Been a bit silent after finishing (and solving) my original thread regarding the fuel pressure issues with my car. Over the past few weeks, I've been enjoying the car now that it's working. Or I'd rather say was working, but I'll get to that.

The car is a real joy to drive. Comfortable, even in 300 spec still ample power, and all the driving aids are so relaxing on long trips. I have the full Distronic Plus with Traverse support, lane keep assist, brake assist etc. etc. So on the highway, you can literally let go of the steering wheel for roughly 10 seconds, the car will find its way, taking care of traffic etc. etc. After 10s it starts warning to grab the wheel. During traffic jams it's even more relaxed, as long as speeds remain below something like 30mph, the car will do all the stop-start-steer actions by itself. Surprised by this level of technology on a 11 year old car!

Some issues popped up, the brake booster seal to the pedal is leaking ever so slightly, so that's on the list for replacement. I already got the seal. There is some knocking noises coming from the front suspension during low speed over bumps. I've replaced the sway bar drop links, no improvement yet. So that needs further work.

The launch creader elite has proved to be very capable. I was able to disable the stop-start eco function and change the parktronic beeps to something more useful. Performed a full B service as that was due, replaced a NOX sensor that probably got destroyed by all the unburnt fuel entering the exhaust (see fuel pump thread).

Over the weekend I installed a Roadtop unit to get CarPlay functionality.



Installation was pretty easy. I decided to remove bits on the center console to lead the USB cable into the storage area there and I installed the Roadtop unit vertically between the glovebox and the NTG unit. Tested things before getting al the pieces back into place, all worked fine. However, after finishing the job, the car had a lot of issues all of a sudden. Keyless go failed, headlights turned on by themselves, rev counter started moving in strange ways etc. etc. Oh bummer, some CAN errors. I re-opened my dash, disconnected the Roadtop unit and all went healthy again. Hmmmm.. I accidentally touched the little dip switches on the Roadtop unit in the final steps of mounting it, changing its configuration and then creating havoc on the interior can. Luckily this went all back to normal with the dip switches in their right place.

The Roadtop unit works perfectly with CarPlay, it's a really worthwhile upgrade, I especially like the voice control through Siri!

Then I ran out of luck:


The fuel pressure issue reappeared all of a sudden after 1000 flawless miles. I got some roadside assistance, checked the fuel pressure with my launch tool (it was flakey, pressure was okay for a few seconds, then dropped, came back etc. etc.). The guy helping me had worked as a mechanic for a mercedes dealership so I could quickly explain the work I had done, he decided it would be impossible to get this fixed on a Sunday night due to a lack of parts.

However, as the issue was transient, it disappeared after a while and some wiggling. As I was only 40 minutes from home, I took the risk and drove back. Fuel pressure was fine all that time, car is once again parked in my garage. The symptoms seem to match the work I've performed on the car, I guess the terminals and crimping I did on the connector to the HPFP are subpar and are acting up. So I ordered a new plug and terminals for the connector to the HPFP from Mercedes which should arrive tomorrow. Probably 2~3 hours of work due to the impossible location of this connector and the car should be fine again.

 

So, that did not work out as planned. I ordered the needed parts at a local MB dealer. The part number I found for the terminals in the connector was not in their system, but the guy on the found found another suitable terminal. When I showed up to collect the items, I found the terminals where identical to the one I used myself, not identical to the original ones on the car. Helpfully, the guy showed me their collection of all different MB terminals in their warehouse and allowed me to pick several different types to find the best match at home.

Back home, I first started measuring conductivity on the loom to the HPFP. No matter how I wiggled the wires, conductivity was all good. I didn't want to gamble things so decided to replace the terminals anyway. Removed the inlet manifold etc to get to the connector, unpinned and removed the old terminals, crimped on the new terminals, replaced all removed parts and took the car for a test drive. All was fine.

However, today, after driving roughly 100km without issues, the car failed once more. The moment I floored the pedal (to test if all was okay on full load), the car hesitated and the fuel pressure was once again all over the place. I was able to coast into a safe spot at a fuel station. This time, the car refused to get back to normal working conditions. Strangely, I could see that the ECU had great trouble getting to a stable fuel pressure:

This is the car at idle. The downward slope at ~345 is me revving the engine to ~3000rpm for a second. You can see the result of the QV control, ramping up the pressure, but way to slow. It overshoots the target 150 bar, finds out too late, slowly reduces pressure and catches it at 150. This takes about 7 seconds in total. Looking at QV control (actuation angle), there is an accompanying wave visible, all the way from max 120° to full closure 0° to bring pressure back. To keep the pressure at 150, the QV is oscillating between 30° and 60° in a ~4s pattern. Clearly, the effect of controlling the QV is not what the ECU is expecting. Pump efficiency seems to be way too low.

