Enginuity question

[mickeyd2005]

Yes. You have to dump the ram using ecuexplorer and then read each each cell.

 

merchgod said that the location of FLKC can be easily determined by looking at the rom logic. However, enginuity doesn't have the functionality to display tables. He said that he is writing something in VB to display it.

 

There's actually a lot more stuff that the disassemblers have done than is publicly known. Modified roms, etc... but a lot of it hasn't been implemented primarily because enginuity needs java programming support.

 

I don't know java. I write stuff in VB.

[itsme]

I believe that that -50 knock count was from a different kind of ECU, maybe the 16 bit, or maybe some piggyback system.

 

You cannot even reasonably pull 50 degrees of timing, think about it. The 2 degrees of pulled timing we talk about is a literally 2 degrees, not an arbitrary knock count.

http://www.xpttuning.com/osecuroms/viewtopic.php?t=491 what do you make of this?

[mikeyan]

It is true that +KC is good, kinda.....

 

When I first started logging, I thought, "I have positive KC numbers, I guess all is well........................WRONG!"

 

If you have access to your tune, you will know how it is set up. Realistically you should always be running maximum + KC values that your tune has. This is assuming that the ECU has not been reset in a while. (if the ECU has been reset recently, depending on tune, it may take a few WOT runs for the ECU to learn its way to full power)

 

Mike is right, if you see a sharp drop in timing, you know that something is wrong.

 

Our LGT ECU is set-up to pull 2 degrees of timing at a minimum if it hears what it perceives as a knock event.

 

If you log total timing and get a knock event you will see a sudden drop in timing of either 2, 4, 6, etc degrees depending on severity.

 

 

 

Sooooooooooo, you could be pulling timing very often due to knock, and still always see + KC numbers.

 

Thanks for the additional info

 

<---- not a tuner

[Deer Killer]

I don't know java. I write stuff in VB.

lol, which one is worse. Python FTW

[mikeyan]

Knock count and knock correction are not the same thing.

 

Mitsus use a knock count. Subarus use knock correction.

 

Knock correction is in terms of degrees of timing. It can't be -50 or your car wouldn't run.

 

KC = IAM x DA + Fine Learning Knock Correction + Feedback Correction

 

FLKC and FC are always 0 or negative.

 

If you look at your dynamic advance table (also referred to as Timing Advance in enginuity), there are some areas where DA is 0. In these areas if you have any FLKC or FC then your KC will appear as negative.

 

Positive KC by itself doesn't mean much. It just means your tuner made your DA table > 0 at that location. Also, not all tuners use a flat DA table (constant value). You should rely upon IAM, FLKC, and FC to determine if there is knock.

 

Keep in mind that KC is purely for the tuner or technician. The ecu doesn't need KC. It uses IAM, FLKC, and FC. I believe that's why Subaru removed KC from the 2007 ecus.

 

merchgod wrote a complete writeup on subaru's knock control strategy on enginuity.org. Unfortunately, it's down. I've read Cobb's ST guide and it's a very similar explanation. However, there are a few additional details on enginuity.org.

 

How soon does FLKC/FC values in the negative change IAM? or is that adjustable via the tune?

 

 

Mike

[Deer Killer]

http://www.xpttuning.com/osecuroms/viewtopic.php?t=491 what do you make of this?

Sounds like he's reading garbage data. You should only believe the logger after you've confirmed it's actually working.

[mickeyd2005]

http://www.xpttuning.com/osecuroms/viewtopic.php?t=491 what do you make of this?

 

What are you referring to? That's a very normal looking log.

 

LBGT should have been more specific. Feedback correction at knock is -2 degrees. It then tapers as time passes. The -2 degrees is set in your rom. You can change it, but I don't recommend it.

 

IIRC, fine learning knock correction is in incremental units of -0.35 degrees. You can change this too, but I don't recommend it.

 

itsme, try datalogging FC and FLKC. You'll see what I am talking about. I have a excel spreadsheet that I use to tabulate the knock in partial throttle locations. It also summarizes VE (volumetric efficiency) so a few guys on enginuity.org found it useful for partial throttle tuning.

