Jump to content


Photo
- - - - -

Knock logging - Accessport


24 replies to this topic

#1 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 02:07 AM

Joe has updated the Accessport BETA firmware to include NCON_KNOCK which I've been testing for a few weeks.

Go get it, post your results here, I'll kick off with some data shortly with an explanation.

#2 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 02:17 AM

My drive to work today attached.

Explanation to follow...

Attached Thumbnails

  • drivetowork.PNG


#3 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 03:49 AM

Right y axis shows RPM (in magenta), rest are on the left axis. I have filtered only to throttle >85%, you will see the runs are spliced together, so when the seconds take a step up it is a different run. Yellow shows the load maxxing at 25.5 through the midrange. Blue is boost pressure in kPa, you can see the small spikes on gearchanges.

The negative purple values that shoot from 0 to -307 are episodes of knock. It jumps down 307 at a time and recovers gradually in steps of 13. 256 seems to be about 1 degree of retard, so it only takes a few quick steps of 13 from -307 to get to -255 and clear the degree of retard. If you get subsequent knocks it will again subtract 307, sometimes a few in quick succession.

To get the highest average timing I propose to tune to have the occasional drop to -307, drops to -614 will take a while to recover from, but may be recoverable from by just voluntarily taking 1 degree out your timing map where the ECU will take 2. It is a general rule that knock takes more retard to clear than it would to prevent.

I will prepare another graph from another car using the same map that I have here, but has probably about 1 RON lower octane but resonated decat midpipe.

Questions?

Edited by thistle, 18 September 2009 - 05:10 AM.


#4 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 04:06 AM

Here is another car with midpipe, 1 RON lower, driven hard on quieter roads. This time only the theoretical pulse width is on the right hand y axis. Same ECU map, probably about 1 RON lower, it shows multiple hits of -307 added onto each other.

All full throttle, again the graphing is splicing together lots of full throttle bits.

Note that when an internal value in the ECU that is much like this one gets to beyond about -2000 you're getting near to a switch to limp mode/low octane/dropping boost. I've not experienced this personally.

Only other thing to say really is that there is a memory effect. The knock value goes back to zero when your load drops below the knock control threshold, when you next go into the knock controlled area it doesn't return to exactly what it was before, but unless it has been a long time it doesn't seem to be zero either.

Summary: Sudden drops of 307 whilst under load appear to be new knock events. Recovery is in steps of 13. Just a bit more negative than -2000 and the ECU will probably freak out. Best results will be like on most other ECUs by cooperating with the stock knock control to get the highest average timing and safest/best performance. Stock ECU on 99 RON on stock car usually runs 0 unless severely provoked. Debate whether we aim for 0 or the odd -307 is open...

Attached Thumbnails

  • Samdrivenhard.PNG

Edited by thistle, 18 September 2009 - 05:10 AM.


#5

    GTRLIFE Newbie

  • Validating
  • Pip
  • 44 posts

Posted 18 September 2009 - 04:08 AM

The scarey thing Thistle is that i am actually starting to understand "some" of the babble you come out with!

Keep up the good work mate!

Will need to arrange a morning with you again soon.

Regards

Scott

#6 David_Yu

David_Yu

    R34 Member

  • Media
  • PipPipPipPip
  • 1403 posts
  • Location:Surrey, UK

Posted 18 September 2009 - 04:53 AM

Two obvious questions from a layman:

1) have the maps Cobb and GTC sent out so far been "safe" from a knock POV?
2) what parameters should we now datalog for a custom tune, as is it true the AP can only datalog 11 or 12 effectively simultaneously? i.e. which parameter Ben previously said I should add to the log list should I now delete, if any?
Posted ImageShare your ownership experience on www.auto-journals.com ! Easy to use, great to read: Real Life with Cars

#7 Martin Donnon

Martin Donnon

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2844 posts
  • Location:Australia

Posted 18 September 2009 - 05:00 AM

Excellent work - AGAIN - Thistle :OMG:

#8 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 05:16 AM

Cobb staged maps have been extensively tested to not knock in any excessive way to harm the engine, any excessive timing is quickly removed by the ECU. Most/all modern turbocharged gasoline engines rely on ECU knock control, this is where you usually get best performance/emissions/economy. There may be room for improvement with custom mapping.

Knock is of course a question of degrees. Mild knock of the sort the ECU hears and corrects before it gets out of hand will not do harm. Severe, sustained knock without knock control will destroy an engine.

The ignition maps used here are stock Cobb stage 2 97 RON, and I think they are good for my 99 RON use at higher boost levels on much lower spec (mine is a STANDARD car), I'm proposing to leave mine as it is shown unless I see any more, but I want to debate that here with other tuners. On the second car it could probably do with a degree or two off for its slightly lower octane. Cars will vary as they always do. Driving styles vary, as do fuels, ambients.

So there is no need to panic with a bit of knock retard, but we may run smoother and higher average performance by using the new info/logging for custom mapping. I had been told that these engines run MBT, I do not believe that to be the case based on what I'm seeing. It is probable that actually the knock control is so good and smooth that tuners adding more timing and not seeing a rise in power will find they have no more timing in their logs and the clever ECU is taking it out behind their back.

Edited by thistle, 18 September 2009 - 05:25 AM.


