Mystery running issue...

A

aragorn

"Stick a V8 in it!"
Staff member
Moderator
VCDS Map User
Joined
Jan 30, 2008
Messages
18,708
Reaction score
909
Points
113
Location
Glenrothes, Fife, Scotland
So i've been working on the track car as per the other thread. On thursday i picked up a MAF for the track car from the scrappy.

The bloke said if it was faulty i could return it within 7 days, so i thaught i'd best test it out.

I connected it to shed and went out for a drive. Everything seemed normal, until i tried a full throttle pull right thru second gear. Above 5k the engine seemed odd, slightly hesitant, and once approaching 6k it felt a bit like a rev limiter, with the power coming and going in a slightly pulsey fashion.

I thaught "baws, this MAF is faulty", and refitted my original one. Then thaught i best check the problems gone away so took the car out for a quick test.... It hasnt.

I drove it again tonight, still with my original MAF fitted, and its still doing the same thing. Feels fine upto 5k, above 5k it starts to "break up" almost like a slight missfire, and you get this pulsey hesitant sensation. I decided to "keep the boot in" and see what happened, and it got even weirder. Beyond 6k, accelleration seems to die away, sounds almost like a rev limiter, albeit a very soft one. Kept my foot in to see what happened, and it slowly climbed to about 6400rpm then just seemed to stop accellerating, like a diesel does at the governor...

Now bear in mind this is 2nd gear, and it will usually happily pull right round to the limiter without any problems. Its not like i'm in 4th or 5th trying some Hero speeds. I tried revving it in neutral, and it bounced the limiter just fine.


So then i began wondering when the issue might have started, and i remember when driving the car on wednesday, i had a blast down the dual carridgeway from stationary off a roundabout. I remember thinking the engine felt flat towards the end of 2nd and in 3rd, but just sort of chalked it up to it being warm and the intercooler getting heatsoaked from a longish pull (car has stock intercooler and stock tune). But ofcourse i've had the car for over 2 years now and never recall actually noticing performance dropping off due to heatsoak. Not that it doesnt degrade, but just doesnt happen in such a noticable manner...

Thoughts, ideas?

(PS, i've lost my VCDS lead, so no fault codes have been found nor have i managed any logging yet, replacement is on order!)
 
I guess i'll throw some new plugs at it just to rule them out. They're not that dear.

Oddly its fine at full throttle at lower RPMs where spark issues usually come to light!
 
Hmm . "Oddly its fine at full throttle at lower RPMs " would that would rule out fuel supply as well then ?

leaves boost/air and sparks.

What type of plugs would you use mine were changed some 8k ago don't know what was used [if any] as it was done by a previous owner....hope it sorts things out for you.

cheers

Mo.
 
They're probably the BKR6E's i fitted nearly 3 years ago when i built the engine. Well overdue for a change :p

As for fuel supply, fuel demand follows the power curve, so fuel demand is highest at the top end. It could be fuel related, the pump is a bit noisey, but no real way to tell as M3.8 doesnt log AFR's
 
OH ***

Replacement VCDS cable arrived today, and its so cheap and ****e it doesnt fit the ****ing socket.

RAGE+
 
Got angry, attacked it with a flapwheel in the grinder

It fits now!

Did some logging, and it looks like the MAF is stuffed.

Timing pull is up in the 15's, and MAF reading peaks at 125g/sec, where it should be in the 140-150's
 
"Timing pull is up in the 15's, and MAF reading peaks at 125g/sec, where it should be in the 140-150's "

Sorry I'm not trying to come across as a mong but trying to learn could you explain that please. I think you mean the ign timing is at 15 degrees advance [??] and the maf is sending a low signal compared to the actual airflow as per rpm...

not trying to be funny just want to learn...

Mo.
 
Its not an actual timing value, but a measure of how much the ecu is having to ****** the timing due to knock, the ECU calls it a "Correction Factor"

When the management detects knock it ******* the timing, and you can log this process. Low single digits is normal, perhaps <6, high single digits is getting a bit iffy. 15 is "off the scale".

Theres a few things happening at once causing the problem. The MAF is the main load input into the ECU, so the engine fuelling and timing are all derived from the MAF reading.

