S-Prihadi

J,

When you installed the new HPFP, you did spin the engine to get the fuel lobe at its most flat zone right ?
And the tightening sequence super careful 1/2 spin per screw alternate-ly till pump lowers itself tight super slow and super straight.

This Bosch HPFP is fragile when not installed as per the WIS or how I described it.
This guy honestly admitted failure to spin the engine to get the flat side of the fuel lobe ( very probably he does not have BMW WIS ) and he broke the mini sliding piston rod

Jeedie

Thanks for that info. I set the engine to 40° as per the WIS instruction. I guess that put the lobes on a flat piece, as there really wasn't that much tension on the spring. After releasing it for maybe 6 or 7 mm, the tension is all gone. I did follow the procedure to loosen (and fasten) each bolt alternately by 180° per step. I'll check that video, however, as I suddenly and briefly got at least a tenfold amount of pressure for a short while, I think the pump itself is probably working just fine.

At least the electrical connection at the pump itself is now okay, looking at the old pump, and the fact that the white plastic securing pin on the connector was missing, someone has been there before. Looking at the date on the old pump, it must have been the original. So some repair job has been done in that area, and they bodged it.

Could someone measure the resistance over the quantity valve? Measuring on pin 5 & 53 on the ECU side of the cable to Y94 now reads 0.00 Ohm, so there might also be a short in the wiring.. I would expect the solenoid to read a bit more than this..

S-Prihadi

I am in Bali my 2nd home at the moment, since 22nd Sept. I will be back to my man cave Jakarta home probably in a week.
I have there a spare HPFP I can measure its Y94 resistance for you.
Agree, zero ohms sounds like a wire shorted, but it is induction coil so best I use my spare one as comparison to make sure.

I forgot you already installed your new HPFP and it will take a lot of work to disconnect its connector to do the load test on the 2 wires to it.

Jeedie

I'll see if I can measure the resistance on the old unit, at least that would give some kind of reference.

Meanwhile I was able to capture the below graph of the rail pressure, just after a startup. It started to spike up immediately after the engine started running, but as quickly as it spiked up, it sloped down again, resulting of course in a rough idle etc. etc.

Jeedie

I was able to find the copper contacts on the old HPFP valve, I'm measuring ~0.9 Ohm on the solenoid. I checked once more on my car, it matches that resistance. There could still be an intermittent wiring issue of course, but at least there is some conductivity.

CaliBenzDriver

You are reading less than 1.Ohm on a solenoid winding: something is bad here! Either the coil is fried or your measurement - Let's see...

Once you account for the two leads own 0.5R resistance, you understand something's really odd with an actual 0.4R. measurement.

>> To definitely measure HPFP solenoid, be sure to disconnect it out of the ECU circuit else you're measuring more components that the solenoid alone.

Anything greater than 10.Ohms can be considered acceptable - That's far from a 0.5R.
Compare values with a known good reference from @S-Prihadi as he return from Bali.


My 2¢ on this topic:
-- I don't expect the ECU to drive any abuse in the HPFP control solenoid ....

-- A nearly perfect 0.5R short-circuit value may damage ECU control or fused protection.

-- Double check your circuit probing is done disconnected. Forget about measuring Ohms in a live circuit.

🤞

Jeedie

I just checked the leads, they show 0.1Ohm. The lowest reading I get on the current solenoid, measured from the disconnected ECU plug is 0.6Ohm, so 0.5 max resistance in the solenoid. With of course 1 or 2 meters of loom cabling in between, adding to that resistance. So there is still the option of some form of short at the pump side..

This is what I'm doing, I measuring from a disconnected ECU connector.

That makes sense. I cannot find a reference online, there are examples that are all over the place. From an Audi I see something in the 500~1000Ohm range, from some other brands I found some lower values. I don't think Mercedes specifies a value here for testing in WIS. So hopefully S-Prihadi can find a good reference.


This cannot be ruled out, I agree on that. Though this is probably not testable without a scope.
I Agree, this is what I did previously. I'll try rummaging around with the loom while measuring the resistance, maybe something will show up.