Unfortunately, even after letting the engine cool for a long time, the issue was no longer transient but here to stay. So the car ended up on a flatbed, the road assistance guy was friendly enough to bring the car (and myself) back to my house.

So, somehow it looks a bit like I ended up back at square one. I'll go over the basic checks tomorrow, see if the wiring to the QV is okay, see if it's pulsing correctly etc. etc. I do expect these things to check out healthy, I'll probably have to dig deep to find the root cause of this issue.

 

Here is a moving car fuel pressure data for you to compare next time. Remember, I am on a 4 lobes for fuel pump.
When under high load and higher RPM, ECM will reduce fuel pressure. That is the NORMAL behaviour.
Attached part of my log in excel.

5 data points per second. I use only 82.8 seconds for sample below.




....


70ish percent engine load is when HPFP pressure get reduced from 189BAR to 150 BAR.
Idling is 120 BAR typical.

I believe your engine being non turbo maybe a bit different for fuel pressure profile, but won't be much difference I guess.

Good luck on the troubleshooting.

 

Dont ignore that front end knock. Most likely a ball joint (tie rod end type connection) that can really bite ya if it goes.

 

S-Prihadi, thanks for those graphs and the underlying data. Looking at that data, it's visible that the fuel pressure gets regulated to the right level within 1s, or even within one measurement interval. That's a very direct response to the desired fuel pressure, and your graph shows non of the hysteresis visible in my graphs.

I tend to jump to the conclusion that there is, after all, an issue with the lobes driving the pump, as the response of the pump is clearly not what the ecu is expecting, and the response changes. Those waves in the pictures above were there for a while, but are gone now. However, I have to approach this more carefully without jumping to conclusions, and think of the logical diagnostic steps to rule out (or confirm) groups of issues.

 

Good to hear you are able to get all your issues cleared.

Can you provide a web link to where you got the roadtop unit ? Thank you.

 

Amazon is usually where people get it. However you can also get it direct from RoadTop or AliExpress for less if you want to wait. AND if you have a problem, Amazon is much easier to return an item.

 

Got the unit from Amazon, with a €40 coupon, for the reasons stated above. I see the unit is currently heavily discounted on Aliexpress, it’s only $178 right now.

 

Aan update on this. Today I measured the most important things, camshaft timing, crankshaft sensor, control of the high pressure pump, I put 12V on the high pressure pump to hear the solenoid switch, everything was ok. Then I repeated the steps from last time. Disconnected the injectors and started again, the pressure immediately rose to a healthy level!

Reconnected everything and started again, and the car works again with good fuel pressure!

What could possibly be going on here? After the previous 'repair' I drove 1500km without any problems. Then it stops completely, shows the strange values above, gets even worse, until I start the engine with disconnected injectors.

I would say that if an injector is leaking, the car will not run (properly) and the rail pressure would drop off quickly after turning off the engine. So that would rule out a leaking injector. At the same time, there appears to be relationship between disconnecting and connecting the injectors and pressure. Please note, these are piezo injectors.

I can't see how this would all add up to a logical conclusion, any suggestions?

 

At this point in time I still think this is intermittent electrical issue.
If mechanical , I doubt the engine can run normal for 1,500KM.

I still am baffled at that time when your HPFP Quantity Valve signal can be "over-powered" by your camshaft position sensor signal.
Was that wrong back probing or was it actual occurrence /glitch ?

 

That must have been bad back-probing, putting in the needle at an angle. Rule that out as a cause, that was a big red herring.

I'm also thinking along the lines of an intermittent electrical issue, now thinking of the injector wiring (as that's what I last touched). It'll be very annoying to troubleshoot, I may have to drive another 1.500km for the issue to come back. So that would mean I have to drive around with enough tools and clothes to get to the injectors while on the road.

