I've been Whitelined!

Great news, my Whiteline Anti Lift Kit is finally here! I ordered it in Nov 07 and it has finally arrived. Actually, I bought a bunch of other stuffs:

KCA359M Anti-lift/caster kit RACE MOTORSPORT - $480
KCA313 Roll Centre Adjust - $340
KSR202 H/duty steer rack mounts - 30mm wide bracket $65.00
W0584M Heavy duty gearbox mnts-RACE $80

The other items arrived in Dec07 (have not installed them yet). Anyway, here are some photos of them:


All the Whiteline items


This is the Anti Lift Kit. A fren told me that he could corner at high speed with this and not feel anything.. I wonder how true that is :-)


The Roll Centre Adjust (RCA).


The steering rack mounts - the guy who did my alignment told me that it's worn (probably from the car where I got the close steering rack from).


My gear box mounts - hopefully this will help my gear syncro issues.

Will update when I get them installed.

Back to Closed Loop MAF Scaling again?

In my last post, I described how I tried to do MAF Scaling for idling and how bad it turned out. But I also did mentioned that something good came out of it as I somehow hit a sweet spot around the MAFv 1.3 region. I used that value and applied it to my MAF table BEFORE I made any changes to for idling. This is now rev5 of my closed loop MAF scaling:


Here are the charts:
Total Corr vs MAFv

Isn't this graph sweet? I mean you can see things have smoothen out over a wider range. I'm so happy to see this!

MAF Check2


Scatter Plot of Total Corr vs MAFv

The scatter plot shows how evenly distributed the data points are. Another nice graph.

Injector Scaling


Here's a screenshot of Learning View for this MAF table. Looks good too, right ;-)


Ok, that's it. I'm done with closed loop MAF Scaling. Off to Open loop now. :-)

MAF Scaling for Idle - Is there really a need?

If you look at the Total Correction vs MAFv graphs, you would notice that at low MAFv, my scaling is really way off. So I thought, let's give MAF scaling at idle a shot. Here's what I did:

1. Use the log data for my current MAF scale (using mickeyd2005 columns), filter out all data with throttle angle > 0 (I only want idle)
2. Filter out all data Open Loop data.
3. Keep only AFR Correction (STFT), AFR Learning (LTFT), MAFv.
4. Sum up STFT + LTFT into a new column Total Correction.
5. Sort data via MAFv
6. Interpolate the data (using williaty's method) and take the mode of values for each MAFv <= 1.37 The first time I did this, I used data logged previously for my current ROM. Once that flash has been done, I did another set of logs: 1. I startup the car, wait till it is fully warmed up. 2. Drive around bit and let it idle. 3. Take logs for about 1-3 mins. 4. Drive around abit (to eliminate heat soak) 5. Take logs for about 1-3 mins. I filtered out the data using the steps mentioned at the top. Re-flash the car. I did this a total of 3 rounds. NOTE: I only changed MAFv 0.98 ~ 1.37 (as I noticed that idling normal takes place within this range only). The changes are shown in the screenshot below. (Idle1/Idle2/Idle3 Values are the final changes that I applied to the ROM)


The idling of the car seems fine. Next I did another round of logging and use mickeyd2005's spreadsheet to analyse my MAF scaling. Here's the result:

Total Correction vs MAFv:


As you can see my low end MAFv (<1.3)

MAF Check3 (Scatter plot of Total Correction vs MAFv):


Overall, you can see that the graph above looks quite good (correction centered around 0%).

My injector scaling:



My findings:
1. Either my method is not correct or my MAFv is screwed up at low idle (hey, it could be due to my APS CAI).
2. While driving the car after the changes to my MAF table for idle, I noticed that idling corrections vary significantly after driving the car hard or at high speeds. Meaning after driving the car at 130km/h for about 10-15 mins. When I bring the car to a stop and idle, it will compensate a lot (up to 10-15%).
3. The funny thing is, before driving hard, the idling had only about 0.78% correction! Strange, isn't it?

