudio).
It worked fine, for at least 5-6 minutes and then it happened again.
It didnt make, again, any kind of sense, and that was when, while i was looking for a solution with the engine still running, i started detaching injectors cables one at a time, starting from injector 1 (B6), which resulted in another kind of damage to the driver, since from then the output B6 is always to ground.

To recap, first board now has 3 healthy outputs and 3 bad ones, of which:
-1 (B6) is damaged by a disconnection of rhe injector while running and now is always grounded.
-2 (B3) is affected by not being able to completely ground and only measures 4.7V when solicited, but idles at 1.6v (different from the 1.1V of working outputs).
-3 (B2) same as B3.

I then ordered some new boards and in the meantime, also some new drivers to try out from mouser.

The boards arived today 19/09, before the components for the repair of the first one.

So i hooked up everything, updated the firmware to the latest build and loaded the tune, and then i fired up the engine. IT WORKED. for 10 minutes.
Then... disaster struck again... injector 2 (B5) started acting exactly as other otputs did on the first board....
I was now without ideas, only one thing was missing to test/blame. The injectors. Code 0280156023 they come from a saab 9.3 and likely have a ton of miles on their back. They all measure a bit over 16ohms (all within 5%), both measured at the injector or at the ech connector side. I decided to try a set of 0280158038, brand new bosch originals.

Again, swapped to a healthy output, hooked everything up and the loop repeated. 10 minutes it ran fine then one output lost, again. this time it was injector 4 (B3) that went.

IMPORTANT NOTE: As it can be seen, the injectors are wired individually to individual outputs, and batch injection mode is selected through software. A message is also present on the software, telling to individually wire in the injectors for batch, and that rusefi will handle everythin correlated to the firing order.

Attached the required picture.

Screenshot_20251019_034154_Photos.jpg

[AndreyB]

telling to individually wire in the injectors for batch, and that rusefi will handle everythin correlated to the firing order.

this seems to contradict what I see on one of the dialogs, could be an error in documentation. What's the source of above statement?




if injectors are individually wired I am puzzled how are we damaging vnld5160

image.png

image.png

[AndreyB]

I.e. my understanding is that safest mode would be sequential injection (with random phase) since that would be true individually wired with all injectors firing. As always, there is a chance that I am very confused.