I'll try wiggling and measuring various things on the wiring to the injectors. But I just cannot imagine a plausible failure mode of those wires to cause this issue. If an injector was stuck open, it would make sense to loose a lot of pressure. But that would not explain the initial pressure rise while cranking.

A mechanical issue with an injector on the other hand, would not simply disappear by disconnecting that injector, cranking the engine, and then reconnecting things.

 

If the issue would return, I'll probably carefully have to disconnect injector by injector, to see when the pressure starts increasing again. The launch tool has the ability to disable cylinders, probably by disabling injector and coil pulsing. Unfortunately, this only works under preset conditions with hot coolant. So it's not a usable test while in the garage.

 

Piezo injector if power is loss during engine running is dangerous, it may stuck open.
Piezo need electrical assist to close and open. There is no spring like oldie electro-solenoid type injector.
The piezo stacks change shape using electricity, hence its so called Open and Close needs electricity in "opposite" way.
https://www.techtips.ie/Autobiz/test...njectors-1.pdf





See how the CURRENT is reversed to close the piezo injector.
To open the injector, the piezo needs current from ECU. To close the injector, the piezo DUMPED or DISCHARGE its power into ECU.
Piezo is both a power consumer and producer, depending on if you want to expand or shrink its shape.



So if an injector connector get loose while engine is running, and so happened it is then OPEN all the time.... it can be a disaster for the engine.
I wonder what is the ECU protection protocol if such bad luck happen ?
Does ECU mess with HPFP fuel pump like in your engine to save the engine from having the effected cylinder flooded with fuel ?
Can you recall which BANK has RICH fuel trim or which spark plugs got fouled when the HPFP low pressure "issue" happen ?

 

So far, fuel trim has been messy with low fuel pressure and I have not yet pulled the plugs. In the low fuel pressure stage, the engine runs very very rough (if at all), so it'll probably be wet plugs all over the place. But, I have not checked so I am not sure about this.

I'm aware of the piezo open / close currents. If it would get disconnected with the engine running, there is a risk of hydrolocking the engine as well as damaging the ECU. I would expect the ECU to throw specific codes, but this will not be a certainty. If the ECU would be actively limiting fuel pressure, a decrease in opening angle of the quantity valve should be visible. However, the ecu is opening this valve to the max, so it's trying to increase pressure.

Tonight I'll be checking all leads to all injectors, see if anything shows up. I'll be measuring the internal resistance of the injectors as well (should be 220kΩ).

 

Yes a decrease on the quantity valve angle would be obvious.

I really wish you have at least a 4 channels scope to see crankshaft signal quality and integrity while seeing also CAMshaft position sensor signal quality and integrity and the VVT magnetic solenoid electrical signal quality and integrity and current clamp at either fuses for injectors ( 2 fuses ) or COPs ( 2 fuses ) and any other suspect electrical components.

If this is an electrical glitch , the only way to see it is using multi channel scope and patience....hoping it will act up while you are doing the scoping.
I seen tough to troubleshoot electrical gremlins from Pine Hollow AD, ATS Bernie Thomson and some others , with real on the road test while scoping is how they got to catch the intermittent Gremlin/s in action.
Some need 20+ minutes drive to show up...damn.

Don't expect too much from DTC. Some super short duration glitch may not be registered by ECU to produce a DTC.

 

Small update. Still not got to the bottom of this. I removed the HPFP, housing etc. Inspected the markings I made earlier on the cam, these check out fine. While in there, I replaced the PCV and oil centrifuge.

Then I pumped 1 liter of fuel through the lines, ran this through a good filter and inspected the filter with a magnifying glass. There was some small dark dust that came with the fuel, as well as a < 0.3mm metal particle in there.




Hmm, not what you want in the lines running to the HPFP. A contaminated valve inside the HPFP can also lead to the fluctuating pressure issues I'm witnessing. So I decided to open the fuel tank, replace the fuel filter and inspect the fuel tank for debris.

Tank was pretty clean, I could scoop up some metal powder with a magnet. Not much luckily. The new filter should keep this in the tank now instead of carrying this to the HPFP.

I flushed the lines again, pumping 2 liters of fuel through them. I use the hard fuel line that gets attached to the HPFP as the flush point, by attaching a small diameter hose to it and fixing it temporarily with a tie-wrap.

