Well, I had similar symptoms of intermittant fuel at the Schrader valve. Sometime fuel, sometimes no fuel. Car would start, then sometimes not. Traced it back to the fuel relay which appeared to be shorting due to oil or grease that inexplicably found it's way into that part of the fuse tray (mine is in the trunk).

 

No issues with starting.

I'll note that I installed new [factory] Bosch injectors. New fuel filter.

 

If injectors are new (leak the ruled out), it is probably fuel quantity valve. Quite common fault.

More rare culprits are primer pump (inc venturi balancer in saddle tank versions), rail pressure valve, hp pump (bolt behind pump are loosening but its usually noticed by a leak outside via peep hole).

 

I ordered a new (reman) HPFP to have on hand when I need it: could be soon, or not for a while. Depending on how things go I might just swap out the quantity control valve: new pump will have a new valve*. I'll perform the same test with a different/new valve or with a pump swap.

* Valves are about $100. For another $300 I get the reman pump AND a new valve.

 

"Here, watch this!"... Great, so I placed the order to then be told that they don't have any pumps in stock and don't know when they will have them. Note to self: wait to post until you know you're going to get the part! I'll look to update status as changes occur.

 

After about 2 1/2 weeks of no issues I encountered the same condition/event. Pulled over and restarted and was OK for a couple miles (down the highway and then off an on to a lower speed roadway) and then it happened again. I then cycled the key and it cranked but didn't start: this had never happened before. After a couple of minutes and after clearing codes (had my diag tool with me and it showed 2017 and 2047, same as I first encountered) the engine then started and I was able to make it home w/o incident. The first incident from the other day was running at roughly 1,800 rpm under minimal load (highway speed) while the second incident was after slowing down coming up to a stop light. It is this second incident that seems telling to me.

Since I cannot get the HPFP that I was aiming for (cost was low enough that I figured I'd just go ahead and R&R) I'm now thinking I need to rule things in/out. Lots of reading seems to suggest these four possibilities (I'm excluding fuel filter as this is freshly installed as well as injectors as they also are new):

1. Low pressure fuel pump (in tank);
2. Fuel pressure quantity control valve (in HPFP);
3. HPFP;
4. Fuel pressure regulating valve (front of fuel rail).

I am suspecting that it's the quantity control valve after experiencing the two different operating states in which I got dropped into limp mode (actually the engine pretty much dies). The [brief] non-start is the first time I've encountered this: the engine was at operating temps at the time; mostly I'm starting when not at full operating temp. A bad quantity control valve is known to cause no/hard starts. Is there a way to isolate this w/o actually replacing it? It's not a very expensive part and doesn't seem very difficult to replace in which case it might be worth throwing in a new part. NOTE: at the time I had the non-start had I know to do so I would have disconnected the connector at this valve in order to test.

I suppose the low pressure pump could be an/the issue but I'd think that it would tend to occur under loads/higher fuel delivery demands.

The fuel pressure regulating valve, from what I can understand, tends to be responsible more for high pressure codes than low pressure ones but I'm not very confident that low pressure issues with this can be ruled out (just seems like it's less of a potential culprit than the other things).

As mentioned, I was thinking I could eliminate two out of four of these possible sources by replacing the HPFP (as they come with new quantity valves). At a low price point this would have been justifiable.

Figure that I can repeat this failure using my Foxwell. I am wondering if I can run the test on the pump with the quantity control valve disconnected.

 

Disconnected the quantity control valve (Y94) and performed some tests.

Rail pressure variations test did not seem to trigger any issues; when done and at idle I showed pressure control valve at 19.93% and Y94 at 12.56 mm^3/H (static value because it's disconnected?). Not sure what it's supposed to do but it did some counts and then ended w/o any discernible affects on the engine (test only had engine at idle).

Running the HPFP test with the quantity control valve disconnected produced the same results as my previous tests- engine speed ramps up around 1,400 rpm and then the engine stumbles and dies and the low pressure code(s) triggered. Given that the quantity control valve is disconnected would this then point at the HPFP? It would be really nice to know what exactly the Foxwell NT510 is doing in this test: I've posted a question on Amazon and am hoping to hear from someone.

