LS1 using secondary spark table. I find no reason for it.

[dermotw]

That log is short... You have driven also for a reasonable period of time (the KL will take many minutes to climb...)?

[Tapi]

Yes, that's second log of the drive. First we drove 20 minutes at different speeds. After that I saved also 2 logs in which the knock was not present.

This car was on a dyno few weeks ago and there was no audible knock (dyno owner used "stethoscope"), neither did knock sensors react. Only thing differing from previous builds was that this didn't go much above 400 hp and few degree timing changes or small AFR changes didn't do a thing to the max power. With these mods we expected about 20-30 hp more.

[dermotw]

In that last .hpt file, the hi and low tables looked the same? I assume you drove with them different, low table -5 or similar? I don't believe the learn will work unless they are different.

[Tapi]

In that last .hpt file, the hi and low tables looked the same?

You're right, my mistake (again). I played with many tunes and the final one that slipped in was with identical spark tables. I have to ad 2 degrees to high octane table and try again. I will write here the results.

[Tapi]

Now I got back to this and I made a 2 degree difference in WOT between low octane spark and high octane spark.

I attach here the results. Still no luck. First we did a WOT run in which we got some knock on third and fourth gear (you can see on the log that there's KR and burst Knock at the same time) but even at that situation the Knock Learn Factor stays 0.00! So this should read the high octane table but it does not. We couldn't get the knock present after that.

I attach here also a long run in which I concentrated mostly in looking the bank-2 O2-sensor which shows very different readings than Bank-1. Between these logs the only difference in tune is that before the long log I removed IAT spark retard from areas we got the car into.

[dermotw]

Burst KR is predicted knock, based on how hard you hit the throttle. It's not 'real' although you get actual KR as well. If you haven't changed the burst KR settings, then it will always do that if you hit the throttle in those circumstances.

As I mentioned, that tune file has the learn level for MAP kpa at 40, i.e. if you are not consistently over 40kpa (consistently means for minutes at a time) then it won't learn and shift the KL off 0.

In your long log, I noticed only ONE period where it was at over 40kpa for more than 20 seconds, thats not enough imho. You would need to be cruising at a constant 120kph (or similar)... or lower the kpa to somewhere that you het all the time...

It will take many minutes to learn to 1 (and also many minutes to revert to 0, each burst of KR would drop it maybe .01 or .02, so if it is up at 1, only sustained, repetitive KR will drag it down to 0. But the same applies the other way...

BTW your o2 and fuel trim readings are quite odd, even though it seems like all is working?? Sometimes the o2's are cycling, sometimes not.. and the fuel trims are miles off a lot of the time (10% is reasonable...)

[Tapi]

I noticed the same thing on O2 signals. I think it's due to long tube headers which let the O2 sensors cool too much on light throttle. The problem seems to occur only on idle and on cruise speeds with almost to none pedaling. This won't explain the bank2 sensor behaving so dramatically though. This situation leads to being only possible to tune with WB and even after that you get the tune messed up when bank2 O2 sensor increases the average fuel adding very much. In open loop the car is not jerky at all and the AFR stays as it should (I commanded the car to OL in the last part of the long log file to prove that).

But to the KR and low octane spark table: When/if Knock Learn Factor stays 0.00, shouldn't it use the high octane spark table? It does not.

[dermotw]

1 = HIGH, 0 = LOW, and anywhere between = a % between, e.g. KL is at 0.5, and at 3000rpm high table says 30 and low says 20, so you get 25, halfway between...

Here is a snapshot of my own car KL recovering from 0, after I had been testing in OL, note following;

Whole log is over 1 hour in the picture, it takes 30 minutes for KL to go from 0 to 1, and almost all the time I am over 100kph. It took about 4 minutes from start up before it moved at all. KL is the white trace at bottom... You will see small places where it goes level or down, thats where learning stopped or there was KR...

Capture.JPG

If you have longtubes (or indeed other mods) then usually you have to alter various 02 parameters to get them to switch better at low speeds, and they will never switch as well as a standard car. If you have a cam its worse. Hopefully the sensors are in the collectors which is about as close as you can get.

[Tapi]

1 = HIGH, 0 = LOW,

Damn!!! I didn't even bother to confirm my opposite belief! If this is as you wrote, I will make a new tune with very small step to Knock Learn and we'll start cruising.

