204 DSG intermittent momentary power loss

[n10n]

If you floor it in neutral, the engine won’t spin up as quickly as that so it definitely cannot do it under partial/full load.

Correct. But if you think the case of accelerating, pedal at 80% or so, rpm is limited and resisted by the torque through gearbox and ultimately grip on the road, right? It that resistance goes away suddenly and unexpectedly, rpm might jump pretty aggressively.

If you floor it in neutral there's no torque request from gearbox present and ECU might behave differently.

It also happens during steady state.

If it would be clutch it could just be more apparent or frequent around shifting.

Well, anyway, above was just a line of thought, let's check what the next measurements reveal!

 

[ben3042]

Correct. But if you think the case of accelerating, pedal at 80% or so, rpm is limited and resisted by the torque through gearbox and ultimately grip on the road, right? It that resistance goes away suddenly and unexpectedly, rpm might jump pretty aggressively.

If you floor it in neutral there's no torque request from gearbox present and ECU might behave differently.



If it would be clutch it could just be more apparent or frequent around shifting.

Well, anyway, above was just a line of thought, let's check what the next measurements reveal!

Yeah I completely see your thought process and it is totally valid! Really appreciate your input!!

 

[ben3042]

It's quite difficult to comment the MAF signal - certainly it is affected by the glitches but I'd like to think it's the consequence. I have thought the error in air mass shouldn't have drastic sudden effect on engine power. Also EGR valve does very rapid movements which affect significantly to air flow.

Re the picture I don't remember seeing the blue stuff on mine (a few months ago) and mine was very clean. Anyways, not a bad investment to replace it as they won't last forever.



Yes, I would put my money on engine speed sensor - especially now knowing about implausible signal (of ?) in the past.
Well, if you drive would be interesting to have a log with these parameters - to trace any delta between crankshaft and camshaft sensors. In the list there are several items related to engine RPM but not exactly sure if they are any different - if even the item named as Camshaft actually is camshaft.

IDE00021 Engine RPM​

IDE00075 Vehicle speed​

IDE00182 Camshaft adaptation intake bank 1: phase position​

IDE00347 Air mass: actual value​

IDE00352 Main injection: duration of activation​

IDE00405 Crankshaft speed (RPM)​

IDE00406 Camshaft RPM​

IDE04685 Setpoint generation interior torque​

IDE07742 Cylinder head: pressure sensor 3 bank 1: raw value​

ENG113719 ESM_Engine_speed​

ENG126001 P_L_Aps_crankshaft_speed​

ENG126043-ENG126092 P_L_Aps_camshaft_drift_angle-Camshaft target wheel drift angle​

​

This also relates to new capabilities with modern controllers of HEX-V2 cable

Hi MMI,
Right, new data file attached from my drive to work this morning. Marker 1 was at the start of the run and I think it glitched but unsure. Marker 2 was an actual glitch but I was about 2 or 3 seconds late with the marker. Marker 3 was an actual glitch and right after the glitch. Im trying to open the data myself but excel keeps crashing. So you might beat me to it Ill post my thoughts once I can get something plotted

 

[ben3042]

Well.... Something definitely isnt right....

Marker 2


Marker 3



Last 500 ish seconds


So the camshaft signal is flat lining A LOT!! and on the glitches the crankshaft rpm jumps. At marker 3 it dropped 500rpm, rose 1000rpm, and dropped 500rpm (appreciate this would look slightly different with higher logging frequency, but none the less) in 0.2 seconds. But what would be absolute sense is that during that 1000rpm rise, the ECU would think 'crap the engine is spinning up, enable traction control' and cut power, which is what I am feeling..... so.... is this a dodgy crankshaft sensor.... or is the flat lining of the camshaft sensor to blame...... Any thoughts?

 

[n10n]

To my eye that crank speed makes much more sense than cam speed, even with the sudden jumps.

For many sensors ECU can provide a substitute value if it notices an implausible signal. For example here ECU could use crank speed as a substitute for cam speed to keep engine running although not optimal. What strikes me as strange though is that no faults are recorded although the cam speed is obviously out of spec.

Quote from Bosch Automotive Handbook, Management for diesel engines, 11th ed, page 914:
"Example 2
If the camshaft position sensor fails, the ECU uses the signal from the crankshaft speed sensor as a substitute value. Different vehicle manufacturers have different concepts which are used to determine via the crankshaft-signal characteristic when cylinder 1 is in the compression cycle. The restart, however, takes a little longer as a consequence of these substitute functions."