[mickeyd2005]

I write stuff in python (on unix systems) too.

 

VB is free with excel and integrates well with Airboy's spreadsheet.

[mickeyd2005]

How soon does FLKC/FC values in the negative change IAM? or is that adjustable via the tune?

 

 

Mike

 

The logic is buried in the rom. It's not that simple. Basically, when certain conditions are met where the ecu is confident that the signal it is receiving from knock sensor is very accurate then it updates FLKC and/or IAM.

 

If the car hits knock in a region or under conditions that it believes is REAL knock then it may choose not to use FC at all. It may start decreasing FLKC or IAM right away.

 

Also, this logic is different in different cars. pwrx (on enginuity.org, he has a 2002 wrx) had his IAM dropping on the STOCK ROM! We tabulated his partial throttle datalog and found that it was due to a very specific region in his timing map. I can't remember exactly where... but I think it was 2200 rpm at 0.95 g/s load. By reducing his base timing by 2 degrees in this very localized region of his timing map, we prevented his IAM from dropping. Remember that if your IAM drops, timing is pulled over the entire map. It's always better to find where the knock is occuring and pull timing locally rather than allow IAM or FLKC to take over. IAM and FLKC affects large swathes of the map.

 

In enginuity, the only thing that the tuner can adjust is the rpm range and load where IAM and FLKC can be adjusted.

[Deer Killer]

The logic is buried in the rom. It's not that simple. Basically, when certain conditions are met where the ecu is confident that the signal it is receiving from knock sensor is very accurate then it updates FLKC and/or IAM.

 

If the car hits knock in a region or under conditions that it believes is REAL knock then it may choose not to use FC at all. It may start decreasing FLKC or IAM right away.

 

Also, this logic is different in different cars. pwrx (on enginuity.org, he has a 2002 wrx) had his IAM dropping on the STOCK ROM! We tabulated his partial throttle datalog and found that it was due to a very specific region in his timing map. I can't remember exactly where... but I think it was 2200 rpm at 0.95 g/s load. By reducing his base timing by 2 degrees in this very localized region of his timing map, we prevented his IAM from dropping. Remember that if your IAM drops, timing is pulled over the entire map. It's always better to find where the knock is occuring and pull timing locally rather than allow IAM or FLKC to take over.

Or is it? If knock should happen the best place is under light load, because it matters less. IMHO maps should be tuned to be more knock-prone under light throttle/loads. If because of physical tuning the lower load column are less prone to knock and/or are tuned that way, and then the high load column are tuned to be on the "edge" for 1 IAM, then engine and fuel conditions that cause knock are only found at high load.. hmmmmm

[Infamous1]

Subscribed

 

-Shamar

[mickeyd2005]

On all the LGTs that I am familiar with, there is some FLKC under light load.

 

None of them drop IAM under normal driving conditions. pwrx's 2002 wrx is the only car that I saw where IAM dropped under normal driving condition.

 

Also, relying upon a single partial throttle location to drop IAM in case of emergencies is not that reliable. There may be times when the car is never driven in that "critical" partial throttle location. In pwrx's case, he RARELY drove at 2200 rpm and 0.95 g/sec. That's a pretty high load for that rpm on a 2L engine. So, his IAM would drop intermittently and for no apparent reason. It was not until he logged ALL his driving that we localized it to one location.

[itsme]

So what does a knock look like on a datalog?

[mickeyd2005]

So what does a knock look like on a datalog?

 

If you log feedback correction, it will be when FC jumps from 0 to -2 or from -2 to -4.

 

Also, if your previous FLKC was 0, and then your FLKC drops then that's knock too. This is more difficult because you have to keep track where you are at. Best thing to do is reset your ecu and then log 100% of your driving.

[bugblatterbeast]

He is right.

 

If you look at the table "Primary Open Loop Fuel Map Switch" you will see that at a certain IAM the ECU will switch to that map. Default should be 0.35.

 

Assuming the IAM is above that point, which is hopefully always, the ECU will use the "Primary Open Loop Fueling" table.

 

I believe there is also a switch over to the high det map if the coolant temp exceeds 212f..... don't ask......

