Geezz you CA guys are doing it tough

yeah the BRZ manual specifically says that the car is designed for 94 but that it can run as low as 91, though performance will suffer. :(

I reflashed to v1.55b of the Stage 1 tune this morning, and it started out at IAM = 1.00 as expected. Stayed at 1 for a while as I was taking it easy, but as soon as I started climbing a hill at moderate throttle it went to 0.78. It's holding there for now so I will give it more miles and time to see where it ends up. Then I'll do some more WOT logging runs, then switch to some 94 octane and see if that helps (seems likely to do the trick).

Do a few logs and see what RPM/Load spots are causing you trouble. I did that with the 1.42 maps and it only took a few days to nail it all down. I pulled a small amount of advance in the areas that knocked.

I flashed 1.55 today and haven't seen any IAM drops yet, even doing steady-state uphill, though I assume I'll find some trouble areas and will adjust as needed.

Just a suggestion: Don't get too worried about Adv Mult moving around on 91oct fuel. It will. That is what it is supposed to do. And you will see more of it as weather conditions get worse (summer). It does not mean that something is wrong but rather that the ECU is doing what it is supposed to do when presented with lower quality fuel.

I have considered making 91oct specific maps but it would be no different than scaling down the values in the knock correction table to what you would get if Adv Mult dropped in the current maps. The downside is that you wouldn't get the nice bump in power when conditions get favorable and Adv Mult climbs to max.

Cheers
shiv

when our IAM drops to .4 and .5...that's something to be concerned about IMO.

Not really. It's going to happen from time to time on CA 91oct. Especially if you drive in the range where the ECU is very sensitive to knock correction (2500-3000rpm) at 3/4 throttle.

yeah, just saw your other post. I do drive in that range when I'm coming up hills on my drive home, and that's where I've had the biggest issues.

I'd rather pull timing in one spot where necessary than globally with the iam.

Google "Vishnu reset"

It's a Subaru ecu adaptation technique we coined back in 2002 for that WRX. It applies to the FA20 too. That rpm zone (during steady state partial load conditions) is used as a diagnostic test by the ecu. If you retard ignition enough to never get any knock correction activity there, you end up castrating torque in that area. I'm my experience, it's best to let the ecu do it's thing and give it what it expects. It's just as quick to raise Adv mult in that area as it is to drop it. After a few cycles up and down, it will stabilize. Definitely not something you can evaluate right away after a flash or ecu reset.

Darn I'm giving away all my secrets jk :)

Good info

Well I finally got around to looking at your logs. I'll point out now that to get the data below I had to remove a lot of intermediate data, leaving only the parts that were 100% WOT pulls over a large range. I've broken it down into some easy steps so that anyone should be able to understand what's going on and the thought process. I'm concentrating only on knock learning and ignition advance tables here, relying on a stable IAM as knock can still happen even if your IAM is at 1 and not affect the IAM or when it's not able to be read.


1st of all you want to reference your ignition advance table, the one below is stock as per your log:
https://dl.dropboxusercontent.com/u/...sClub/KCL8.jpg


Next I checked the logged loads for the pulls:
https://dl.dropboxusercontent.com/u/...sClub/KCL9.jpg
If you look at these, you'll see that the load can vary at any given RPM as much as 0.1. This is worth remembering when looking at the resultant KC learnt values. You'll see from the advance table above that in some RPM ranges that the values are copied across multiple ranges, in others a 0.1 change in load will take affect more. I'll note below where this may have to be checked.


So, I've broken the following down to consecutive runs. 1st pull:
https://dl.dropboxusercontent.com/u/...sClub/KCL1.jpg


2nd pull compared to 1st:
https://dl.dropboxusercontent.com/u/...sClub/KCL2.jpg
A jump in timing at the start of the map and at the end - this is good.


3rd pull compared to 2nd:
https://dl.dropboxusercontent.com/u/...sClub/KCL3.jpg
This pull starts later but timing has dropped in a few areas, notably around 4k and 6k. This could be knock or due to being in a different load area. Possibly run on a non-flat or different gradient road? This highlights the importance of using a flat or same road for logs.


4th pull, compared to 2 and 3:
https://dl.dropboxusercontent.com/u/...sClub/KCL4.jpg
Timing has bumped up from run 2, suggesting an increase in IAM, the reduced timing from previous pull doesn't seem to be taken into account, which would suggest that it wasn't knock that caused it. The next pull will confirm/refute that idea. Lets have a look.


5th pull, compared to 4:
https://dl.dropboxusercontent.com/u/...sClub/KCL5.jpg
More timing added at the beginning and end, bump in IAM. No reduction in timing again so all looking good. The fact that no timing is added mid range is worth checking the advance table.


6th compared to 5:
https://dl.dropboxusercontent.com/u/...sClub/KCL6.jpg
Added timing again except 4.5-5k. Maybe we are now seeing knock in that area but IAM is still increasing. That was the last WOT pull provided. The next pull would be an indicator as to how the ECU is responding further. Hopefully the above has given you an idea of how to compare runs to see the indicators. Finally I've included the overall timing of those last 2 runs. Remember this is base ignition timing table + (advance table x IAM). This will illustrate the effect the KC learn values have on the overall timing:
https://dl.dropboxusercontent.com/u/...sClub/KCL7.jpg

i've found that 3-4.5k and 7k+ are the most common areas to see flkc applied during full throttle pulls. i've also found that having even a little knock correction in fine learning (-1 flkc) will absolutely murder your power curve. -1deg of fine correction was costing us 15 lbft and ~10hp on the dyno. it seems to be very important that the tune has absolutely no flkc, or you're probably just losing power.

i'm often able to eliminate the FLKC by removing 0.3 deg where it is occurring, thus netting a 0.7deg gain in timing and gaining power, without losing consistency. making multiple pulls back to back to back and ensuring there is no correction applied is important to ensure that all that power you just gained on the dyno will stay there in real life.

I really appreciate you taking the time to go through the runs and show us noobs a few things. And yes some of the runs were done on uneven slopes, not my ideal testing grounds but it was closer by and convenient.

Just to give an update, I went and reflashed to stage 1 v1.55b, and IAM only dropped as low as 0.78 this time. It started to climb slowly back up which was encouraging. Could be that the tank 2 fillups ago was a particularly lousy batch of 91, and this last one was better. Or that 1.55b had adjustments to the advance tables that helped. However, I was about to be empty so I decided to tank up at a station that offers 94 octane. Shouldn't be any surprise to anyone reading this thread, IAM very quickly rose back to 1.0 shortly after. As it's rather expensive stuff here, I will probably go back to 91 again on the next tank, and from there we'll see if the IAM dips down again; and if so I will be doing some more logging and perhaps getting my hands a little dirty in the advance tables. Or taking the plunge into E85. :D

Maybe run a mix of the 94 and 91

Sent from my SGH-T999 using Tapatalk