Also I find it very weird that the camshaft speed seems to be correct in lower rpm but almost like capped to a maximum value. I'm not sure how the cam speed sensor is technically implemented in T6 but that kind of failure mode just feels unlikely.

I need to go out and see what my cam speed reports.

 

[DXX]

I'm not sure how the cam speed sensor is technically implemented in T6 but that kind of failure mode just feels unlikely.

 

[n10n]

So, a hall sensor. As that is sending out pulses dependent on speed still not sure how that could cap on some max speed. If the pulse wheel would be damaged it should show also on lower rpms.

 

[DXX]

If there is a fault I would expect it to be a wiring fault rather than a sensor.

 

[ben3042]

Thanks guys.
I have seen hall sensors fail at high frequency before, so I'm hoping a new sensor (I have just collected one, fitting it this evening) will resolve the capping. Annoyingly Eurocarparts let me down for the crank sensor and that now wont be ready for collection until tomorrow. But I'll get the cam sensor changed this evening and report back with some data.
I'd disagree with it being wiring because the sensor would drop out irrelevant of engine speed, whereas here it is always capping above approx 1500rpm.
I'm suspecting crank sensor issue as the signal increases in speed, at the point where the engine is cutting back and slowing down. But welcome anyones thoughts on this!

 

[DXX]

I'd disagree with it being wiring because the sensor would drop out irrelevant of engine speed, whereas here it is always capping above approx 1500rpm.

Engine vibration changes with RPM / load.

 

[mmi]

So the camshaft signal is flat lining A LOT!!

Well, mine does the same - interestingly the data item marked in VCDS as "Camshaft speed" saturates at approx. value 1400 RPM. It follows the Engine RPM only below. Also the point of saturation seems to vary a bit.

I'd like to state the behaviour is perfectly normal - otherwise would fairly soon see some kind of "unplausible signal" fault.

 

[ben3042]

Well, mine does the same - interestingly the data item marked in VCDS as "Camshaft speed" saturates at approx. value 1400 RPM. It follows the Engine RPM only below. Also the point of saturation seems to vary a bit.

I'd like to state the behaviour is perfectly normal - otherwise would expect fairly soon to see some kind of "unplausible signal" fault.

View attachment 210924

View attachment 210923

Yeah, I've just replaced the cam shaft sensor, for the sake of £20 and 20 minutes work. Just been out for another drive and yes, it still saturates and the van is still glitching..... data sheet attached along with a new blockmap file which completed properly this time.
There are 13 markers in this data file...!!

 

[ben3042]

Another log attached but a very short run with 3 markers in. And another new blockmap. I noticed in the last blockmap that the battery voltage was only 12.5v ish when idling, so i took the main engine ground cable off and cleaned it all up. Idling voltage is now near 14v, but it's made no difference to the glitching.

 

[ben3042]

Crank sensor fluctuates first, then injection is cut... so the ECU is reacting to the engine suddenly speeding up "apparently" and then cutting injection..... thats my take on it anyway..... new crankshaft sensor will be going in tomorrow.
What do you think @mmi ?

 

[n10n]

battery voltage was only 12.5v ish when idling, so i took the main engine ground cable off and cleaned it all up. Idling voltage is now near 14v,

Of course good to clean but 12.5 is perfectly normal. Mine can go down to 11.9 as measured from diagnostic port. Smart alternator is on only when battery SOC is considered low enough. All the consumption will pull apparent voltage down when alternator is off.

 

[mmi]

the van is still glitching..... data sheet attached along with a new blockmap file which completed properly this time.
There are 13 markers in this data file...!!

Crank sensor fluctuates first, then injection is cut... so the ECU is reacting to the engine suddenly speeding up "apparently" and then cutting injection..... thats my take on it anyway..... new crankshaft sensor will be going in tomorrow.
What do you think @mmi ?

That's my thought how it's happening.

---

Anyways, below an interesting detail - at least to me. Plotted cylinder pressure because the drop in that can be felt at driver's seat.

Initially plotted IDE00021 Engine RPM and IDE07742 cylinder pressure vs. markers. Noticed cylinder pressure (blue) drop at 393.5 seconds (because of marker at 394.4 s) but no glitch in Engine RPM (dark blue).