To top it all off, my AFR Learning A has worsened from -4 to +9 (see screenshot below). My AFR Learning D also suffered abit this time around from 0 to -0.49 (I'm surprised about this - might need some monitoring).


Well, I'm not sure why it behaves this way but I have decided to revert back to my original MAF table BEFORE the changes for Idle (I considered it a failure). But I did make one more change though, to my MAFv 1.29 and 1.33 as in my logs for Idle changes, it shows the 0% correction in this range (but not for MAFv less than 1.29)

The top MAF table are the changes (changes in grey) and the bottom MAF table are values BEFORE the idle MAF changes.



I'm going to go through another round of logging to see if my MAF table has improved at MAFv 1.3 range. One thing's I've learned this weekend, MAF scaling for idling is not easier that CL Scaling as well. But given that it only affects idle, if a car is already idling fine - is there REALLY a need? Hmmmm...

Will keep you guys posted. :-)

Getting close to Closed Loop Scaling

The MAF scaling that I have been doing has been Closed Loop - one always do close loop scaling first before scaling your open loop. (close loop has to be right before open loop scaling can be done).

I will not bore you the details of each iteration but will give the major milestones of my MAF Scaling. After using several rounds of williaty's method, I switched over to mickeyd2005's method (it was much much easier using his spreadsheet). Here's some of the results (from mickeyd2005's spreadsheet):

My correction vs MAFv:


As you can see, my scaling looks ok except for spots at 1.66 (-5.88), 2.38 (+3.07).

There are two minor spots at 1.44 (+2.75), 1.96 (+2.22) but these are close to 2% total correction so that are not too bad.

The other graphs for these data set is:

Maf Check2:


Maf Check3 (Scatter Plot):


Injector Scaling:


Concentrating on these 4 spots, I went through another 3 rounds of iterations and here are the results:

My correction vs MAFv:

Compare that with the earlier graph (I reproduce here for easy reference):

You can see that my curve has smoothed out quite alot in those areas I mentioned earlier, right? Heehee.. Finally I am getting somewhere (after 1 month of trial and error and wandering in the dark).

My Learning View also shows 0 correction for AF Learning D:

I'm quite happy with my progress :-)

Anyway, here are the other graphs from this dataset:

MAF Check2:


MAF Check3:


Injector Scaling:



Lessons learnt from the many many rounds of closed loop scaling:
1. Although williaty's method can be quite laborious, it helps to ensure that you are picking the right set of data for scaling.

2. mickeyd2005 spreadsheet is simply heaven but you can't just pick up the suggested corrections because you really have to study the graphs and number of data points to ensure that you are making the "right changes". I made the mistake of just copying the corrections directly and my results were worse (I made up to 11 iterations without getting anywhere!). In the end I started over again from where I left off using williaty's method (mickeyd2005's advice).

3. Once I became more picky on which changes to by looking at the graphs and the number of data points (the more the better), I began to have better scaling results.

4. When scaling for close loop, it's good to concentrate on a range when making changes rather than changing so many points at once. Patience is key here. I concentrated on those 2 major points and when they came close to 0, I made changes to the minor points.

5. For close loop, the higher MAFv ranges (>2.0) are harder to get data - you need to drive around 110km/h ~ 130km/h to get data. Here are some speed ranges and MAFv values

For MAFv around 1.6 ~ 1.8 drive at about 60km/h
For MAFv around 2.0 ~ 2.1 drive around 90km/h
For MAFv around 2.4 ~ 2.6 drive between 110km/h and 130km/h

6. Open the dashboard view when driving and just keep a quick watch on the MAFv and Close/Open Loop indicator to ensure you've got data for all the ranges you need.

7. I can only scale close loop up to MAFv 2.6 for my car. I guess I have to use Open Loop scaling for higher MAFv

8. Ignore high total corrections variations for MAFv < 1.3 as this range falls mainly within the idle range and it can be very erratic (maybe its my APS CAI that's causing this - what others have reported too).