The O2 sensors are in the LT headers collector and the headers are wrapped in insulation wrap. I already made some adjusting to O2 parameters but I didn't put much time on that.

[Tapi]

After thinking too much about this KL-thing, I need some education of you more experienced ones:

Is the Knock Learn "memory" just one number for all the load chart or is it specific for each cell? If it's specific, then it could be very hard to get the car tuned for right spark advance if you do not know which KL value would be in each cell.

[dermotw]

One number, and when its not at 1.00, you are <<always>> using a "blend" of the HI and LO tables at all speeds and loads except if the idle spark tables were triggered (might sound silly, but on some cars, for example my 99 TA as standard, the idle table is switched to at about 60kph! - No idea why... anyway I changed that, lol.). When the KL is 0, you are 100% using the LO table.

BTW a very typical value for the LO table is "5% less" i.e. you take the whole hi table, copy to lo, and then multiply it by .95....

"Normally" the car should be at 1 for KL almost all the time. A burst of KR will shift it down by a small amount, e.g. to .97 or .95, then it will (hopefully) learn back to 1 quite fast. It will only go anywhere near 0 if you had heavy, constant, KR...

Actually you can see in that snapshot how it climbs to 1 about halfway through, then it hardly deviates from that, just small drops if there was a burst if KR (this log was from before my car was setup BTW, nowadays it is on 1 all the time, basically)..

[Tapi]

I wish I had found the problem from KL but not: I made a tune which allowed all above 20 KPa map values to alter Knock Learn Factor but first 0,5 h log didn't change it from 0,00 to anything. Neither did the second 0,5 hr log. That second log and the tune attached.

[Ben C]

I am on phone so can't look at log or tune, but Based on the commanded timing is it definitely running on the low octane table?

[Tapi]

Yes it is. On WOT high octane table commands 28 degrees advance and low octane commands 26 degrees. The log shows 26 degrees.

[dermotw]

Hmm, well, that should be all you need!

The only other odd thing (to me anyway) is that idle spark table switching value that I mentioned to you before;
Capture.JPG

I suggest you set to e.g. 20kph not 402, "as a test". A '99 ECU like mine has it set to 60kph standard (the other values are same).

If it is holding you in idle tables then of course the KL will not operate (and your idle tables are at 26 deg like the mains...)

EDIT: In fact I'd set the MIN TPS and its hyst. to some value as well, e.g. 1% TBS and .4 hyst. Again, as a test.

[Tapi]

That max required speed for disabling main spark tables I understand. The lower the value, the earlier idle spark will be ignored. But that throttle % is more a question mark. Is that value a low limit or a high limit for the throttle? This one idles at 5.5-6% throttle when warm but I suspect it will alter a lot depending on various things.

It says: "This value is the throttle position required to disable the main spark tables". To my understanding a low value here would make the pcm use idle spark but if this is high, it will never use it.

[dermotw]

Below "max speed" value = use idle tables.
Below "min TPS" value = use idle tables.

The explanation for this is very badly written, imho, it is almost deliberately confusing. The thing which is also not said at all is;

Is it "below min speed AND below min TPS" or "below min speed OR below min TPS"....?

Another point is that sometimes value 0 is used as a special case, meaning "disabled". So in an ETC car, the throttle value could be set to 0 to disable the table. I have not seen it documented though... If that WAS true, then your settings are saying: "dont use TPS, below 400kph use the idle table" !.

Other point, of course the throttle CAN'T go below 0. So if in use, that value should surely be above 0...? In case of an ETC car then its probably not actual throttle but the pedal input that it reads (if it is used at all...). If it was actual throttle it'd have to be e.g. 7%!

Anyway, you can test it: just set the speed to eg 12kph and throttle to 1% or similar, and see what happens,

[Pekka_Perkeles]

Please log cylinder airmass as well.

Now it's missing from the log.

[Chris@HPTuners]

Try setting your P0102/3 DTC's to 3 - No Error, then clear DTCs in the scanner and see if that fixes it. Either of these codes will disable knock learn, also any AIR Pump codes P410, P1415, P1416.

[dermotw]

THATS interesting... I didn't know the DTC could get 'stuck'...