Another thing to add is that I had never coded the new injectors. I am going to do this (after reading this) but am doubtful it's going to have any affect on this low fuel pressure issue: but while I'm focusing on the fuel stuff I ought to take care of this.

Going back to the "spare" fuel rail that came with the parts car, I'm thinking that a similar problem may have been plaguing the previous owner (the guy, as I was told, was a good mechanic [aren't they all? LOL], meaning that the work wasn't likely performed by a shop). Perhaps it's been the HPFP all along?

 

An update (of sorts)...

I posted a thread in the diesel discussion forum and got nothing: disappointed. I am going to just let that thread die and continue with this one.

I have determined that Foxwell's HPFP test is utter crap. Running it on another CDI (which has ZERO issues) produced the same stumble and low fuel pressure codes (killing the engine). Totally unreliable, which tends to pull back on the probability of this being due to a failing HPFP.

I talked to my [extremely] trusted truck mechanic in depth about this issue and he told me it was worth it to replace the QCV (Y94). He said that he just doesn't see CP3 pumps failing (very rare). I obtained a new QCV and am waiting for the next low fuel pressure event before changing it: very slim chance that the coding of the new injectors might have resolved this issue- I figure I'll test this theory by running as-is for now (if no more errors after a couple thousand miles then I might be able to claim the coding as being responsible).

I went back and reviewed correspondence I had with the person that I got my parts car from (the engine that's in my car now) and I note that he'd said the spare fuel rail (and engine wiring harness) was good (these were off the car's original engine, one that a quickie oil-change place failed to refill with oil).

I'd like to know of a reasonably priced tool that can actually test an HPFP on-car. I reached out to ThinkMyCar(?) about their ThinkDiag-2's HPFP testing and got horrible customer service in which case I WILL NOT purchase anything from them: I'll happily pay more for actual customer support.

 

Took a couple of weeks but the low fuel pressure codes (2017 & 2047) popped back up which means injector coding had no affect (on stomping out the errors).

I replaced the QCV today and ran the HPFP test via my Foxwell. NO DIFFERENCE! As I have no idea what the Foxwell is doing I have ZERO confidence in the test results. Yes, the test doesn't generate codes on another om648 I have access to but without knowing what the test is actually doing it's inconclusive. Anyone out there that has a Foxwell NT510 (or other Foxwell that has an HPFP test) could you test it to see what you get?

At this point I'm looking at an HPFP or the rail pressure regulator. Pretty much assured that if I replace just one of these that I'll still end up with these codes!

I could only dream of having a code that ONLY goes into limp mode. Losing complete power is BAD! (no way my wife is going to drive this car until this is fixed)

 

Not going to let this topic/thread fall of the cliff...

I'm kind of in limbo right now as the weather has changed and I'm not driving the car right now (also some other reasons why I'm not driving it). Spending some time better understanding how the system works and how to diagnose issues. I had been thinking of picking up an oscilloscope and while doing research I stumbled across this potential use of an oscilloscope to isolate fueling issues:

In my case I think I can use an oscilloscope to see what is happening when I run the HPFP test on my Foxwell NT510 (as it trips a low fuel pressure error every time the test is run). The above video seems to answer my question as to what the Foxwell test is doing that causes the engine to shut off; what's happening is that the rail fuel pressure is jacked up to the point that the ECU shuts off the injectors- the test holds the fuel pressure valve (Y74) closed and the ECU sees the pressure being high and shuts down the injectors. What is interesting in my case is that I'm getting LOW pressure errors.

 

FINALLY got around to testing the Y74 (fuel rail pressure regulator/valve) via leak-off. Here's what I got after about 10 seconds of cranking (engine cold):

That's 2 mL(?) Clearly this is more than ZERO, in which case it would appear to indicate that the Y74 is bad. I have to wonder how I could have only had very intermittent/infrequent problems with it being this bad. ZERO starting issues.

I'll note that I clamped off the injector return lines and when I first ran this test and saw the fuel coming out of the tee off the Y74 I doubted that my injector return line was clamping off sufficiently. I cut the injector return line (couldn't get it off the Y74 tee), routed it into a catch and pinched off the stub that I'd left on the Y74 tee. Had my wife monitor as I cranked and she saw no fuel coming out of that pinched stub in which case the fuel I was collecting off the tee was ONLY coming from the Y74 itself: she did say that when I stopped cranking there was a small belch (perhaps just a bubble?).