Thanks for all these suggestions! I think I'm close to solving this issue, especially as, since replacing the pump, I've seen the rail pressure spike twice to the needed pressure for a short while, and the car drove fine until it suddenly decided to die.

S-Prihadi

MED177 ECM we use is quite smart and safe.
On many sensors/devices the ECM is sending power to, the ECM can detect overload or dead short and shut power to the defective device.
On 5V sensors such safety measure from overload or dead short is available from/at the ECM.

The Y94 valve since it is PWM signal using some sort of driver circuit, I am sure it is a smart one.
So when the ECM detected overloading from bad induction coil or mild short circuit on the wires, it will shut itself off.
Other examples are most incan bulbs on oldie headlight or rear tail light controlled by Front or Rear SAM, they are smart power circuits.

As long as your HPFP pump is genuine MB, we can rule out defective new HPFP with high degree of certainty.

There was a good case study from Pine Hollow AD, on a customer of his using MB GL class SUV.
The customer bought an aftermarket HVAC compressor, and the command to power up the compressor refrigerant flow solenoid keep failing from the
HVAC computer. So Ivan the techy used a bulb instead and if using a bulb, the HVAC computer will power the bulb.
Using a scope Ivan can see initial command from HVAC computer to power up the aftermaket compressor solenoid exist and within seconds,
the HVAC computer then cut power.

The final and actual cause was this :
That GL class SUV HVAC compressor, if genuine MB, is supposed to have the solenoid with built in flyback diode to absorb the reverse voltage spike an induction coil
will produce. The aftermarket HVAC compressor installed soleoid does not have flyback diode at its solenoid. So the reverse voltage spike caused by the coil winding of the
aftermarket solenoid is seen by HVAC computer as a short circuit, hence HVAC computer shut down power to the solenoid.
An incan bulb is pure resistive load and does not produce reverse voltage spike, that was why the HVAC computer was allowing itself to power the test bulb.

Ivan then installed a diode to act as spike absorber and all is good.
I saw a seller who sells aftermarket refrigerant solenoid for the Denso type compressor used on that MB GL SUV stated : For Mercedes, has built in diode.
So , some car brand will put flyback diode inside HVAC computer, and some expect the solenoid to have the diode.

Therefore when we install something using induction coil, such as solenoid or a relay, we need to make sure the specification matches MB spec.

One example where flyback diode is an ADD ON for our W212 and probably this is for most MB cars, is that green special Starter Circuit 50 relay.
That is the only relay on my non-airmatic W212 where there is a flyback/clamping diode at its output pin to absorb the starter motor solenoid reverse voltage spike.
The coil of this relay has the regular resistor to absorb the relay coil 85/86 induction reverse voltage.



So, if HPFP is genuine and not a suspect, there could be mild short circuit at wires #53 and #5 of F connnector of ECM.
If the short is a true DEAD short, I doubt there is enough power to power up HPFP to 45 BAR.

Jeedie

This is a real head scratcher. Today I unmounted enough to get to the solenoid connector to test the wiring etc. All seems good. The HPFP I installed is a genuine Bosch item, it's the same manufacturer as the original HPFP. All the prints etc are the same, except for the MB logo of course.

I tested the resistance on the wiring between the solenoid and the ECU. That's 0.2 Ohm in total (I made a loop on the Y94 connector and measured on the ECU connector). So that would give only 0.5 Ohm on the solenoid itself, very very low.

When nothing is connected to the Y94 connector, no matter how I wiggle the wiring, the resistance is infinite. That would rule out a shortage in the wiring.

Then I tried my old solenoid, made sure I could properly connect to the (broken) wires, and this one really measures 0.5 Ohm. So in that respect, the old and the new measure similar resistance and I think, when reading 0.7 Ohm on the ECU side with the solenoid connected, I'm actually reading the internal resistance of the solenoid at 0.5 Ohm.