So, fuel lines should be clean now. Ideally I would be able to flush the HPFP itself as well, but I cannot find a suitable attachment for that (even tried a local hydraulics shop). As I have 2 new High Pressure pumps (one Bosch, one MB), I'll stick the 'other' pump on there that should be free of contaminants. See what that gives.

 

So hows the progress J ?

 

Thank for asking!

I mounted everything again and flushed fuel through the HPFP, siphoning of the fuel through the rail connector which was not yet attached at that time. Then I mounted the rest, started up the engine and found the fuel pressure to be still way too low. No difference with previous reports.

I think I have the same HPFP installed now which was last on the car when it ran (1 month ago) and failed it's a new MB unit. I have 2 new HP pumps, the Bosch one and the almost identical MB branded one. So it might be that the current pump has a mechanical issue, I can still swap in the Bosch one to see if that makes a difference.

While fuel was running through the HPFP, I controlled the QV manually (connecting it to 12V). I could hear the solenoid click, but this didn't affect the fuel flow. I was kind of expecting the QV to close the fuel flow, however, there can be several reasons why this didn't work in my case. It may be designed in such a way that it only closed if the high pressure side actually has a higher pressure than the low pressure side, or I may have have put to little power through the QV to fully close it. It's a 0.5Ohm solenoid, and I only provided it with 1amp to prevent overheating the coils.

Anyway, no progress as you can see. Meanwhile I made a small extension lead for the QV electrical connection. I'm awaiting delivery of a current probe for my picoscope, I want to see how much current actually runs through the coil during operation. This will give a further indication whether this is an electrical or mechanical issue. Once I validate this, I might swap the HPFP.

 

I fear you might have intermittent issue with the camshaft position sensor at the intake Bank 1, the CAMshaft the HPFP fuel pump is working from.
I wonder what would ECM do if such information is lost for say 1 second ?
I know if a 4 cylinder inline engine and the intake valve cam-pos sensor lost com for a mere 1 second ( signal on purpose shorted to ground ), ECM lost 1st cylinder TDC and engine stall.
I saw a tech seminar video on it, but the engine is not GDI, only PFI and single bank being a 4 cylinder in line.

Intermittent signal loss is hard to detect.

 

So far I've seen a very reliable signal coming from the cam sensors. However, once my current clamp is in (out of stock, grr....) I'll make sure to capture a combination of Y94 current and B6/5 signal!

To make an extension cable for the 2-pin Y94 cable, I did a lot of hunting for the right connectors. While I was crimping the leads, I glanced across my workbench, where a spare Coil on Plug from my Lotus was lying on a pile of whatever. Guess what, same connector! My Lotus has a Rover engine, and at the time this variant of the engine was engineered, Rover was owned by BMW. So BMW used something from there existing Bosch parts bin. Kostal number 50290937.

 

Ohh I did not know it is by Kostal, assuming mine is the same as yours for HPFP connector, does look similar but not the same.
I bought it from MB Indonesia for the plastic body only and the female terminal from Germany.








Do note, the two wires to HPFP is to be twisted like CAN BUS wire, is yours that way ?

 

My M276.8 turbo dumb azz wiring diagram does not state wire is twisted, but they are twisted.
M276.9 NA wiring diagram shows wires are twisted.






.


Grey color on my ECM and Front SAM wiring is very bad, looks like light brown.




.





.

 

There is way too much shielding around the wires to spot any twist. But I wouldn't expect these wires to be twisted, they carry power (~4 Amps), not a sensitive signal like CAN.

I've received my current probe by now and did some further tests. First, the result of the current test combined with the intake camshaft hall sensor:


Blue trace is the current clamp on the quantity valve of the HPFP. Red trace is the signal from the hall sensor. The red signal looks very weird, this is caused by the common ground of my picoscope. You can still see the spikes indicating the normal block wave, it just 'floats' back to zero, making it a bit hard to read.

The blue trace follows the pattern of the PWM signal to the quantity valve. To compare, here is a single pulse of the QV in the PWM signal:


Comparing the current to the PWM, you can see the initial closing of the valve draws ~5Amps. After this, the ECU regulates current to the QV to 3A to keep it closed.

To me, this assures me the electrical connection between the HPFP and ECU is fine, otherwise there would not be such a current draw. Meanwhile, the position signal from the intake camshaft is present. Hmm..