NOTE: I've read where it's claimed that it's sufficient to just disconnect the cam sensor to keep the engine from firing. NOT TRUE! I ended up having to disconnect the wiring from each injector.

Next step is to swap my spare rail [of unknown condition]. QUESTION: what's the best way to ensure that a fuel rail is clean enough to install? The spare rail wasn't capped.

 

Easier to just spray brake cleaner in the orifice and let it flow out. You can't get all of it out, just get the major ones out.

Fuel injectors all have a screen built in precisely for that purpose.

Did you measure the lpfp output? Either with a gauge or a sensor (if installed)? I dont have any experience with 648, but it's similar enough to my 335d.

 

Did not measure LPFP output. My reasoning is is that I have ZERO fuel delivery issues under hard loads. The car exhibited no issues with this pump (unlike everything else, the LPFP was not swapped out). NOTE: based on the cranking fuel pressure in the rail being about 400 bar I'm assuming that that's an indication that there's no issues with the LPFP. COULD the LPFP be flaky? Maybe, but as I said, there were no fuel pressure issues from this car before (with the other, original engine).

Going to add some additional observations on my testing...

If I disconnect the Y74 harness fuel rail pressure shows about 35 bar and valve at 0% open; zero fuel leaking/dumping past. With Y74 harness connected pressured registered about 400 bar and the valve 28% open; fuel leaking/dumping past. 28% open would indicate that fuel SHOULD be dumping. Why is it open at cold crank, shouldn't it be closed fully?

 

It should NEVER start with cam sensor unplugged, because cam/crank sync has not been done. If it fires, you have something in cylinder which can burn. Oil or fuel.

Fuel pressure is controlled by Y74 at first seconds, or cold engine, or idle.. On other conditions, Y94 controls pressure, Y74 is held just slightly open then.

Your persentage is PWM percentage, so depending on definition, 0% may actually mean full command.

 

It should NEVER start with cam sensor unplugged, because cam/crank sync has not been done.

This is a diesel. Diesels do not need a cam/crank sync because everything is straight. No variable valve timing. No camshaft adjuster. No advancing or retarding. Straight time.

Yes, diesel still use a cam sensor, but it is used as an additional point to verify the crank. Cam/crank sync is used to make sure everything is in time, otherwise limp mode. Depending on the ECU and the level of failsafe the engineers built in the ECU, diesels can start and run using only the crank sensor or cam sensor as a rpm source.

To answer the other one, my understanding of Y74 is that its reversed PWM. You're correct that 0% is actually fully open, 100% is closed.

 

Deplore, thanks for the clarification on the cam sensor issue and Y74.

Here's a video I was going by for my testing:

What is confusing is is this guy didn't disconnect the injector wiring, only the cam sensor. I think I'd read somewhere of disconnecting the wiring to the injectors but thought that if just disconnecting the cam sensor did the trick then that's a lot less work (not that it's a big deal). I was somewhat caught by surprise when cranking the starter and the engine started to pop- it was an uneven firing. Anyway, unless any of this interferes with the object of testing the Y74 I think I can move on from it...

On the Y74, this is the confusion, whether the default state is OPEN or CLOSED. When running an HPFP test on my Foxwell I can see the percentage INCREASE as the test increases load:

What is going on here? I wish I could see what the Y94 is doing. I'll have my hands on a Star setup soon in which case I hope to have a lot better look.

All said, however, can anyone state with certainty (provide documented reference) to whether that Y74 leak-off test is the correct way to verify operation of the Y74? On face it would seem that my Y74 is bad as it is dumping a fair amount of fuel; but, it also seems that with this amount of leakage that the engine wouldn't run as well as it does, that the frequency of problems (tripping codes) would be far greater. I'll note that the Foxwell HPFP test triggers this error every time: also note that it is supposed to kill the engine at the end of the test.