However, adding ENG126001 P_L_Aps_crankshaft_speed into the plot (by offset of 1000 here) reveals that there is indeed a glitch in RPM data there.

Well, the IDE00021 and ENG126001 are actually 0.05 seconds apart from eac others and obviously from the same sensor. Thus in this case it seems we managed to double the sampling rate for crank signal - revealing the secret of power loss at 393.5 seconds. Or am I misinterpreting this? A good theory anyways?

 

[ben3042]

That's my thought how it's happening.

---

Anyways, below an interesting detail - at least to me. Plotted cylinder pressure because the drop in that can be felt at driver's seat.

Initially plotted IDE00021 Engine RPM and IDE07742 cylinder pressure vs. markers. Noticed cylinder pressure (blue) drop at 393.5 seconds (because of marker at 394.4 s) but no glitch in Engine RPM (dark blue).

However, adding ENG126001 P_L_Aps_crankshaft_speed into the plot (by offset of 1000 here) reveals that there is indeed a glitch in RPM data there.

Well, the IDE00021 and ENG126001 are actually 0.05 seconds apart and obviously from the same sensor. Thus in this case it seems we managed to double the sampling rate for crank signal - revealing the secret of power loss at 393.5 seconds. Or am I misinterpreting this? A good theory anyways?
View attachment 210944

Completely understand what you are saying. I ‘think’ the engine speed trace is an amalgamation of the crank sensor, the ECU’s smoothing factors and influence from other sensors like cam sensor which is why there is a short offset in engine speed trace but also explains why the spikes seen on the crank trace are more prominent than on the engine speed trace. I’ve noticed that same pattern at every single glitch. It’s the rise from the initial drop in crank signal that triggers the fuel injection cut - I.e. the rise is so steep that the ECU thinks it’s overrunning to cuts injection and subsequently dips in cylinder pressure and air intake.
This theory all seems to check out but I’m now just itching to get the new crank sensor tomorrow to prove it all!
I will report back as soon as possible!!

 

[n10n]

I ‘think’ the engine speed trace is an amalgamation of the crank sensor, the ECU’s smoothing factors and influence from other sensors like cam sensor which is why there is a short offset in engine speed trace but also explains why the spikes seen on the crank trace are more prominent than on the engine speed trace

This is a very good summary and I completely agree.

In general it’s just such a shame we have to guess the significance of many measuring values. But they probably correspond to different measurement points within ECU - eg IDE000021 as "release" value and ENG126001 as the actual input. If the crank sensor turns out to be the issue I think ECU detects the implausible signal and substitutes with a more plausible value for release, yet ramps injection down any case in order to protect the engine. Might be the fault is not recorded because it's just a glitch and returns to normal soon after. But again, just guessing.


"The level 1 software contains the desired application functions. These check the plausibility of input signals supplied by different sensors and thereby blank out invalid sensors and implement appropriate substitute reactions. The level 2 software calculates in parallel in an independent computer core parts of the application software in a redundant, independent implementation (e.g., generation of torque or triggering the starter). The output stages are enabled only when both levels arrive at the same result. The level 3 software checks hardware-related faults (e.g., damaged memories)."

Good luck with the crank sensor change and hope it’s the culprit!

 

[CJW]

Well, mine does the same - interestingly the data item marked in VCDS as "Camshaft speed" saturates at approx. value 1400 RPM. It follows the Engine RPM only below. Also the point of saturation seems to vary a bit.

I'd like to state the behaviour is perfectly normal - otherwise would fairly soon see some kind of "unplausible signal" fault.

View attachment 210924

View attachment 210923

I wonder if this clipping is to do with the sampling rate of the ECU topping out. 60 teeth on the disc on the end of the shaft, at 1400rpm, it needs to sample to find the on and off edges of the signal, so it makes sense that the sampling resolution might be, say, 20 samples per tooth. At 1400 rpm that’s 28khz.

 

[mmi]

Well.... Something definitely isnt right....

Marker 2

That's quite a fascinating capture.

The Camshaft speed was "latched" at 1413 rpm, during the glitch the Engine RPM went below threshold, and immediately above the threshold. Thus Camshaft speed was "latched" again at 1599 rpm (by the way, record high value seen this far).

Did the recording miss the 1599 rpm's - perhaps not?