TD_D mod journey. From stock to the 'bastardmobile'

Forums New posts
Td_d

Td_d

Commander In Chief
^ yup, that's what I'm going to do - give you 2 /3 short logs to make sure the scaling's now buttoned down (already checking the morning drive! :tard:) looking good so far on the learnt correction. On the comps - I know that it will take some as we say locally, seeeeerrrrriiiiiiiaaaassss driving! I saw some recommendations on Romraider forums stating 2 hours! Once I've confirmed (and uploaded the logs), I'll make a time to do a nice long drive, and I'll also make sure I do it in S mode, limits the C/O transitions nicely.

BTW - this morning's drive felt very smooth, nice power delivery, looks like it's definitely working. Ok, just finished crunching the numbers - learnt parameters are getting nice and close to zero - worst learnt correction is -3% early on, but mostly around -0.5% to 1.5%. Woohoo!
 
Last edited:
Td_d

Td_d

Commander In Chief
Even on the combined learnt and corrections, the graph's not looking bad at all - between 1.6% and -7%, but mostly around -2% to -5%.
 
HolyCrapItsFast

HolyCrapItsFast

Drinks beer!
Okay this is good news!

Yeah it could take a bit of driving for the trims to settle in. To be honest when I make changes to my maps i usually wait until the next day to start logging and after I have driven it for atleast 15 miles first. For some reason I notice the learned trims always start out not perfect but after a good amount of driving they settle in and are fine.

It sounds like yours is doing the same and this is a good sign. Okay then, I'm standing by and awaiting the next logs. Take your time, no rush.
 
Last edited:
Td_d

Td_d

Commander In Chief
Td_d

Td_d

Commander In Chief
Oh - another odd thing - today, with the weather about 8-9 degress celcius colder than the first day rom was flashed, suddenly very lumpy on the bottom end (up to 3000rpms)...
 
HolyCrapItsFast

HolyCrapItsFast

Drinks beer!
Td_d said:
Ok, I've done a couple of logs, looks like the MAF scale is about -3.5% off - straight as an arrow, but between -3 and -4% off - here's the spreadsheet on the learnt corrections http://dl.dropbox.com/u/3099895/TD_D_3-26-2011_RoughTrimErrors.rar and here are the logs it's based off - http://dl.dropbox.com/u/3099895/TD D 26 Mar 2011 Logs.rar

I think maybe I should make the adjustements based on the proposed MAF scale in the spreadsheet, and then try out the load comp drive?
Click to expand...

Okay this is good. What you want to do instead of messing around with Maf Calibration (because it is so perfectly flat) is apply a global correction to injector scale.

What the data is telling us is that the ECU is correcting for 3% to much fuel. To correct this using injector scale you need to tell the ECU you are using a 3% larger injector. Apply a 1.03 multiplier to injector scale.

Td_d said:
Oh - another odd thing - today, with the weather about 8-9 degress celcius colder than the first day rom was flashed, suddenly very lumpy on the bottom end (up to 3000rpms)...
Click to expand...

Yeah. I'm not gonna worry till load comp is adjusted.
 
Td_d

Td_d

Commander In Chief
Great stuff - I'll do that, reflash, let it settle in and then go for the long drive!
 
Td_d

Td_d

Commander In Chief
Ah, now this is a pretty looking graph - learnt corrections are almost all at zero, slightly richer just before going open loop (which can be adjusted by tweaking the MAF scaling marginally)

album.php

http://www.igotasti.com/vBforum/album.php?albumid=28&attachmentid=863

Even with learnt and correction combined, range is between -5% and 5% with most of the middle pretty close to 0% as well (so, pulling fuel at very low g/s, and adding fuel just before going open loop).

Right! I will do my upmost to do a nice long, hour drive today to sort out the comps!

This is the log, by the way: http://dl.dropbox.com/u/3099895/log0008.zip

[JJ - LOOKS LIKE IMAGE LINKING'S BROKEN ON THE SITE]
 
Last edited:
HolyCrapItsFast

HolyCrapItsFast

Drinks beer!
Holy Fucking WOW. That is great! Mine doesn't even look that good.