I could just swap out the rail and retest but I want to refrain from parts-swapping w/o fully understanding how things are supposed to work. For all I know the current rail and Y74 are fine and installing the spare will show it as being bad or performing the same way. I'll note that the spare rail came with my parts car, it was in the trunk; the PO's comment on it wasn't all that clear, saying that I could toss if it I wanted to. Have zero confidence in the existing rail or this spare. And to make things even more convoluted, there's also the rail from this car's original engine that I could try (it's really trashed looking- I don't want that look on this engine). There could be an issue with the engine wiring harness? Parts car also came with a spare wiring harness. Hm... BUT, if there was a wiring issue I'd think that I'd readily get errors triggered.

 

Yes, cam sensor thing keeps me confusing too... because it VERY COMMON culprit for non starter (also in diesels), that cam sensor is faulty. Yes I digged WIS documents but there was quite overall description about limp and substitute sensor usage if cam data is inplausible. Cam sensor is not giving any angle data, just 1 or 0 (I have seen that in live data)and cylinder injection (compression stroke) is triggered instead of scavenging stroke. I still suspect your cylinder had some fuel and it autoignited... perhaps some injector is not closing properly?

I remember not seeing actual Y74 leak test procedure in SDS

 

Yeah, no idea what was going on with it firing. The engine had not been running for a while and was stone cold. Based on what I'd read/seen others do (video of 0m647 above is an example) the ECU would withhold signals to the injectors if the cam sensor wasn't plugged in. This is now clearly debatable! The sure-fire way, pardon the pun, is to disconnect the injector wiring.

This entire concept of testing the Y74 is puzzling. I'd have to go out and read all the stories again to see if dealers were performing this leak-off test. Seems that in so many instances dealers would just swap out rails (applies to om642s as well).

But back to the understanding of how the Y74 works, how it's commanded... Is it not PWM controlled? That's the only way to get a percentage, right? If so then is Y74 positive or negative switched? (does that even matter?) I can understand that at 50% it would be in more or less allowing fuel to be bled and fuel to be contained HALF the time. There appears to only be TWO wires to this (same as the Y94) in which case there's no CAN signaling going on in which to verify (only real verification of the operation of the Y74 is via the rail pressure sensor?). STILL, this doesn't seem to correlate with what the supposed leak-off testing suggests is supposed to happen.

 

All these proportional valves are PWM controlled. Remember it is pressure control valve, not flow control. So its actually pump which defines how much flow is delivered to rail. Then Y74 setting controls which pressure it needs to open. Flow over Y74 is the defined by pressure difference over control edge (area) by orifice equation. So its interaction and between balance of pump delivery / rail pressure / Y74 flow.

 

Excuse the lack of updates here. I am currently not using/needing to use this car in which case working on it hasn't been a top priority. I instead spent a bunch of time cleaning up in my shop: I could barely walk in there- all the parts from this restoration project (had two entire cars sans one body in there at one point!). Now that I've got a more comfortable environment in which to work AND that I now have MB Star I am hoping to get back at this issue.

Can anyone tell me whether there's any specific adaptation requirements for replacement of Y94, Y74 or the fuel rail pressure sensor?

This engine was from another car and I'm thinking that perhaps the ECU might not be in sync with this engine. I DID perform some sort of fuel-related adaptation with my Foxwell but I don't recall whether it would affect any of the fuel rail stuff: at the time I was concerned about adapation for the injectors (though I did go through recoding).

With my ALH VW engines I know that the Injected Quantity [IQ] if set low (which is actually heavier fueling) can result in the engine stumbling when dropping off the accelerator. The results here are somewhat similar to what I'm having with the CDI, though the CDI's behavior is very random

 

So, spent some time poking at this yesterday. I've found that the leak-off test [as seems typically specified as] cannot be relied upon as the ECU is commanding the Y74 to be partially open during cranking.

I've found that Oregon Fuel Injection's documentation (for Sprinters, om647s in which case all is applicable) is one of the best sources out there (outside of MB Star, which I soon will be finding out more about in depth). Here's the relevant [to my issue] part (under Fuel Pressure Control Solenoid, In Fuel Rail):

2. In a DE-energized state, the fuel pressure solenoid is held closed by spring force, like a relief valve. During starting the fuel pressure solenoid is held closed by
magnetic force to allow enough pressure for starting. When checking the fuel pressure solenoid there should be no fuel coming out the return during cranking,
and it should hold 870-1450 psi when unplugged during cranking. If it doesn't, and the injectors are OK, then the fuel pressure solenoid is bad.