Hopefully someone can provide a measurement of another HPFP to compare, it would be a pretty expensive bet to get another HPFP..

Jeedie

A further note on some info I found regarding the HPFP used on a completely different engine (Chevrolet Cobalt SS 2010 2.0 Turbo), where it states:1. Ignition OFF, disconnect the harness connector at the high pressure fuel pump.Note: The DMM and test leads must be calibrated to 0 ohm in order to prevent misdiagnosis.
2. Test for 0.35-0.6 ohm at 20°C (68°F) of resistance between low control circuit terminal 1 and high control circuit terminal 2.

So, at least some vehicles have solenoids where 0.5 Ohm is actually a good working solenoid.

S-Prihadi

Please do a load test too on the #53 and #5 wires, minimum 5 amps now, not 2 amps.
If load test pass with flying colors, its either the HPFP fuel pump is bad or the driver circuit inside the ECM for Y94 is bad... I hope its only the wires having many broken strands and
not able to carry the amperage of the Y94 load. Such wire will pass resistance test with ease.

Jeedie

Flying color indeed, a 50W headlight bulb was shining brightly, connected with a loop through #5 and #53. So that rules out the wiring.

I'll get a scope to monitor the ECU output on these pins, there is indeed the possibility that the driver circuit can be at fault. Or the solenoid. Not much else I can think of right now, unless there is an issue on the low pressure side. But my previous testing seemed to rule that out, with at least 50psi pressure during idling and a steady flow.

There is another suspect, if injectors are stuck open the pressure on the rail would not get to the max. However, to basically stay at the same pressure as the low pressure side, would indicate an enormous amount of fuel to leak through the injectors. I don't think this would be a realistic scenario.

CaliBenzDriver

> RECAP so far:
The main issue is proven to be related to low rail pressure.

We are all over the control of an already NEW HPFP. What gives?

HPFP testings :
- The ECU wiring run to HPFP is load tested good.
- The solenoid direct connection shows 0.5R (?!?)
- The ECU can control pressure then self-protects.


> Asymmetrical protected circuit:
- Is there a built-in diode inside that solenoid and/or a resistance ???

- Try Ohming both ways by flipping leads... see if value changes or identical either way.

- If indeed 0.5R is a valid test value that's from an internal bypass resistance to bleep spikes.


> Live actuation:
It is odd a new genuine HPFP would be electrically detective already. Let's test it live with ECU off...

> Can you energize HPFP directly through a test light and listen for it's actuation ? DO match harness polarity in case of internal diode.

🤞

S-Prihadi

Cali,
There is no feature in Xentry to actuate the Y94, as far as I recalled on my engine.
01 - Injector kill yes.
02 - VVT kill yes.
But engine has to run to do these test 1 & 2, otherwise Xentry wont allow.


J,
When you get the scope, do also get the low current clamp to read the amperage value, aside from the voltage value.
Pico 20/60A current clamp can do decent resolution even for under 1 amp.

Now, can we try this "crazy" test :
Take a 3 watts incan bulb, use it as a voltage spike absorber, wired in parallel to Y94 wire #53 and #5.
3 watts should not overload the driver of Y94, I assume.
Or even 1 watt incan bulb if you can get one.

The fact that your wire #53 and #5 is good at 50 watts light is both good and bad news.
You know what I mean.

Jeedie

Another batch of useful advice from you guys, thanks a lot!

It measures both ways..

Interesting idea! I have the old pump on my bench, I'll wire it up and see what happens!

I'm currently awaiting the delivery of a scope and some test leads etc. to start probing what's happening and whether or not the ECU is driving the solenoid. I'll need those readings before I can device a PWM controller to drive the solenoid manually, as I would need some readings of the frequency and signal type first.

The test with a bulb in parallel to validate if there's any control at all is a useful suggestion, I could even go as simple as connect a LED with a resistor to really drive down the power consumption and prevent overloading the driver in the ecu.

Meanwhile, I'm trying to get hold of the right connectors to pigtail the HPFP connector, instead of removing insulation. Hopefully I can source some connectors locally, I would prefer a pig tail to a back probe.

