Eliminating Knock - Tune Differences

[Hotwire]

Hi All,

 

Recently I have had some minor knock corrections (.5 to 2.0 deg) being constantly applied, particularly in the mid range. I have run numerous 98RON fuels and still get the same results, so i'm confident its not fuel issues. I have had the car inspected at a well known subaru workshop in my hometown, and they tell me everything else is 100% so I can only conclude its the factory tune on my car.

 

I have been looking at My tune compared to another MY04 Auto Rom I have, and the only differences I can see is in the requested torque and throttle position tables in the DBW settings. All other areas (timing, boost targets fuelling etc) are exactly the same between the two roms.

 

However, in the DBW settings, in a few areas around 25-30% accellerator there is less torque requested, and over 36% accellerator there is average 8-10% more requested torque.

 

The requested throttle opening data doesnt seem to quite match the above though - so i'm wondering if 1 map is slightly superior to the other. Would anyone in the know mind having a look through the comparison csv sheet I made and give me your comments? (wouldnt let me attach original excel file or the zip!)

 

Looking through the logs i have, the knock correction I get (FBKC & FLKC) are usually at around mid throttle, and alost never under WOT so I'm thinking this may be part of the problem. Also is there a way to tell which ROM would be the newer of the 2 released by subaru? as Im thinking they identified a problem and fixed it here....

 

Cheers

Lee

Tunes Comparison - MY04 Auto CSV.csv

[Hotwire]

Can anyone shed some light on this for me?

[NSFW]

I can't see how the DBW tables would affect knock at all. I'm guessing you've got some kind of mechanical difference going on, maybe a rattle causing false knock, maybe a bad injector, something like that.

[Hotwire]

OK Thanks,

I guess when I posted this I was thinking that the increased requested torque may be causing lean/knock conditions, but now when I think about it and look back at past logs, the AFR is generally around 11.14 (min) to 11.5 under WOT with a worst case of 13.5 under very mild loaded conditions. The odd thing is I get -2 FBKC every now and then under cruise in CL fuelling and the AFR at front O2 sensor is still 14.7.

 

I have had the car inspected for issues that may be causing false knock (turbo heat shields loose etc) and they came up with nothing so thinking something more internal.

 

Would this indicate that it may be 1 failing injector - i.e. get slight knock in 1 cylinder, but others are all ok, therefore AFR is on track?

[NSFW]

My car had some knock in cruise even with the stock tune, so I'm not convinced you actually have a problem. Can you post a data log where this happens? It would be interesting to see:

 

AFR

RPM

Total Ignition Timing

Load (preferably the 2-byte version)

Knock Sum

FLKC

FBCK

IAM

 

And take a Learning View snapshot before and after.

[Hotwire]

Here is another thread I posted:

http://legacygt.com/forums/showthread.php?t=117988

 

Following my first post i was thinking I had a bad batch of fuel as I haven't had that much knock again. There are a couple of logs there, however I dont have knocksum as an enabled field for my ECU.. I will post some more logs off my laptop tomorrow shoing most of what you listed above.

 

Cheers for the help, much appreciated

Lee

[mwiener2]

Recently I have had some minor knock corrections (.5 to 2.0 deg) being constantly applied, particularly in the mid range.

 

Negative .5 to 2.0, or positive .5 to 2.0?

[Hotwire]

Negative .5 to 2.0, or positive .5 to 2.0?

 

Sorry, that is -0.5 to -2.0.. mostly -1 or -2 applied for up to 1-2 seconds. I have attached a couple more logs:

 

Row 27 of romraiderlog_20090824_070428 shows FBKC then a whole lot of FLKC as well as som other random corrections

 

Row 1061 of romraiderlog_20090824_072058 shows ~4 sec of FBKC corrections,

 

Row 106 of romraiderlog_20090824_074855 shows corrections AFTER giving it bit of WOT

 

(note I had to trim the logs down as they exceeded file size limit)

 

Thanks for helping guys.

romraiderlog_20090824_074855.csv

romraiderlog_20090820_070428.csv

romraiderlog_20090824_072058.csv

[Hotwire]

Here are some learning view shots from during (7:20am) and after (8:01am) on the 24th for the logs above.

The only "before" I have is the 19th August which I have also attached.

[Hotwire]

NSFW: How do these logs compare to what you were seeing in your car? is it similar knock to what you experienced (and therefore could be sorted with tuning?)

 

Thanks!

[NSFW]

Those learning views are pretty clean, there's no sign that you knocked more than once in any cell. I'm not that worried about a single correction (-1.40 to -0.35) unless its consistently seen in the same cell. Correction of more than 1.40 indicates that the car was running with at least (X - 1.40) correction applied, and then knocked again, so those are the kind that I think need to be fixed any time they occur.