For the purpose of this we are done with Maf and Injector scale but you may choose to further refine this by continuing to perform these steps and achieve perfection (if it is even possible at this point) but for us now, we will continue on to load comp as suggested. Can you please send me a copy of your latest ROM?

------------------------------------------------------------------------

In the future (not now though) if you feel the urge to refine your closed loop MAF calibration you will do it a bit differently because you are fine - fine tuning it. The filters you will apply, if not already, are...

Closed Loop Switch = 7,10
Engine load = .5 (AND) A/F Sensor < 13
Engine load = .5 (AND) A/F Sensor > 16
A/F correction < -14
A/F correction > 14

And when you run the Maf Trim Error tool, this time you will apply some smoothing. 3 cells is sufficient.
 
Td_d

Td_d

Commander In Chief
Yeah - I was pretty amped when I saw that! Now how's this for a :tard: move (facepalm). I specifically make time over lunch to drive for an hour, so I can do the load comp, mission off, get onto the highway, put a good solid 1/2 hour at S and I mode, nice and smooth, then come off to do some town driving. At which point I realised that I had forgotten to take the OP2.0 with me... argh!

So - I did manage to get 1/2 hour drive mostly highway, and will do the town driving this evening, so we can append it, to get a full hour. Couldn't help myself, so plugged it into the load comp spreadsheet, and indeed it comes out with corrections - now if only I understood what the hell it means :lol: What I did see is that on the % correction vs. RPM chart, there is a predictable huge jump in corrections up to between 10% and 23% around... drumroll - 2600RPM to 2900RPM. Here's the log - I'll post the town driving component as soon as I have it:
http://dl.dropbox.com/u/3099895/log0009.rar

Check this graph out - exactly what we expected to see!

http://www.igotasti.com/vBforum/album.php?albumid=28&attachmentid=864

Here's the ROM: http://dl.dropbox.com/u/3099895/750RC_3-26-2011_Rev-A.rar
 
Td_d

Td_d

Commander In Chief
Awesome, thanks! So - any pointers as to how this actually works? I'm curious as to the mechanics of what the compensations are actually doing.
 
HolyCrapItsFast

HolyCrapItsFast

Drinks beer!
Yes. Aside from just associating a maf voltage with g/s and than deriving fuel, the load and manifold pressure also play a huge part in determining fuel. These compensation table simply account for other factors that may effect the actual required fuel.

The easiest way I can explain it is if your load compensation tables are not adjusted properly you will notice this in fluctations in the unfiltered error trims vs filtered trims and during certain load situations. Properly adjusted and you will see that your unfiltered error trims are very close to your filtered error trims.
 
Td_d

Td_d

Commander In Chief
Hmm... that makes sense, since the filtering, I've noticed, removes the lean then rich condition that the 'stumble' causes at around 2800 RPM - you can see it quite clearly as the AFR lean out significantly to over 16, and then the ECU overcompensating and overfueling. Ok, it's beginning to fall into place. Was there enough data points in the logs - it's quite difficult to get readings at some points in these arrays...
 
HolyCrapItsFast

HolyCrapItsFast

Drinks beer!
Yes I believe there was enough data. You got a good spread. :tup:
 
You must log in or register to reply here.

Similar threads

D
EJ257 - 2008 STI Hatchback, engine replacement?
Replies
0
Views
455
deraeler
D
A
WOT stumble/lean
Replies
0
Views
678
Abaddon04
A
J
2014 wrx sti; 90k miles; columbus, oh $19.5k
Replies
0
Views
785
jhestyr
J
Alin
State Of The Forum In 2024?
Replies
1
Views
315
IGOTASTi
IGOTASTi
cyclefiend1123
Horn Stuck On?
Replies
0
Views
353
cyclefiend1123
cyclefiend1123
Top