S-Prihadi

Post #7 and #8 has the screen capture of picoscope for the Y94, the red channel.
Post #8 https://mbworld.org/forums/e-class-w...ml#post8554312





.




If the original file, it is in my PC back home.

Jeedie

Just been bench testing the old HPFP, put 25W at 12V through it and it audibly clicks. It already clicks at a lot less power, but 25W was a limit I gave myself. According to the resistance of 0.5Ohm, it might go all the way till 250W, but I’m not risking a local fire on my bench

I’m not yet willing to test the new hpfp this way, I have no idea if it can lead to any damage internally.

Scope should arrive this Thursday, let’s see what that brings.

Jeedie

Nice graphs! This gives me a much better understanding of the workings of the quantity valve. I was under the impression it was held at a certain position for a long time, driven by a relatively constant PWM signal. It's totally different, the valve has to open and close during the pumping actions dictated by the cam speed and # cam lobes,, so there is constant, high speed movement in the valve. At least I now know what to look for once I get everything wired up for testing.

S-Prihadi

in red
Yep, at post #18 in this thread I tried to explain to you how complicated Y94 duty is timing wise....already, but you seems to missed that important information.
That was why I have always wanted to find the camshaft true ZERO degrees in respect to Y94 operation.
Y94 able to CLOSE/Pressurize is one thing, it has do to it in a very short accurate window is the challenge....because we have VVT phaser which is dynamically changing.
The VVT change has no effect on the Y94 physical relationship towards the 3 or 4 fuel lobes on camshaft physical position, but the ECM has to re-calculate everything on the fly
due to VVT, since crankshaft sensor to camshaft sensor relationship is dynamically variable with VVT.
That was why I am always worried and ask myself, can Y94 actuation be late ? I mean mechanically weak solenoid and slow to work mechanically.

We can scope PWM signal, that is the electrical command, but what is the delay of that electric signal into actual mechanical action of closing the Y94 normally open valve ?
I don't know the answer, but Bosch engineers will know the delay value.
Is our Y94 a piezo based ? I doubt, if the voltage is system voltage only at 14V. Piezo need 100V+ and the piezo stack is to be long/many to produce enough movement.

Imagine the Y94 has to travel say 2mm from an fully open state to a fully closed/pressurize state. That mechanical 2mm movement need time to perform.
If the Bosch HP fuel pressure+temperature combo is a high speed enough sensor ( I doubt, because it is too cheap ), you can correlate the pressure rise when every time
the Y94 is closing or has closed shut 100%. The pico WPS500X is a fast 2 millisecond sensor, but it can't handle working pressure of 200BAR and making a safe 300 BAR test fitting
is a DIY challenge.

Lets do some numbers for fun sake. M276 non turbo, 3 fuel lobes camshaft.
5,000 CRANKshaft RPM = 2,500 CAMshaft RPM.
3 fuel lobes for your engine CAMshaft. So that is 7,500 times a minute the Y94 need to actuate and de-actuate accurately.
2,500 CAMshaft RatePerMinute = 41.6 revolution per second.
41.6 revolution x 360 CAMshaft degrees = 15,000 degrees per second, simplified.
15,000 degrees divided by 60 degrees = 250 segments of 60 degrees, in 1 second.
( 3 fuel lobes camshaft need its Y94 to do 60 degrees per lobe as closed/pressurize out of the 120 degrees per lobe )

Per segment of 60 degrees at above scenario has only 1,000 milliseconds divided by 250 segments = 4 milliseconds of actuation window worth 60 CAMshaft degrees.
So ECM has to actuate and de-actuate Y94 in time within the 4 millisecond window, this is piezo based injector speed territory.

I am beginning to think now of another question which been bothering me....
Why the hell is MB HP fuel pressure strategy on my 4 fuel lobes CAMshaft M276 3.0 turbo is lower than 188BAR when at higher RPM ?