So because the MAF is under-reading, its only injecting fuel for 150hp, rather than 180. This causes the engine to run lean, which causes knock. Furthermore, its also trying to run the timing values for 150hp, which will be over advanced, and also casues knock. Thirdly, because the engine doesnt have a boost sensor, it uses the load from the maf to calculate boost. When it realises it isnt getting the 180hp it wants, it tries increasing boost to get the airflow up, meaning you've now got even more air, and are running even leaner!

Log is here if you'd like a look:

https://www.dropbox.com/s/7hxxlx2df7wcoqd/LOG-01-007-002-013.CSV
 
Thanks aragorn, that's a clear explanation of how the ecu uses mass air flow to derive it's whats nots.

So they only have the one heated wire element as a sensor, bit crap that considering what runs of it.

Gotta get me some kind of vcds, no option really if you want to run and maintain these cars...
 
okay got that log thanks, quite a lot of info on that; but I think it was the columns H-K along with timing columns m-p read in conjunction yep, see that.

You need to take a step back and understand what you are seeing to confirm what you may/may not be thinking.

thank you very much, good stuff.

cheers

Mo.
 
VCDS logs three blocks simultaneously so there are three data sets, B-F, H-K, L-P. If you look at the timestamps you can see they're logged one after the other then it starts again at the first block. This means logging less data (ie 1 or 2 columns instead of 3) gets you a faster sample rate.

Unfortunately i logged lambda (b-f) to see if i could see any weird fuelling, but M3.8 doesnt really log lambda properly unlike the later ECU's when at full throttle. It does log lambda voltage, and that voltage stays >0.8v throughout the pull, which suggests the fuelling is at least richer than stoich, which is about all a narrowband sensor can tell you. IE i dont know if it was at 13:1 or 12:1 or whatever.

If you also look at the timestamp, you can see it jumps from 240 to 280, this is where i paused the logging after the first pull and then started it again when i did the second pull. I havent logged throttle position, but you can see where i've gone full throttle because the charcoal filter solenoid goes to 99.


Aye this is the problem with diagnostics. You can never take things on face value, you need to look at the whole picture and consider whats going on. If you did a simple "scan for fault codes" you'd never have picked that up, as the ECU was happily saying no fault codes found.

Fault codes are obviously useful, but folk are sometimes overreliant on them. Quite a common example is folk get a fault code such as "Coolant temperature sensor: Implausible Signal". Without thinking they rush off and buy a new sensor, and oh look, the faults still there... You really need to step back and think about whats happening. If the coolant sensor never reads above 70c after a long time driving, the ECU is sensible enough to know thats not normal and something is wrong. It doesnt know WHAT is wrong though, so it simply sets a code saying the signal its getting doesnt make sense. Now that might be due to the sensor being faulty, but it might also be that the sensors fine, but something else is causing the temperature to be low (such as a faulty thermostat). You can usually apply some simple tests to try to narrow down the fault. This is when the live data comes in useful. If the sensors saying -38c then you know its knackered and can replace it. If its giving a plausible reading, you can compare it to the instrument cluster reading and see if they match, if they do, its probably a mechanical issue etc etc.
 
Ok getting ****** off with this now, tried two different used mafs and it still wasnt right, so i stumped up and bought a new one from euro car parts.

And its exactly the same.

The last ebay maf and the new maf have also introduced a new symptom, its now not boosting past 5-6psi.

Now with the lack of boost the airflow has dropped to 100ish and timing corrections are as high as ever.

New log: https://www.dropbox.com/s/hr6v2sr9ncgxt3h/LOGNEW-01-022-013-xxx.CSV
 
Did you check the plugs? Mine doesn't boost over 10psi but sometimes cuts out at 5psi as well. I went through pretty much everything now boost related other than turbo and still the same. Why do you think it is maf? Maybe your maf was bad and when you changed to new one it highligted another issue?
 
Replaced the plugs with new, but the old ones werent really all that bad.


I thought it was the MAF becuase the airflow figures in the log are all wrong.

