RomRaider

I've done a lot of research and been through the write up of Subaru's knock control strategy a few times. I'm trying to get a better understanding of FBKC that is related to the ECU guessing there might be knock rather than reacting to real knock conditions via the knock sensors.

I'm running on E85 currently so pretty much the only FBKC I've seen in logs is from being too rich and having timing that is too low at lower loads. When I see this FBKC in logs, I see it in increments of -1* because this is what I've set in the tune for the FBKC retard value.

Today, I logged a quick pull in 2nd gear and looking at the log, I see FBKC showing -1.4 right at peak load. It stays all the way until I let off at about 5600 rpm. Now I did hammer on it right after going around a corner and it spooled very quickly. I've read that changes in load can trigger the ECU to pull timing and I'm guessing this is what I'm seeing.

What seems confusing to me is that it shows itself as Feedback correction when really it isn't using feedback from the knock sensors. I would also like to tune this out so that no timing is pulled when the ECU assumes there could be knock but I've never seen a table defined for this kind of FBKC. I assume it exists but hasn't been discovered yet.

Can any others confirm my theory on this "preemptive feedback knock correction" that I seem to be getting? I thought about asking for someone to look into defining this in the ECU Defs sub forum but I want to make sure my understanding of this is correct first.

FBKC is defined, and it is instantaneous knock compensation (not learned). The ECU is hearing knock and pulls timing (typically -2° per event to a max of -8°) as you see in the log and the total correction decays over a defined time period, about 10 seconds. If you have knock then adjust the tune in the related RPM/load.

I understand FBKC exactly as you explained it above. If I have any that is tied to the knock sensor, I see the amount of timing pulled show up in steps of 1* because this is what I set in the tune.

This FBKC I'm talking about pulled 1.4*, not the 1* I have defined in the tune. I know I don't have any real knock in this area and the knock sum isn't incrementing and I'm running conservative timing/fueling on E85 which is why I think this must be some sort of preemptive type of timing reduction.

If I'm completely wrong, then can you help me understand why if in my tune I tell the ECU to pull timing per knock event at a step value of 1*, I would ever see it pull timing in steps of 1.4*? Again, when I've seen "real" knock, I see FBKC show up as -1, -2, -3 etc. When I see this "preemptive FBKC", I see it as -1.4, -2.8, etc. I don't understand why I would ever see it in different steps that what I defined in the tune. Furthermore, when the ECU starts reducing the FBKC for real knock, it steps back .5* at a time as I've defined in the tune. This other type of FBKC steps back .35* at a time. I know the stock tuning retards timing 1.4 per event and advances .35 at a time. I feel like there is some other logic being used when the ECU thinks there might be knock and this logic references 1.4/.35 for its values.

Thanks for your help.

Can you post the log?

I need to post my logs of this as well. FBKC on a few where changes in load triggered FBKC yet knock sum and knock count stayed at zero. Was a recent thread on it and I could induce FBKC with rapid throttle change or odd wgdc values

There are four knock counters for the new STI, one per cylinder. Which one does the RR logger read?

I have multiple logs from Forester, LGT, WRX and STi all with the same issues.

Knock sum is different to knock counter. I get neither parameters increasing and even knock ears get nothing.

Pull timing and add fuel does nothing

There is knock sum per cylinder in the new Roms - I managed to dig it out for my rom viewtopic.php?f=34&t=8506

Post a log, will be interesting to dig around that area of the rom - I've seen multiple instances of the 1.4 value in this region, so there might be something behind this.

Here's a couple logs. I had to dig around a while to find one where I saw some cruise knock. The step values in the cruise log are what I'm used to seeing when I was tuning out some rich knock. The 2nd gear knock is where I am seeing this other type of feedback knock that steps differently, at 1.4.

I'm looking at the logic around this area, and (at least in the Forester XT, likely in the STI's), there is a branch off of the routine, based on a certain switch, that leads to a value (of 1.4) being subtracted from Feedback Knock Correction, other than the defined feedback knock correction retard value. The red elipse at the top is the 'normal' retard value. See the green elipse.

I KNEW IT! Insert Happy Dance emoticon here!!
I been chasing my tail on this for months and it was the only logically assumption! I only had looked at logs and pre-emptive was the only logically course.

I pull -1.51 on my FLKC and add back 0.25 so the 0.01 shows up in LV so I know how many times it has knocked and FBKC is -2.11 with 0.35 additive so having -1.4063 in FBKC was confusing the sh*t out of me

Thanks for confirming my theory!

Aye - I'm not going crazy.

Even with FBKC correction set to 1, I still get 1.4 pulled sometimes.

Also, I notice sometimes it doesn't do a full countdown to zero in normal increments. Sometimes it only pulls for ~15 cycles and then goes all the way straight to zero.

I have been stressing about this for some time now!

At long last, I feel relieved. No more chasing may own tail... thanks chaps.

Great find td-d! I'm glad as well to see there is a good explanation for this odd behavior.

Now - how much would it take to understand this sub logic to know what exactly triggers this?

Could the value being pulled be defined like the normal FBKC retard value so that it could reduced or eliminated?

Sure - very easy to define. Which roms would you guys want?



Well - that could take a little doing, as it's a mixture of following the logic and logging. First clue would be to log the switch that determines whether the logic continues to the 'normal' retard, or branches to this one.