I think MB engineers knows the Y94 mechanical side is too slow a reaction time for a 4 fuel lobes CAMshaft....LOL.
For my 4 fuel lobes CAMshaft, it will be 10,000 actuation per minute at 5,000 CRANKshaft RPM
15,000 degrees divided by 45 degrees = 333.33 segments of 45 degrees, in 1 second.
( 4 fuel lobes camshaft need its Y94 to do 45 degrees per lobe as closed/pressurize out of the 90 degrees per lobe )

Per segment of 45 degrees at above scenario has only 1,000 milliseconds divided by 333.33 segments = 3 milliseconds of actuation window worth 45 CAMshaft degrees.

Jeedie

Very interesting, and indeed you outlined the working of the PWM signal earlier in the thread, I just wasn't fully up to speed at that time to comprehend the workings of it.

You posted this in another thread:

Interestingly, on my 3 lobe cam, the max value is 120°, while yours, on a 4 lobe cam, is 90°. If you read that as crankshaft angle, it would translate to 60°/45° camshaft angle, exactly one slope of the cam. If that's true, the activation angle is the angle over the slope of the cam, under which the quantity valve is closed and pressure is generated by the HPFP piston.

Confusingly, in this scenario, the PWM signal from the ECU, as displayed in your graph, has no relevant duty cycle, but only a duration, dictated by the required angle. Then the solenoid basically would work as an open or closed valve, not a partially open valve. While I was under the impression so far that the duty cycle would translate to the amount of opening of the valve.

In the graphs you posted that would explain the difference at the blue arrows (post #41, first graph). In that scenario, the duration of the actual closing of the quantity valve would be dictated by the duration of the PWM signal, where the 1st occurrence has a much shorter closing than the last occurrence. The PWM signal would only control the amount of current delivered through the solenoid, and it might control the speed of opening/closing of the valve, to reduce noise (see Bosch patent S 2016/0076501 A1).

Taking all this into account, the differences between the first and last PWM signal length on post #41 could be explained as a control measure to increase rail pressure. At the same time, the duty cycle of the PWM signal itself might be the current control mechanism to allow ECU control of a 0.5Ohm solenoid without destroying the driver circuit.

Most of this won't help me fix my issue, but it's interesting and entertaining nonetheless. And one idea for an extra failure mode has occurred to me, what if the cam which drives the HPFP is rotated on the camshaft itself? I see more and more issues with modern camshafts, which are nog longer manufactured on a lathe, but are actually separate cams pressed on the camshaft (see http://www.jandeengineering.co.uk/ca...alignment.html). Sometimes, the lobes get out of alignment. If that would happen to the lobe driving the HPFP, everything would get really messed up.

Jeedie

Reading more, I found this piece from GM:


A typical voltage profile to energize the solenoid valve is shown in Figure 3. A pull-in phase corresponds to 100% duty cycle that is intended to move the solenoid armature from rest, closing the valve as fast as possible. Once the valve is closed, the duty cycle is reduced in the hold phase for as long as the valve is closed.

This pull-in phase nicely matches your scope readings, and would give some indication on the delay of the valve, which is taken into account by the ECU.

S-Prihadi

I do not know if M276 camshaft lobes are press fit like how Jandee Egn UK is showing, or a proper CNC/cast one from solid bar. I never seen and touch my camshaft, yet.
Our harmonic balancer pulley is known to spin out of its timing position, due to the internal rubber casting failure.
When one uses such shifted-out pulley as timing guide............. you know the result.

My scoping of the HPFP Y94 was more into understanding Y94 timing for its actuation and de-actuation because Xentry shows that data as actuation angle of the fuel quantity valve,
and I am curious to what it actually meant, in real mechanical action. I was more interested at that point in time on my worry of the possibility of mechanical delay of Y94 actuation and de-actuation with an aging HPFP causing poor fuel pressure boosting due to delayed mechanical action of Y94. Or maybe Y94 could also be in a failure mode of not fully closing...after all it is a mini valve.

