If you have the ross-tech VCDS dongle/SW, there's a free software add on called "VC Scope" you can download that shows some of the data visually and in real time, which may be a help in chasing this down.

Only ever used it a couple of time, but you can have this connected and showing graphs during a drive, with a passenger holding the laptop, and looking for any sudden changes/clues in the data when the issue happens. I think load is one of the data points and might help to see if there's any correlation, along with RPM, fuel, etc. Maybe taking a video of the screen when driving so that you can audibly or otherwise signal when it's occurring in real time to compare the data 'reaction to' or perhaps cause or correlation of the occurrences. You'd want to try and duplicate the condition several times to see if the reaction/behavior is consistent at a certain rpm, load, fuel ratio, etc., etc,

Download link is here, if you don't have it already.

https://www.ross-tech.com/vag-scope/index.html

 

Thanks for the link, I do have VCDS so I'll download this and hopefully do some logging tomorrow evening

 

Ok, did a bit of logging tonight for the first time.

I couldn't get address 31 engine other to display anything more than 1 data stream at a time so I've had to record a bit of driving from each bank individually. Not ideal I know but not sure how to get round this?

I've attached the 2 csv files (hopefully)

Thanks,
Lewis

 

Hi Lewis,

I looked at your posted files. These aren't going to be that helpful as the timestamps from the three sectors and the second file are so different that data do not track enough to compare bank 1 with bank 2.

I manipulated data from both docs so that at least you can track data from bank 1 with bank 2, but BC the time stamps are different, data will not make sense to track an anomaly. I personally am concerned both of your ECUs are good, but again, with time stamps so far off, I cannot tell for sure.

If you look at the file I attached, for simplicity, I pulled some data that took up columns but isn't going to help with anything, such as - Coolant temps (these were so steady that not helpful) and throttle position angle (%). Throttle angle has three actual values - Commanded, virtual, actual. Even though the data collected will be actual, the throttle angle is reflected enough in several other data such as MAF (g/s) and timing (to an extent).

I'm looking at these numbers but it is a little bit like a 1000 piece jigsaw puzzle, BC of the timestamps.

 

Hi BWings,

Thanks for taking a look.

I did try to record both banks simultaneously by going to address 31 engine other but I could only view 1 data stream at a time and when requesting to view 33 long term fuel trim the data returned didn't make much sense as it just said something along the lines of 'bank X'

So I had to do a bit of driving recording bank 1 and and bank 2 sequentially which to be honest I did not think would be very helpful.

Is there something else I need to do to view and record data from both banks?

Thanks

 

Lewis:

I further edited data to isolate just Lambda %s bank 1next to bank 2. Lambda is a terribly flakey number using just ODBii as opposed to using a wideband. AFR is essentially estimated BC O2 sensors are simply not capable of producing accurate AFRs. The difference between a WB and an O2 sensor is like a scalpel compared to a battle axe.

So, keeping that in mind, if you look at the Lambda outputs Bank 1, Bank 2, Bank 1 is producing some extraordinary swings, numbers t you don't typically see. Bank 2 appears a little more normal. But, if the bank 1 O2 sensor 1 is failing it should throw a powertrain fault code.

Remember because I pulled these two columns from different logs you cannot compare any rows (row to row) just a big look and rough comparison of raw data in the two columns.

You should try to capture just fuel trims and O2 sensor voltage outputs both banks. Use the instructions for VCDS that wee published above and use the final graph feature.

 

Hi BWings,

Thanks for your help so far.

I see what you are saying looking at the file attached to your reply but equally I understand we can't directly compare these 2 sets of data as they aren't from the same drive cycle.

What do you mean when you say extraordinary swings, numbers you don't typically see?

I will do some more research and have another go at logging both banks simultaneously tomorrow.

P.s. no check engine light and no fault codes stored at present

 

Might also help if you have noticed any additional correlation between when this happens and any particular set of RPMs and Speeds. If it's completely random, then that information is also helpful. Does it always occur at 4th gear 3k rpm? Does it only happen at that point? What other combos? Does it feel like it might just be a hard shift? Do you notice any RPM drops, pauses in the speedo or tach, other noises or anything else that happens at the same time? (dimming lights, ac blowing less, etc.)