#9 Martin Donnon

Martin Donnon

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2844 posts
  • Location:Australia

Posted 18 September 2009 - 05:23 AM

I work with many modern ECUs Thistle, and as you say there is nothing to fear from indicated ECU knock as long as they are performing the correction, most of them seem to have a fair degree of it from the factory - and with your background in the scene I am sure you would agree with this :OMG:

#10 sam elassar

sam elassar
  • Guests

Posted 18 September 2009 - 05:30 AM

weather we need to run close to the odd knock and let the ecu deal with it isok as long as it does not happen all the time. my car was consistently provoking -307 most times it went around 4800 rpm, so i will need to take couple of degrees out around that area and do couple of logs.

my theory has been usually is to map the car 1 degree away from the knock thresh hold, that is what i have always done, some tuners will take advantage of this safety margin and make more power by advancing the same map by a degree or two. i just like to err on the side of caution. the R35 engine is far too new to push to destruction !

if i was not busy this weekend i would like experiement with different ignition curves and different fuel octane i have access to some race fuel as well and see how far will the ecu advance the timing when not getting any signal from the knock sensors. i will probably do that next week.

the other thing i need to find a good place to fit my det cans and see if the ecu is actually hearing det or was it my driving style that was provoking the engine noise ;)

saying that the boost improvement at the top end is very good :OMG:

Edited by sam elassar, 18 September 2009 - 05:33 AM.


#11 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 05:37 AM

The knock sensor(s) are checked each cylinder event I believe, so a single knock may be retarded very quickly without us even hearing it in det cans. On Evos I found this to be the case, you could not hear anything with the knock control active and thought it might be false knock. Turn the knock control off and find it was actually saving your engine's ass.

Indeed Sam, I agree the key to validating this further on the GTR is to see whether it is timing and octane responsive. I expect once a few people log knock and take out a few degrees we'll soon know. I've just not had time to play with all the boost stuff going on.

Edited by thistle, 18 September 2009 - 05:46 AM.


#12 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 18 September 2009 - 06:06 AM

Sorry David, my present list is:

Ignition timing
RPM
Theoretical Pulse Width
Boost
Wastegate duty
Throttle position
Mass airflow
Injector pulse width
Air fuel ratio
Vehicle speed
Knock

Room for one other, this might be a trim or Mass airflow on the other bank, or air fuel ratio on the other bank depending on your task/need.

#13 JoeGr

JoeGr

    COBB Tuning Firmware Engineer (Hacker)

  • Members
  • PipPipPipPip
  • 1615 posts
  • Location:Salt Lake City

Posted 18 September 2009 - 08:52 AM

If some of you can't find NCON_KNOCK that is because it should be called "Knock Sums". NCON_KNOCK is a variable I use so I can translate the name into different languages.

Joe
COBB Tuning Firmware Engineer (Hacker)

MBA Class of 2012

#14 thistle

thistle

    R35 Member

  • Members
  • PipPipPipPipPip
  • 2067 posts
  • Location:UK

Posted 06 October 2009 - 12:42 PM

Only Brits and Aussies interested in knock? ;)

Where are the US results?

#15 Kunani

Kunani

    R34 Member

  • Members
  • PipPipPipPip
  • 1335 posts
  • Location:Las Vegas, NV

Posted 06 October 2009 - 12:50 PM

Only Brits and Aussies interested in knock? ;)

Where are the US results?


If I can get some updated maps and software I'll put my car on a dyno and do some comparisons. I can also log on the upcoming GTR track day on the 16th if I get something on time.

I'm running a custom 100 octane tune now.

#16 'Q

'Q

    Spec Mi

  • Members
  • PipPipPipPipPipPip
  • 4308 posts
  • Location:Kingdom of Bahrain

Posted 06 October 2009 - 01:00 PM

Please post the link for the download.

One thing, my laptop that runs windows just died on me couple of days ago. Can this be done with a Mac?

Facebook Admin/Content Creator:
Facebook.com/MEGTROC

Formerly known as member GT3240x2


#17 Ben Linney

Ben Linney

    SpecV Member

  • Newbies
  • PipPipPipPipPipPip
  • 3233 posts
  • Location:on your PC

Posted 06 October 2009 - 01:22 PM

If I can get some updated maps and software I'll put my car on a dyno and do some comparisons. I can also log on the upcoming GTR track day on the 16th if I get something on time.

I'm running a custom 100 octane tune now.



for interest why not just go down the road now and log current map knock ? 3rd-4th redline :doh: dyno not needed

#18 JoeGr

JoeGr

    COBB Tuning Firmware Engineer (Hacker)

  • Members
  • PipPipPipPip
  • 1615 posts
  • Location:Salt Lake City

Posted 06 October 2009 - 01:24 PM

Please post the link for the download.

One thing, my laptop that runs windows just died on me couple of days ago. Can this be done with a Mac?



You'll need to update your AccessPORT with a PC. We are still working on a updater for the Mac.

Joe
COBB Tuning Firmware Engineer (Hacker)

MBA Class of 2012

#19 tomgtr

tomgtr

    R32 Member

  • Members
  • PipPip
  • 58 posts

Posted 06 October 2009 - 01:27 PM

Please post the link for the download.

One thing, my laptop that runs windows just died on me couple of days ago. Can this be done with a Mac?

Same issue for me, running macs at home. Apreciate though it's not worth redoing Cobb software for the few running Macs.
1 more month to wait!

#20 'Q

'Q

    Spec Mi

  • Members
  • PipPipPipPipPipPip
  • 4308 posts
  • Location:Kingdom of Bahrain

Posted 06 October 2009 - 01:29 PM

You'll need to update your AccessPORT with a PC. We are still working on a updater for the Mac.

Joe


Any idea when?

Facebook Admin/Content Creator:
Facebook.com/MEGTROC

Formerly known as member GT3240x2




Reply to this topic



  

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users