I have Googled a lot back then, and no one is discussing actuation angle of the fuel quantity valve with scope waveform capture assist for us to learn.
It is common people only wants to learn of things when they have issue with the "thing". Me I want to learn before those things go whacky on me.
Complete failure of an a device is easy to troubleshoot , whacky behaviour and intermittent is not easy to find.
Hence capturing waveforms on my engine while all of the devices are still healthy, is part of my list to do.

There was a case study on a Bosch crankshaft sensor, at Diagnose Dan. He is a Dutch techy, very smart man.
He got a brand new CKP, healthy one..... but it has reverse initial direction of the waveform., so ECM does not work with such waveform and crank no start.
Example below , of what I described as reverse initial direction :

Here is the video :

S-Prihadi

A lot of solenoid uses this power pulsing method. Such pulsing is called peak and hold. Solenoid based Injectors often is a peak and hold
Supposedly faster response and cooler operation of the coil. This is still only an electrical signal and does not reflect the actual mechanical result, which need some sort of pressure
transducer to verify. Electrical signal is the input, mechanical pressure is the output. We must see both.

S-Prihadi

My schedule to be back is next to Jakarta is next Monday 9th Oct.
By 11th Oct I hope I can assist you in reading my own Y94 amperage values while in operation, I have never measured it yet. So you have something to compare it with.
Reading that amperage value is a bi-etch on the need to open up my wire tray, its clips are not nice to work with if I do not want to break them.
M276 3.0 has wire tray with cover, unlike M276 3.5 NA.

While at it I will look into the WIS and give you the MB official cams waveform + crank waveform + ignition firing sequence based on crank degrees/waveform.
If you already have a scope by yesterday Thursday, you best collect the actual CAM + CRANK waveform , so you can verify timing chain stretch in real time.

I have another troubleshooting request I need to attend to on my friend's yacht. His 10kw 24V DC bowthruster is not working.
So, next week will be fun , I get to exercise my itchy fingers for 2 good reasons, your Y94 and my friend bowthruster.

If my PWM tester box can handle Y94 current, maybe I do table test on my spare HPFP too, from PWM persepective and use air pressure to verify the valve opening/closing.
I have a cheapo China made PWM generator. This one : https://id.aliexpress.com/item/1005004255904120.html

Jeedie

The scope arrived yesterday, first I put into to good use troubleshooting a new alternator on another car with so called Ford Smart Charging. Well, that new alternator is not playing ball, now proofed by the scope.

Next was the E300. I attached back-probe pins to the HPFP solenoid, carefully arranged the wiring close to the bulkhead, refit the manifold etc. etc. and started the engine.

This is the first startup, measuring the PWM signal on the solenoid 12V input side. The engine did not ignite here, the pressure in the rail got up to 25bar. Luckily, the ECU kept the quantity valve closed at the end, so I could record the live fuel rail pressure decline on the real, which would show any leaking injectors. Pressure drops by 0.6 bar (9psi) every 10 seconds, so it seems the injectors are actually closed. Happy I was able to capture that moment.




So, the ECU is at least controlling the solenoid, that's a great insight.

Zooming in on the duty cycle of the PWM signal gives this info (my scope runs at 25Mhz):



That gives some info in how the ECU copes with the low resistance of the solenoid, the duty cycle to keep the valve closed is pretty low it seems. So the PWM signal is managing the current through the circuit.

Anyway, second run shows a similar pattern:




Here, the pressure ran up till 40 bar and then dropped again etc. From the scope, it's clearly visible that the quantity valve gets the correct control signal. Once the pressure does build up, the valve is closed for a shorter duration, once the pressure goes down, the duration increased up till the max, where the valve is closed for 50% of the time.

With that info, I see the following possible causes:
- The new HPFP is actually not working properly and needs to be replaced.
- The ECU is misreading the cam position sensor and driving the solenoid at the wrong intervals.
- The cam lobe has moved / is moving, which is messing up the timing of the quantity valve.

Any suggestions to test these scenario's, or should I just splunk out roughly $750 and get another HPFP?