Mercedes Fuel System Diagnosis
History: I have a 2008 ML63 giving me some trouble. It originally came in a year ago for the same code. It needed other things and it apparently wasn’t a priority to them at that time. From what I understand it has been on and off since. Suffice to say it has absolutely no effect on the driveability of the car.
Now the truck is back to specifically have this issue fixed. (I assume due to our bi-annual emissions testing in Maryland). Looking at the freeze frame the fuel rail pressure was 52.6 psi, Open Loop, 4 seconds run time. FP definitely doesn’t seem high? Due to the code description I immediately went after the fuel rail pressure sensor to see if anything was obvious there. The Verus Edge with 17.4 shows very poor factory level data. Global honestly has a vastly better data set. Taking the car for a drive I never saw more than 62.2 psi and less than 49.1 psi. Numbers that definitely do not draw any suspicion. I then verified clean reference and ground at the sensor and scoped the output voltage (screenshot included during code set). I found that if I clear the code and start the car cold, or partially warm, I can get the code to reset in 4-5 seconds after 2 to 10 starts. Always in park, no throttle necessary. The pressure has never been more or less than 50.2 psi or 52.6 psi in those conditions. Hardly enough to elicit a Upper Limit Reached code being set.
ProDemand doesn’t have much information on this truck other than a fuel spec of approximately 3.8bar (55.11psi). Absolutely no help on the DTC index. I’m almost ashamed to say it but I have never purchased OEM information before, but this truck seemed like a great candidate so I took the leap. We spent $60 on a Mercedes one day subscription and after a lot of digging and familiarizing myself the best I could, I found their DTC index. It was as helpful as ProDemand with no information. Basically the trouble tree is to see if the data is plausible and that was it. No mention of tests or specs whatsoever. I played around in their WIS program looking for fuel information and the like. Maybe it was my frustration of the situation, or that I know time is limited, as well as other cars that needed to be fixed. Either way, I am here for any and all insight that my colleagues may have. The only things I could find on Google and AMG forums were saying the dealers were recommending pumps and those that went that route had success. I know dealers have the techline protection and after time spent on the phone running specs are given the grace of installing parts without the worry of financial burden. This gentleman, however, has a extended warranty and I need to be a lot more sure before we go down a potentially $2000 dollar path for parts/labor/&diag.
If the engine runs normally, Based on those codes alone i would Replace fuel pump assembly, found some information related to the code 3699 but cannot post pdf here, is not an accepted file. I would also flash and SCN the engine module.
I would feel a lot better about the situation if I could find some set criteria. What could the computer be looking at considering the pressures look good but it sets the code like clockwork? I don't have a PID for pump duty cycle but that doesn't mean it doesn't exist in the computers logic. Also you would think if the duty cycle was higher to produce appropriate pressure then it would hesitate under heavy load, which it doesn't.
Having the factory scanner we rarely have to get involved with such drawn out technical diagnostics. based on information i have read the normal fuel pump Idle voltage should be 5-6 Volts and current draw is 7 Amps, but this is actual values from the Mercedes scanner, not sure if yours will show this value.
Does your fuel pressure bleed off or go up when static?
I do not have the voltage and amperage PIDs in the Verus OBD2 but I could check another tool that may.
I just checked the pressure like you asked and it rises.
rises when off?
found this in my notes from training
Check engine light concern, no drivability concern. If you encounter a ME 3698 checkThe correction factor: PWM signal fuel pump via the path shown below.A 3698 will show this factor at a -15% or more--check for a restriction in the return,check return flow to tank from fuel filter pressure regulator. If there is no returnflow all available fuel/ pressure goes to the fuel rail, causing the ME to correct the PWMto slow the pump to maintain system pressure, hence the negative correction factor.A 3699 will show this factor at a +15% or more-- check the fuel pump for mechanicalwear. If the pump has mechanical wear and has trouble providing pressure, the ME willTry to speed up the fuel pump to maintain system pressure, hence the positive correction. On known good vehicles this value looks to be anywhere between -0.3% to -5.0%
Here's what you need to do:
Start and idle the engine, go to Engine > Actual Values > Further Actual Values > Fuel System
Here you can see actual and target fuel pressure, as well as "Correction Factor". 3699 will set if the correction factor exceeds 15%, a normal value is 0% to -5.0%. This problem can be caused by a bad fuel pump but it is much more commonly a vacuum leak. You also want to look at your mixture adaptations to see if the engine is running lean.