You might need to try to duplicate the symptoms, observe the speedo and tach, and keep note of what those two pieces of information are whenever the jolt happens. If they happen to occur at specific speeds/rpm combos, it might indicate a transmission related issue, if they match the shift points. Also, might help to give some clues as which time points to look at in the data, based on RPMs as the correlation could potentially be with any of the data points.

I think the hard thing is that if it's only happening very briefly, it won't leave a big tell-tale signature in the data, and depending on the capture rate of the logs and the brevity of the event, might not show at all.

Another reason I prefer the visual / real time approach to help narrow down which data points to consider. You could do a screen recording with the various readings showing real time in Scope graphs, but you would have to have a way to 'mark' when the jolt occurs. Perhaps a mouse move or something to indicate the event just occurred so that if any 'needle jumps' happen on the graphs that might happen around that time can be noted. You'll need a partner in crime to do this safely, unless you can set the laptop to record and perhaps have a wireless mouse you can quickly jostle or move in a deliberate way, without looking down and losing your attention on the road. Capuring 4-5 of these events may help reveal something.

Were it my car, if it doesn't seem to be associated with shift points, I'd take a good look at the MAF, throttle body and vacuum lines etc. to make sure they are clean, clear, unbroken have full freedom of movement, and are operating per spec., just as a pre-emptive strike to see if that helps, because it's easy to do and helps to potentially rule out one thing. Visual inspection of O2 sensors also, if you can see them.

I'd also see check the plugs for any fouling or crossfire in the dark, because that's also easy to get to (but vastly more quite time consuming vs. other cars that don't have the coil-paks under the intake manifold)

As full disclosure, I'm a hobbyist at best with respect to troubleshooting cars and the above 2 pre-emptive measures are certainly not a scientific approach to the order of operations. I would also fully concede they are speculative at best, but sometimes you get lucky!

 

Hi Lewis,

Yes, I am talking about the +- swings in each column, but independent of each other. There is a very wide range on bank 1 that is nothing I've seen, but the range is a little more normal on bank 2.

Even though there is no way to align data from bank 1 to bank 2, I looked at RPM which I left on the sheet. Even though, here too, you cannot align anything, it appeared you tested a pretty steady rpm range from idle to about 1800 to 2500rpm. You should have been fueling in closed loop during all of these ranges. Even if you opened loop with a throttle burst where you would expect Lambda to be maybe as low as .7/.65, but that didn't show in the RPMs. In closed loop those columns should be looking like .8 to 1.2

I'm in some of the same space as Sam thinking that somewhere along a path of sensors on Bank 1 intake I would want to ensure everything checks out but I'll add a couple to Sam's list. Look at MAP too. I wouldn't rule out possible leaking or clogged fuel injectors especially on bank 1.

Finally, if the car is newer to you and you might not know history, suggest you try an easy throttle body alignment with VCDS. This is pretty intuitive following VCDS directions. It just might yield something.
http://wiki.ross-tech.com/wiki/index...Alignment_(TBA)

 

Sounds like the figures for bank 1 are not you would be expecting to see in that case. I was driving normally from 0 up to 30-40 mph with occasional speed increases to 50mph and slowing down to 30 mph.

I was just cruising along, no large throttle increases or bursts of acceleration.

I will take a look at everything Sam has suggested and include MAP in my next log as you have requested.

I've owned the car since March/April this year but barely managed 3 days of driving before things went wrong on bank 1. Here is my thread for that issue which turned out to be a failed boost control solenoid - https://www.6speedonline.com/forums/...ine-light.html

I don't recall whether or not this sputter/jolt when going on or off throttle existed prior to the above issue as I hadn't really driven that car a lot at that time.

I have already tried a throttle body alignment via VCDS which succeeds no problem but doesn't make any difference. The car has 12 new spark plugs and 12 new coils (fitted by myself purely as preventative maintenance)

 

Hi @BWings

Please bear with me here, I'm out trying to log both banks simultaneously however when going to address 31 engine other and selecting group 32 VCDS is only showing me one bank and not both banks as appears in the screenshot on the ross tech website.