The first logs the car was boosting MORE than normal (over 12psi), but only registering 150hp worth of airflow on the maf sensor.

The last two logs its even worse, down to 120hp worth of airflow, albeit making less boost.

The timing corrections on the log are also indicative of the engine trying to run WAY too much timing, which would happen with an under-reading maf.
 
leaks would put the maf figure up, not down.

I suspect the knock is getting bad enough that the ECU is pulling boost to try and sort it out.


Leaning towards the possibilty of a blocked cat now. Going to try driving it tomorrow with the lambda probe unscrewed.
 
Sploob question, how does the ECU know the "knock" value, sensor or some calculation.

OR is the knock actually "real"...

unrelated story, bare with me ok, ignore if you want.

I rebuilt the top end of an cvh engine after a head gasket fail post stat fail ok. Head skimmed true, compression calcs done [ok not on a fag packet but A4 sheet] head heavily flow worked and a high lift kent cv22 can used.
Engine ran perfect "around town"UNTIL you gave it some berries then pre ignition all the time no matter what until engine cooled.

What it was in the end AND I REMEMBER LOOKING AT THE PISTON CROWS AND SEEING 0.005" ON EACH ONE but just not thinking "out of the box" was the fire ring of the hg protruding into the cylinder it was ok running about ie it could transfer enough head away during town driving but give it enough boot and the fire ring stayed hot enough to cause detonation after that...fitted the hg from a turbo engine, sorted.

just throwing that into the mix as it was, don't want to step on anyone's toes.

hope you get to the bottom of it aragorn.

atb

mo.
 
If cat was blocked would you not notice a differance in engine/exhaust tone ??

If your maf is under reading would that not point to air leak before the turbo ?? Ie pcv or evap ??

Whats the lambda correction reading ??
 
Do you see a difference between cold and warm engine?
Mine runs fine when cold. As soon as it reaches temp, that's it.
 
Checked the Pre-turbo pipework, its all fine. The TIP looks a bit perished and probably wants replacing, but theres no actual splits or holes. The dump hose for the DV is similarly perished, so i need to get a new one of those, but again, its not actually holed.

And ofcourse, actual leaks would throw lambda correction out. Lambda correction seems normal, fuel trims are <1%

Car has a factory exhaust, so theres not really much in the way of exhaust noise to notice changing.

I havent tried driving it hard when its cold, i dont like doing that and tend to keep it off boost or very low boost until the temp guage is up to 90.

mojo: your story sounds similar to whats happening here, but ofcourse my engine internals havent changed but its obviously getting hot after a pull and thats letting the knock continue. The engine has two knock sensors attached to the block and the management uses the data from them to control knock. If the ECU is managing to control it, then its not actual knock. Within normal bounds you'd get some CF as the ECU just trims down the ignition timing, thus meaning the timing is always optimal, so even on a stock tune you'll get some low CF's being recorded at full load etc. However the amount this is recording is pretty much off-the-scale, and its a little worrying that if its bad enough the ECU becomes unable to actually keep it in check and it turns into real knock that can wreck the engine.

Continuing the blocked exhaust theme, i noticed its also very slow to come on boost. Usually you jam the throttle open above 3krpm and the boost guage flicks instantaneously round into >5psi, whereas these last few tries, its a slow lazy sweep of the guage from 0 upto about 6.
 
What are the actual "knock" signals from the sensors?

Hmm..

Knock Sensor : Volkswagen UK


"The knock sensor prevents damaging spontaneous ignition in your engine. The knocking effect is created by uncontrolled combustion at the wrong phase of the engine cycle, causing extreme temperatures in the cylinder. This places a severe strain on engine components such as the pistons, valves and cylinder head, and can lead to damage. The knock sensor registers the sound of the engine and converts it into electrical pulses. The engine control unit cross-checks the measured pulses against the stored specification values and intervenes to control the engine, fuel injection and ignition as necessary, before the combustion reaches the knock limit. The sensor also detects the fuel quality. The higher the octane number, the more knock-proof the fuel, as it ignites at a higher temperature than normal petrol. If only a lower quality of fuel is available, the engine control unit automatically corrects the ignition timing, based on the information from the sensor. This prevents damage to the engine. "

Any signal conversion done at the sensor? yes.