You need to BOTH smoke this thing and use your favorite propane or carb cleaner and search for vacuum leaks all over the engine. There are a million places it could leak. All of the sensors and connections on the front of the intake, the intake gaskets, the rubber boot from the air cleaner, the vacuum hoses/fittings going around the intake, the PCV hose going from the oil seperator on front of engine, underneath intake, to PCV valve (check this too), and the PCV hose from the valve to the intake. The brake booster hose and fitting, the purge connection to the intake, the secondary air injection servos and vacuum supply line. With the air cleaner taken off the intake at the back, make sure the purge line is plugged into the port in the intake so it goes straight to the line on the outside. You can see it looking straight down. I forget I think an intake air temp sensor is right there too.
Here's what you probably want to know:
This code is setting because a fuel system Monitor is failing. On a PWM system when the target fuel pressure is not achieved the module will increase the duty cycle to the pump to make up for it. Alternatively if the fuel pressure is exceeded the duty cycle will be corrected negative (pinched fuel line?).
With PWM fuel control there's a balancing act between air, injectors, and the fuel pump. As the intake air g/s increases, the injection increases, and the rail pressure drops. When the rail pressure drops the fuel pump command is increased to maintain the setpoint. The module stores a performance map of typical rail pressure setpoint, injection quantity, and fuel pump duty. It's all measured in advance when the system is engineered. Then the pump duty is adapted to make up for discrepancies between expected and actual pressure. 3699 and 3698 are set by a monitor specifically for this engine based on the idea that emissions will degrade if the pump can't maintain the correct pressure.
So how would a vacuum leak affect this system? Fuel injection is calculated on the basis of measured airflow, so unmetered air does not count toward the fuel system rationality. Unmetered air is recognized as a lean condition and injection is increased which lowers rail pressure which increases pump correction factor. A bad enough vacuum leak will cause the problem just idling the engine.
How would a bad fuel pump affect this system? If it fails with reduced output the rail pressure will fall below setpoint (remember performance map is based on a healthy system). Correction factor will increase. Not nearly as common for it to be a fuel pump issue as this would affect loaded driving mixture adaptation much worse than idle fuel supply. (Watch correction factor and lambda, target/setpoint pressures while driving).
What about a bad fuel pressure sensor? Possible. Never heard of it on this application. Check sensor against gauge, may have to compensate for atmosphere (relative or absolute type sensor).
There are two fuel pumps I think you can scope the pair and compare the total amperage and the commutator quality.
Here are example quicktests:
Thank you very much for the excellent response. I am going to have to see if my other tool will give me those correction factor PIDs.
I did smoke test the intake system the end of last week. I made discs to fit into the inlet ovals so I could get a good positive pressure seal. The only thing I found was the seal at the base of the two to one inlet. I would like to see if I can see that correction factor PID before putting it on so I can see what kind of change it makes if any. It was very minor and I didn't think it would play much of a part as the fuel trims are in order all the time, even when the code sets.
Generally what I do is put a block off on the throttle body and go into a vacuum hose. I do love your custom block off plates though.
Mercedes has a bulletin for this where they tell you to like take a piece of sheet metal and trace the intake plumbing onto it, then run it through a band saw and glue it together!
Andy make sure you check for leaks with propane/carb cleaner, not all leaks show up under pressure. Did you check the vacuum lines for the secondary air valves and the valves themselves? If they are leaking they may only cause a vacuum leak when secondary air is active so make appropriate tests. Check the mixture adaptations you should have "idle speed range, left cylinder bank" and "idle speed range, right cylinder bank".
If you want to know if you fix a vacuum leak you can take a look at your g/s at idle. It may change noticeably if the leak is removed.
Thanks, Andrew. Unfortunately I am feeling very electronically under tooled for this. As I said the Snap-on data for this car as of 17.4 is literally worthless. I tried my Auto Enginuity to no avail as well. That had much more data, but oddly enough didn't even have fuel rail pressure, much less the correction factor that I would really like to be able to see. The only other thing we have is an old Autoland Scientech iScan2 which I think is updated to 2011. I forgot to even try it because I haven't used it in a long time, but it has surprised me on more than a few occasions.
I was working on something else all day but was popping out here and there to try a thing or two. I will check for leaks again tomorrow using propane or carb cleaner after I install the new TB inlet seal.
My coworker and I were thinking today. I know for a fact that the correction factor is higher than 15% at idle because I can consistently get the CEL light to come on. If I make absolutely sure there are no intake leaks, it seems like the most likely thing for it to be is pumps. I could go as far as checking the commutator pattern to identify any glaring issues and idle speed. I can't tell you how much I hate not having the right tools for the job.