9. Therefore when performing close loop scaling, concentrate on MAFv range 1.5 ~ 2.6. You can work on MAFv 1.3 ~ 1.5 when 1.5 ~ 2.6 have stabilised.

10. Keep your throttle constant and light when logging and always watch that open/closed loop indicator (8 is for closed loop). If you not, you could be just logging open loop which is useless for analysis.

Anyway, here are my MAF tables for two set of graphs:

First:


Final:

Ok, that's all for now. Hope this post has been useful.

Getting started on my MAF Scaling

When I first plugged in the stock Sti ecu to my WRX, I was constantly getting the error code P0171 (System Too Lean). %he main reason for this is that my MAF scaling table was way off.

I read on the Net that the WRX stock intake is 65mm. Since the APS CAI is 70mm, this works out to an increase of 1.1 (70/65 = 1.1). So what I did was to increase the MAFv scale by 1.1 from the stock values.

Having dones this, I did my first log using williaty's method and here are the final corrections I got:

MAFv Mean Mode
0.94, ,
0.98, 16.41 ,
1.02, 14.26 ,
1.05, 15.31 ,
1.09, 15.14 , 18.75
1.13, 18.01 , 19.53
1.17, 18.54 , 16.4
1.21, 18.69 , 19.53
1.25, 18.52 , 18.75
1.29, 17.59 , 18.75
1.33, 16.79 , 17.18
1.37, 17.50 , 17.96
1.41, 16.42 , 17.96
1.48, 8.725 , 6.25
1.56, 12.54 , 11.72
1.64, 12.60 , 9.37
1.72, 13.98 , 13.28
1.8 , 17.77 , 18.75
1.88, 20.20 , 18.75
1.95, 20.78 , 22.65
2.03, 19.87 , 19.53
2.11, 17.26 , 14.84
2.19, 16.63 , 16.4
2.27, 13.91 , 12.5
2.34, 12.49 , 12.5
2.42, 10.47 , 9.37

As you can see, at this point, my MAF table is still way off (corrections are mostly > 15%).

Luckily someone (airboy) on nasioc pointed me to someone who has a similar intake and has the calibration data: http://www.iwsti.com/forums/cobb-street-tuner/54421-aps-70mm-cai-intake-calibration.html

I compared my corrections with the calibration data in that link that I found that I'm not that far off, so I guess I am doing ok. in fact, after a reflash with the new corrections and the car runs much better, hardly any P0171.

Here's a screenshot of my calibrations up to 9 iterations I did using williaty's method:

MAF Scaling Methods

As mentioned in my earlier post, since I have changed my stock intake to an APS CAI 70mm, I need to re-scale my MAF.

After doing alot of research on the Net, I found 2 methods to scale my MAF, found on nasioc.com and on romraider.com. Both methods uses spreadsheets to run statistical calculations/analysis to determine the final MAF calibrations.

1. williaty's method: http://forums.nasioc.com/forums/showthread.php?t=1427448
2. mickeyd2005 method: http://www.romraider.com/forum/topic1871.html

There's actually another method here but it's not as detailed, though it's for exactly the same intake as mine. It still makes a nice read though: http://www.iwsti.com/forums/cobb-street-tuner/54421-aps-70mm-cai-intake-calibration.html


williaty's method requires abit more time once you have logged the necessary parameters since you perform each step manually (perhaps it can be automated using macros).

mickeyd2005's method is faster since he has already built a spreadsheet to automate all the tasks but be warned, you need to study the data points and determine which one is valid. Also, do remember how your driving style was when you were logging as it will determine which MAFv ranges are ok to use.

In my opinion, both methods works fine but one thing's for sure, don't expect to get quick results, you need alot of patience to log and study/understand the data before making changes or else you are only making matters worse.