From how I am reading this there are TWO separate qualifiers:

1) there should be no fuel coming out the return during cranking;
2) [the rail] should hold 870-1450 psi when [regulator valve is] unplugged during cranking.

My tests completely fail under the first condition but as noted my ECU is commanding the valve to be open in which case fuel can only have to be coming out. It's the second condition that I found to show up a problem and Oregon Fuel Injection is the only folks that I've encountered that present this test scenario.

For comparison (cranking- ample RPMs):

Y74 Wire Connection State​	Y74 %Open​	Rail Pressure (Bar)​	Y74 Voltage​
Connected	30​	500​	2.3 to 2.7​
Disconnected	0​	48​	8.2​

I briefly saw maybe 52 bar when disconnected. This is roughly 754 psi which is significantly less than the lower threshold of 870 psi that Oregon Fuel Injection specifies: I suppose one should keep in mind that this is w/o adjusting for atmospheric level in which case for my location I'd reduce psi by about 15 which would adjust to, at best, about 740 psi.

I had a helper with me and he is pretty certain that he saw some fuel, very minor, coming out of the Y74 when cranking with the Y74 harness disconnected. In a case such as this it would appear that the Y74 IS bad. BUT...

I will be reconnecting up all the fuel lines and going through Star's testing to see what shows up (rather excited to do this as there seems to be a lot of different tests [makes me wonder how MB techs can struggle with figuring out these problems; I shouldn't speak too soon though]). As my Foxwell is able to trigger the error I'm thinking that Start should be able to do so also and that, perhaps, there's an ability to capture data that'll clearly show where and what things are failing.

Because this engine has over 150k miles (actual mileage unknown but estimated to be somewhere between 150k and 175k miles, perhaps a bit more) it is far closer to the end of life than beginning of life for the Y74. It is only a matter of time before the Y74 fails. Parts costs and availability are only likely to get more tenuous over time, so... With this in mind I have ordered a new rail. If the Star tests point elsewhere then I'll have a new rail standing by for when the inevitable occurs.

Stay tuned...

 

I ran some DAS diagnostics. The HPFP was very steady at 1,400 bar: it over-shot 1,400 for a brief moment right at the start of being ramped up but then flattened out. I would surmise that the HFPF is fine. Tests of the Y94 showed it varying by about 20 bar: I think the max deviation allowed is up to 50 bar. Recall that this is a new Y94. I am unable to readily perform the specified Y74 tests because special adapters are required: I tried back-probing the connector but couldn't make a proper connection (same with pressure sensor).

What I was "encouraged" to find out was the information that DAS reported about a stored 2047 (Y94) fault code. This might have been from some time back, I'm not certain (the km reading doesn't make sense- 20292! in which case I cannot say if this occurred prior to being in my shop and messing around with all this). Here's a pic of that logged fault code:

Obvious low pressure here. Why is the Y74 at 29% when, clearly, there's a need to increase rail pressure? Perhaps this is a normal requested percentage and the Y74 was sticking and therefore continuing to drop pressure when an increase was desired? I'll note that during cranking and with the Y74 connected I was seeing roughly this same percentage -29%- being requested. The amount of fuel available to the rail seems plenty sufficient as "Current injected quantity" seems to indicate.

I cleared this code and will be running the HPFP test again to see if it triggers a code. If it doesn't trigger then I'll use my Foxwell and run the HPFP test (which has been dead reliable in its consistency of triggering a low pressure code) and then dig into the frame data using DAS.

BTW - Where the heck is that "Kilometer reading" value coming from? Seriously, 12k miles? Is this from some last service reset or something?

 

From EIS.

 

Any idea on how it could come up with such a low reading? EIS has been replaced?

 

most of the code mileage readings are only last 5 digits.

 

OK, the "last 5 digits" maps out. Doing the conversion I'm within about 3 miles of what the odometer reads: I have to drive another 3 miles before the error occurs! Before I spend time poking around, can anyone tell me if I can change DAS's reporting to be in miles? (trying an internet search gets nothing but a flood of how to change the odometer's reporting units!)