I've replaced the fuel filter in the tank and flushed the system. I also opened the old filter out of curiosity. Once the housing of the filter is pierced, the fuel contained within the filter can be drained out. This is what I got:



Funnily, it looks very dirty compared to the fuel I drained in the front of the engine. So clearly, dirt accumulates in the bottom of the filter housing. I've left this bottle untouched for a day, and the dirt indeed sinks to the bottom. Probing with a magnet, this dirt is actually a kind of iron dust, it sure follows my magnet. Not sure if this is normal for petrol. Further, the filter was in good shape, no holes or other damage.

So, what can be learned from all this?
- The electrical control of my HPFP is okay.
- Timing of the control is still relatively hard to analyse. Though it all seems normal.
- I found some contamination in the fuel lines. This HPFP might be contaminated inside.
- The position of the cam lobes driving the HPFP has not changed over the last 1.500km. Or, better put, the cam lobe was at the same position both times I examined it.
- The injectors are not leaking.
- Low pressure side is all okay, with a new filter, plenty of flow (2 liter in 30s) and pressure.
- High pressure fuel pressure is still not there. Only a short ramp-up during cranking of the engine, which drops back to 6 bar the moment the injectors start their job.

There are only 3 faults I can think of that can be remaining:
- An internal issue within the HPFP.
- A timing issue with the control of the QV.
- A mechanical issue in the drive of the HPFP (cam follower or cam itself).

Easiest for now is probably to replace the HPFP, I have another one still on my workbench, even though it looks slightly dirty on the inside. This pump has ran for only 1 minute of idling on this engine (which at the time did not solve the issue).

 

When you use 2 channel scope even though non floating ground one, if the 1 channel is used using current clamp for QV, the second channel for cam position sensor ( hall sensor ) is therefore not very contaminated
by the 1st channel...because current clamp is kinda isolated.

Only when both channels are used for voltages and one channel get high voltage like COP 2 wire type, direct big power from ECM ( with 20:1 attenuator ) and one channel goes to CKP, there you can get disturbance.

This is the same 2 channel picoscope like yours, my 2205A 25Mhz, Testing my friend's M271.820 EVO. Blue channel for COP amperage at fuse 24 and red channel for CrankPos-Sensor
https://www.picotech.com/oscilloscop...specifications

I also attached the waveform, 1 page of it.




That is why I said last post, I worry your CamPos Sensor could be having intermittent issue.
You got 4 of them, try each one and see how.
Compare between Bank 1 and Bank 2, INTAKE and then EXHAUST ones



====================


Is your fuel filter design as bad as mine ?
I mean the brown body of the fuel filter element is part of the sealing system.
There is no o-ring. Thus when this brown plastic get hardened, it actually leaked a bit of unfiltered fuel to the clean output of the filter assy.




Supposed to be clean side...look at that small contamination allowed to pass thru by the dumb-azz sealing design




After 5,000 liters of fuel. Fuel filter dirty side. Yours is worst than mine. More liters you have consumed or the fuel station underground fuel tank maybe below seawater level line at high tide and is not healthy.
I once got lots of water from such incident in a marina based fuel station. The underground tank is old and at high tide water actually enter thru small corrosion of the tank body.


This stupid azz sealing design for the fuel filter output ( clean side ) can ruin my HPFP .

...

Fuel processing plant >>>trucking>>underground tank at fuel station>>>>fuel dispenser maybe not having a good built-in filter>>>> your fuel tank.
That is the dirt you get at filter element UNFILTERED side.

Its very difficult to explain to car owners about fuel cleanliness because they burn so little fuel per day.
But when you have a 3 x 250HP outboard dive boat (mine) or 5 x 250HP outboards as used by fast ferry operators in my country, these boats burn 1,000 - 1,500 liters a day easy for 5-8 hours daily run,
you see the accumulated dirt from fuel easy.

Your engine has burn easy 5,000+ liters, so no surprise to see dirt in the fuel filter.
I am limiting my E400 fuel filter now to 25,000KM or = 3,500 liters and 5 years.
This way I do not loose more than 0.75BAR at the fuel filter towards the HPFP fuel pump at WOT redline.

 

Look at how different you cam-pos-sensor signal is, disregard the spike, see the pattern is wrong on your capture.
You call it HALL SENSOR, CKP and CamPos-Sensor are also HALL SENSORs, but you meant CamPos-Sensor ...correct ?

.