Sorry I actually wrote this last night but didn't post by accident.
So today I installed the Y outlet/throttle inlet boot and re-smoked the intake manifold. Even wiggling the Y around it did not leak any smoke whatsoever (Although I found the MAF g/s to be extremely close to before). I then taped over the small holes under the filter housings, and around the filter inlet necks. After that I felt like the tract was as sealed as it is going to get. I then removed the foam on top of the fuel rail for better access to the intake and applied propane while looking at the trims. I found absolutely no change in the trims due to propane ingestion.
I checked the hoses at the secondary air valves before that test and after. I found that the left side held well colder, while the right side wouldn't hold vacuum under 10 psi, higher than that it would seal up. When checking them both hot, the left would not hold vacuum at all now, and the right would pause at 10 psi for a second, then drop off. I am definitely not happy with the way they are operating. I pressed on even with those in the condition they are in, and started the car with a coolant temp of about 150 - 170 F about 15 times and did not receive a CEL. So she is now on her best behavior. Although the secondary air never came on while I was trying to start it these times, further solidify to me their role in the CEL.
After that I scoped both the fuel pumps at the fuel pump control units. I found the left to have a very good looking pattern, while the right I am less than satisfied with. Overall I think there are multiple things contributing to this code. As I said before I am very unhappy about not being able to see the target fuel pressures as well as the correction factor PID.
In your opinion is the right side pump waveform poor enough to condemn the pumps and filter? As implied before I am a huge fan of identifying set criteria, observing the fail condition and verifying the repair. I have gone rogue plenty of times (with varying degrees of comfort). I really don't want to see this one again for the same code after all the work I have put into trying to positively identify the problem.
Ok, so let's go with what you have and what you know. I'm not seeing anything in the fuel pump wave form to warrant replacement just yet. Did you get an rpm on those pumps?
You may not have a "factory" scan tool, but you do have a damn good global OBD2 scan tool. We can fix this using obd. If you aren't intimately familiar with all the different make's correction rates and fuel strategies, then global OBD2 will be your friend.
What is your long term/short term at idle? What is it at 2500 rpm?
I'm with Andrew. I think you have a vacuum leak. This test will prove/disprove it. If anything, it will give you a direction to go in.
Put that Snap On scan tool to global mode and read some data. Make nice short custom data lists.
I'm not sure how many commutators the pumps have. If it's 8 then I'm seeing about 12.5 seconds of the left pump waveform = 4800 rpm. Each segment in the screenshots is 2 ms.
I have been using the global OBD2 of the Verus through the whole diagnostic. The Mercedes data on this software revision/vehicle is far too poor to do anything with.
I have provided a cold start data stream, idling then held around … rpm the best I could for a throttle by wire, then some revs at the end. The check engine light comes on around the 490 timestamp. The trims dip very negative at the beginning showing over fueling, the opposite of a vacuum leak. I didn't see it do this in the past. I have always seen them around the 0 mark.
I have cleared the codes but not any adapted values since I have been looking at the vehicle.
Another thing I have learned this during this quick go around is that while the CEL was always coming on very early into the start in the past, this time it came on at about the 5 minute mark. The freeze frame is again showing the fault at the 5 second since engine start time, although I watched the light come on around 5 minutes in.
I'm calling this one fixed.
Well after a lot of testing I decided it was time to pull the trigger on the pumps. I replaced the fuel filter unit on the left side of the tank, and both pumps on the right side of the tank (also if you do this job remember to replaced the sealing ring kits. They come with tank threads, lock rings and sealing rings).
Also the pumps do indeed have 8 commutators.
Well, hopefully that did the job.
Mercedes is great about most things and I love them, but they are not great about code criteria. The actual monitor criteria are a secret that only the engineering teams have documentation on. In general, guided paths in the test tool and experience are what directs atypical engine performance diagnostics. If you have a subscription to Startekinfo even for a day the best place to look is "Launch OBD/DTC". Over there you can pick a vehicle model and find what drivetrain controllers it uses and a full list of their fault codes. There's also usually a link to a WIS document written for that engine that does give you some documentation on DTC set/clear/erase criteria. Not monitors, DTCs. I checked for your vehicle but the 3699 and 3698 codes were omitted. You can search by a DTC globally for all modules as well.
There it will give you some of the guided testing information for the code that would be in the scantool, but a lot of those tests are designed to retrieve information from the vehicle at the time, and a lot of the text is still in German. They're a helpful clue though.