Is their output analogue, frequency or some sort of digital?? Frequency from above ^^ although some state 0-5mV????



Is the FPR actually allowing the fuel rail to develop the correct pressure for the feed side of the injectors ie a physical limitation on fuel flow into the engine?

But it's only doing it hot [?] so that's a long shot for sure.

What fuel is in the tank?

How old is it??

Another unrelated story..Shell stuff [optimax I'm talking about ok] had a fairly short service life before additives fell away. When I was running the mojo it was on twin 40's, clockwork ignition, no cat.
I mixed up my own brew in the 20l tank with optimax and tetraboost to give about 103-6 octane leaded, car ran like a dream despite lumpy cam, open jetted carbs etc. On normal fuel it ran like a pig, I carried a 13mm spanner to ****** ign [ie turn dizzy] to the "low" mark if caught without some tetraboost in car, nasty stuff mind...

I take it there are no code lurking for any ecu problems?

Mo.
 
Ok small update.

After driving around looking for the lambda socket and get getting everything roasting the lambda simply unscrewed with a spanner.

I also noticed that i'd forgotten to reconnect the actuator vac line, meaning the last run i'd done was making only 6psi with no boost control at all. Everything pointing at the exhaust restriction now!

So i just did a log with N75 unplugged and lambda removed. 100g/sec peak, but with only 3psi of boost (actuators clearly fubar then!) and no knock at all.

Plugged N75 back in and tried again...

With N75 on it hit 10psi immediately, however it got to about 4krpm and started missfiring really badly so i lifted off. Airflow got to 110, which isnt too bad considering it didnt rev, and again no knock at all. Getting somewhere!

The lambda probe on the A4 points straight at the airbox, which when i stopped was too hot to touch... I also dont have the intake snorkel fitted, so i'm thinking the missfiring was due to pouring 130hp's worth of exhaust straight into the airbox :p That or the ignition amp was overheating, as again its right in the line of fire and was scalding hot.

So got home, let it all cool off and unbolted the cat, and after much f-ing around wrangled it out of the engine bay...

Ah:

https://www.dropbox.com/s/g9cnltuq5353oe8/2014-07-11 13.03.44.jpg

https://www.dropbox.com/s/vo8bwnfbemwsa6w/2014-07-11 13.12.50.jpg

Monolith has broken free, and dropped into the bottom of the casing, where its then shaped itself to the outlet and forming a "plug" with only a tiny portion of the core actually letting gas thru (you can see the black ring). Explains why the cars tailpipes have been white and dusty recently!!
 
This is my daily, so it needs a cat. I know the thread starts talking about the track car, but all of these issues were on my normal daily driver. Its easy to miss though now i look back :p

The wifes car got a nice 3" DP with sports cat a few months ago and i'd kept her old one just in case, so i've fitted that.
 
So it's all good now? Did the cat make any noise under acceleration?
 
Thinking about it.....I went through 3 cats on the old tdi with a tuning box fitted, think it was just **** cat construction tbh. Now about the tensioner issues....
 
Yeh its fixed.

The only noise i heard (and have been hearing for a while) was a sort of tickety chuffy noise, most noticable driving round the street at low loads, with the windows open. Sounded almost like a blown manifold gasket or something.

Since swapping the cat, that seems to have gone completely.

This was a genuine original cat and has done 160k and 14 years so TBH its not bad going! The one i've fitted is equally old with a bit less mileage, so it probably wont last forever.
 
  • Like
Reactions: slepy
You must log in or register to reply here.

Similar threads

MaverickTDi
Engine running issue
Replies
0
Views
499
MaverickTDi
MaverickTDi
MaverickTDi
AFN running issue
Replies
4
Views
621
MaverickTDi
MaverickTDi
Blackie_No1
1996 A4 2.6 (ABC) running issue??
Replies
2
Views
1K
aragorn
A
MattS5
Interior illumination mystery
Replies
5
Views
691
OutLore
OutLore
A4-Newboy
Running lean?
Replies
4
Views
1K
A4-Newboy
A4-Newboy
Top
Back