Some tips:
1. Log good data, not just ANY data. What I mean is for both methods to work fine, you need to be logging MAFv which is in constant state, i.e. when you are driving on a highway with constant throttle and steady driving. If are logging in conditions where you are in constant stop/start conditions, your data is probably not that great.

2. Go for more data rather than more iterations. Of coz, log good data.

3. Try to make small number of changes as you progress into more iterations. Reason being you are probably in fine tuning mode as the number of iterations increases and thus big changes will most likely have a big impact, wasting your previous iteration efforts.

4. Study each data point and value you are going to change, if possible look back to older data/graphs to study the impact of previous changes/values.

5. Be patient. :-)

It's all about SCALING.

No one puts it better than williaty did on his post on MAF scaling on nasioc.com:

"One of the most common mods people do to our cars is to change the intake. One of the most common results of changing an intake is to change the relationship between the amount of air flowing past the MAF sensor and the total amount of air entering the system. Without adjusting the MAF scaling to reflect the changes to the intake, the ECU will constantly mis-calculate the amount of air entering the engine. This will affect nearly everything the ECU does from fueling, to timing, to the Closed-Loop to Open-Loop Transition. Re-scaling the MAF is a process of altering the programmed relationship between MAF sensor voltage and the grams per second of air entering the engine." (see the full post at: http://forums.nasioc.com/forums/showthread.php?t=1427448)

Fuel injector scaling works in the same way. When one swaps the stock fuel injectors for another set, the injector scale table must also be re-calibrated.

If you have read my earlier post, I've changed both my fuel injectors (to STi 550cc) and intake (to APS 70mm CAI). So both my MAF and injector tables needs to be re-calibrated.

Actually changing both at the same time is not the right way to do it as you can NEVER scale both MAF and Injectors at the SAME time (both tables affect each other). The proper way to do it is:

1) Install new injectors with a factory intake. Scale injector size to minimize fuel trims. Check your fuel table against logs and see how accurate it is. It should be close even in open loop and really high load with the stock intake.

2) Install aftermarket intake. Scale the whole MAF scaling table.

In my case, I'm lucky coz the ecu which I have got (I swapped my ecutek enabled stock WRX ecu) is an STi 03 WRX (see screenshot below from romraider):


The injector flow table is already set to its correct value (see screenshot of injector flow table from romraider):






I'm assuming Subaru's calibration is correct (it should be since the engineers in the factory have spent so many hours calibrating it) so, the only scaling I need to do is my MAF Scaling. I'm glad it has turned out this way or else it would have been difficult to do both at the same time. :-)

I hope to talk more about the different methods of MAF scaling I've tried on my next post.

Romraider now supports LM-1!!!

After going through so much trouble with my LC, guess what? The latest version of romraider (rev033) now supports LM-1. I've just did some tests and it works even better than LC-1 (occasionally, my LC-1 gets hung at 20.933)

Yippee.. can't wait to do my open MAF scaling.

LC1 + Romraider

It's been awhile since my last post. Sorry, I've really been busy. Some updates since my last post. I've managed to hookup LC-1 to my WRX. Since I wanted it to be portable, I hooked it up using my exhause clamp and powered using a cigarette ligher power source. The inspiration came from the following link: http://www.m2motorsports.com/files/lc-1_write_up.pdf

It's pretty good and gave step by step on how its done. I'll upload some pictures when I have time. The only difference in my setup and the one in the link above is my grounding - I ground everything together to power source's ground. It seems to work but I'm not sure if it's the right way to do. The LC-1 manual says both grounds have to be grounded to the same source though.

See my LC-1 connected to my exhaust clamp on my tail pipe:

Anyway for those interested in hooking up LC-1 in a permanent way to a subaru, here's a link that shows how its done: http://forums.nasioc.com/forums/show...php?p=19084368

For some discussions on LC-1 which I posted on NASIOC: http://forums.nasioc.com/forums/showthread.php?t=1463068