[GTWILLY]

Wow this conversation took off over the weekend.... I am still wondering about one of the tunes i am analyzing. Is there any reason or sense to having the (fail safe) fuel map not primary map have leaner values? And I mean much leaner... in the 11. range vs. 10.51

[LittleBlueGT]

I would look at an OEM set of maps, see how much richer the fail-safe map is (vs OEM primary fuel map), then if you feel you want to make changes, make your fail-safe map app the same amount richer (vs your primary fuel map).

 

I don't think too many people do this. But hey. Why not.

[GTWILLY]

No no I have two maps that are made for a stage 2 car, and both (primary) fuel maps are identical except for a small change at 4400-4800 RPM at mid load levels. However one of the tunes has slightly RICHER (fail safe) fuel map values and the other one has much LEANER values. I am confused by that tune as I would have thought that a (fail safe) fuel map sould be RICHER right??

[LittleBlueGT]

No no I have two maps that are made for a stage 2 car, and both (primary) fuel maps are identical except for a small change at 4400-4800 RPM at mid load levels. However one of the tunes has slightly RICHER (fail safe) fuel map values and the other one has much LEANER values. I am confused by that tune as I would have thought that a (fail safe) fuel map sould be RICHER right??

 

I just checked a COBB stage2 map (real quickly) and the failsafe map is about .5 point richer.

[GTWILLY]

See that bothers me as I used to run this tune and yes it was a faster tune but I am wondering why this person tuned it with a LEANER fail safe map... The tune I am running now, the fail safe map is about .5 leaner in midrange RPM at full load but tappers off a lot by redline and is down in the 9.##'s.

[GTWILLY]

I am not going to throw any names out there but the tune I am running now is a proffessional tuner's tune that was bought and the other one I am analyzing is someone off of Open ECU forums.

[LittleBlueGT]

I am not going to throw any names out there but the tune I am running now is a proffessional tuner's tune that was bought and the other one I am analyzing is someone off of Open ECU forums.

 

When you read a ROM that has been married with an AP, a few things look different. Notably, some/all of the RT tables.

 

For example, I can read my ROM with ecuflash, and many tables I have examined don't seem to make sense.

 

 

I can flash a new base map to my car with ecuflash and the old RT map is still there. COBB does something different with that info.

[GTWILLY]

Oddly that one is the one that makes sense, it is the one from Open ECU that is weird

[SeeeeeYa]

I am not going to throw any names out there but the tune I am running now is a proffessional tuner's tune that was bought and the other one I am analyzing is someone off of Open ECU forums.

 

Not that this necessarily pertains to you, but at some point I thought I had the same situation, i.e., a High Det Fuel table that was leaner than the Primary. What I discovered, however, in my case, was that the two maps had two different load scales and what I perceived as leaner was not so.

 

The High Det table is usually left untouched, as it probably should be. As such it retains the 2.5 max load scale Cobb designed it with. When someone buys, or makes, a map it is usually for increased performance and the fuel, timing, and DA tables are rescaled. This will create perceptual errors when comparing them with tables with other load scaling, such as the High Det map.

 

Just a thought. Otherwise someone made a grevious mistake. No High Detonation Fuel table should be anything other than richer than the Primary Fuel table.

[LittleBlueGT]

Oddly that one is the one that makes sense, it is the one from Open ECU that is weird

 

That is not what I am saying.

 

The map you read off the ECU probably is displaying the correct failsafe map (as it is not a RT table) but the primary fuel map that it is displaying is likely not the table your ECU uses for primary fuel (as that part of the map is a RT table).

 

Does that make sense now?

 

I don't know, but your map may be just great, but you cannot rely on what it got from the ECU in certain tables.

 

 

RT tables in AP:

 

-AVCS

-Boost limits

-Boost targets

-TD High

-TD low

-TD proportional

-WGDC high

-WGDC initial

-CL modified load

-CL TPS

-injec lat

-injector scale

-intake cal

-primary fuel

-DA

-primary ignition

-rev limits

 

 

Any of the above tables that you are looking at in some other software is likely not what yor ECU is actually using.