 

The first log shows some FBKC around 1.64 load and 2510 RPM, and the second log shows FBKC around 1.35 load and 2500 RPM again - seems like a good idea to remove a little bit of timing in that area.

 

The second log also shows a lot of timing being pulled via FLKC around 1.8 load and 2700 RPM, then again at 2.0 load and 2800 RPM, when 1.4 degrees were already pulled. So I'd pull about 1-1.5 degrees from that area. (It seems to unlearn that FLKC pretty quickly - that's something I haven't seen before, I'm pretty sure my car takes longer to reduce FLKC.) There's another FLKC at 1.91 and 3332, right after it removed all of the FLKC from the previous knock, so I'd consider removing 0.5 degree from that area.

 

The 2nd log also shows you hitting 1.5 load at 2590 RPM without knock, which makes me wonder if there's a lump in your timing surface at slightly higher load and RPM.

 

The 3rd log had a small blip of FBKC at 1.5 load and 2800 RPM, but you also go through the 2.0 / 2800 range twice without signs of knock, which is interesting. Since the knock isn't repeatable every time, the timing isn't terribly excessive... It's possible this is just noise, but I'd still pull 1 degree from that area to see if it makes a difference see if that does the trick.

 

Can you post your map? I'm curious if there are any odd lumps in the combined timing (base + advance).

 

If you're tuning yourself, consider saving the current map and leaving a note with it (in a text file or something) that says this is the ROM before you started pulling timing to try to fix knock. If you end up pulling 3+ degrees and the knock doesn't get any better, that would point to noise rather than knock.

[Hotwire]

Here is a copy of my MAP/ROM

http://www.filehosting.org/file/details/54952/MY04_A2ZJD02G_(my_car).hex

 

I have been looking at comparing an STi tune to my default tune, and was looking at adapting it to my own map. So yes, I am looking to have a go at mild tuning, so removing some timing is something i'm willing to try.

 

Let me know your thoughts on the map.

Cheers

Lee

[NSFW]

Any chance you could post that on a site that doesn't require me to give them a working email address?

[Hotwire]

Happy to - any recommendations? I would host it here but hex and zip are both not allowed :S

[Hotwire]

http://rapidshare.com/files/272014965/MY04_A2ZJD02G__my_car_.hex.html

That should be better

[Hotwire]

I have also loaded the STi ROM that I have that I was looking to use/adapt to help improve throttle response a bit.

 

http://rapidshare.com/files/272015857/ADM_06_LGT_STi_AT_-_A2WD010Y.hex

[NSFW]

I downloaded your ROM, but it's bedtime now. More later...

[mwiener2]

you can post it on the forum if you change the .hex to .pdf or some other allowable format. Just say you did so, that way people can change it back to .hex after they download it

[Hotwire]

NSFW have you had a chance to have a look at the map yet? Would really like to hear your feedback

 

Cheers

[NSFW]

NSFW have you had a chance to have a look at the map yet? Would really like to hear your feedback

 

Cheers

 

Oops, sorry to keep you waiting. I'll get to it tonight.

[Hotwire]

Thats cool - I appreciate you taking a look.

Thanks!

[NSFW]

The timing seems kind of lumpy to me... and there's a lump around 1.35 load and 2100 RPM, shown in the first picture below. I'd smooth that out.

 

The 2nd picture isn't where you were knocking, it's just something that I find weird. A lot of stock tunes are lumpy too, so it's not necessarily that bad, but I'm not convinced it's really deliberate either. It just seems strange that the engine would really behave so nonlinearly.

 

The suface already looks fairly smooth around 1.64 and 2500 where you were seeing knock, so I don't know what to think... and my car is running more timing than yours here, by a couple degrees. That might just be noise.

[Hotwire]

Thanks for that - what is the program you've used there to look at the timing? Is it freeware to download? Looks like the sort of program I need.

 

Cheers

Lee

[NSFW]

Oops, I thought I mentioned it already but I just noticed I haven't. It's something I wrote, and you can download (free) it from the RomRaider forums:

 

http://www.romraider.com/forum/topic4439.html

[Guest]

The timing seems kind of lumpy to me... and there's a lump around 1.35 load and 2100 RPM, shown in the first picture below. I'd smooth that out.

 

The 2nd picture isn't where you were knocking, it's just something that I find weird. A lot of stock tunes are lumpy too, so it's not necessarily that bad, but I'm not convinced it's really deliberate either. It just seems strange that the engine would really behave so nonlinearly.

 

The suface already looks fairly smooth around 1.64 and 2500 where you were seeing knock, so I don't know what to think... and my car is running more timing than yours here, by a couple degrees. That might just be noise.

 

What tool is that?