Am I doing something wrong here? Also i couldn't find which group MAP was under.

Thanks,
Lewis

 

Hi Lewis:

I wanted to show you what a Lambda data log should look like. I found a data log that was done logging AFRs. Lambda is a much simpler way to log and review AFRs and what is supposed to be logged using the VCDS (not AFRs). I created a column on the sheet and converted the AFRs to Lambda.

In the data log I am posting, the log was created using a progressive and steady increase of RPM up to redline. The data is very easy to discern this way BC closed loop can be clearly seen and as loop opens and transitions into power enrichment, that too can be seen. I always DL this way.

I know you were doing something very specific by varying RPM from idle to about 1800/2000RPM, back and forth, and every time you let off the pedal (0% commanded throttle) which triggers overfueling then DFCO, it does tend to make an alphabet soup out of Lambda numbers, but not the spread. You were doing the right thing of course, but I am still not satisfied with what I saw even given the DFCO and overfuel. As an example, overfuel might produce a Lambda of 0.6, then immediately followed by DFCO which often shows up as Lambda of 1.5, maybe as high as 1.6. I can roughly follow that by reviewing the oscillating RPM dataon your sheets, but given the spread of #s in the two Lambda columns, Lambda just doesn't look like that.

On the Xcel I've created, I've highlighted closed loop (fuel trims maintaining stoichiometric {Lambda =1}) in green. The red area shows when loop opened and fueling commenced by a fixed table in PE.

Long story short, these are what should be seen for normal fueling.

 

Hi @BWings

I've had a look at your attachment and I feel what I need to do is try to log the data from my car in a similar fashion so It can be compared to your log.

Unfortunately as I said in my post above I just cannot seem to get VCDS to show me data from both banks of the engine simultaneously.

On the ross tech website they state to access address 31 'engine other' then data should be shown for banks 1/2 and 3/4.

I've attached a screenshot showing my screen when attempting this.

 

Lewis:

I cannot help with DL using VCDS. Never DLed with it. I use VCDS for lots of other diagnostics, but not DL on this platform.
Sorry I cannot be more help there, but if you can get to the point to just gather fuel trims and O2 sensor voltage outputs in graphs, that would make it a walk in the park. Doesn't even have to be the same DL. You can DL each function separately

Good Luck!

 

Lewis,

Let's do another take. I am looking back at the original data that you posted. Forget about the tables I've been sending, instead lets look at the original Lambdas (banks 1/2 and banks 3/4 from your first two files. These, in fact are not lambda data as purely as lambda is defined, instead these data are actually fuel trims. Lambda still works here, just what it is being called.

I will repost one sheet I created that just shows the RS/LS data columns we now want to look at like they are fuel trims.

So, interpreting them as fuel trims, the bank 1/2 (bank one, right side when sitting in the front seat looking out the windshield) is making some giant fueling corrections compared to the LS bank. It is what I originally said did not look normal and still doesn't. It looks like you have an AIR leak that occurs PAST the RS MAF. The MAF is located right below the air filter, so anything that could possibly be sucking in air on that side past the MAF is suspect. Let me suggest a few places:

1. This might or might not seem odd, but the oil dipstick must be completely sealed. It should have an O-Ring, it must be in good shape and the stick MUST be seated. On mine I actually zip tie the dipstick closed. Check that first.
2. Next, right below the air filter you'll find a snorkel that funnels the metered air to the turbo. It will look like this:



At the base where spring retainer is, there is a neoprene o-ring seal. It can dry out creating an unmetered air entry point.
3. Then as the air enters and leaves the intercooler, you have to check the condition and the connections of those hoses. The bottom one (under the belly tray) has a tendency to slip off, but not fully.
4. Finally you need to check the integrity of the intake manifold on the RS. You said that you changed plugs and coils, removing the intake in either of two ways could have caused an unmetered leak in numerous places.

If there is no unmetered air on the RS bank, then the last possibility is a bad O2 sensor 1, bank 1

See if any of this is making sense to you and reply back