Latest work for 2004 USDM STI, GT30 turbo map

Latest work for 2004 USDM STI, GT30 turbo map

Postby Freon » Tue Sep 19, 2006 12:13 pm

I have identified a few extra maps. Mainly the CT timing compensation and IAT timing compensation. I threw a few guesses in there (the question marks), but I'm pretty darn sure about the CT and IAT corrections.

Also, the latest build of my GT30 map. It is still very conservative.

AJ243 XML:
http://freon.shackspace.com/car/ecu%20a ... ZJ710J.XML

AJ243 ROM for my car:
http://freon.shackspace.com/car/ecu%20a ... 3-rev3.hex
edit 9-24-2006 new revision:
http://freon.shackspace.com/car/ecu%20a ... 3-rev4.hex

Factory AJ243 ROM (for comparing):
http://freon.shackspace.com/car/ecu%20a ... _AJ243.bin

Be warned it is still an ongoing development build of the ROM. Still only running about 18psi. I do think there is some interest in opening this map side by side with a stock map so you can see what kinds of things I have done, but I wouldn't recommend using it in your car. Particularly with my boost control solenoid, my boost control maps are rather violently different than you'll see elsewhere.

Relevent mods:
Perrin GT30 rotated mount turbo kit (T31 .82 AR turbine)
Perrin BigMAF intake
Tial 44mm external gate
Full 3" catless exhaust
ARO Premair 4-port boost control solenoid w/speed control
SSAutochrome front mount intercooler (24"x9"x3")
Stock BOV

Accomplished:
Extended map load ranges with reasonable values
Idles and cruises fine
AF in fuel maps is close to what I actually get per my wideband
MAF scaling reasonable
Relatively det/knock free map
Drivable every day
Spools up great

Still need to work on:
Actually running 20-22psi and pushing the loads the turbo is capable of
AVCS map needs lots of testing above 4800rpm
Still tweaking boost control a lot, tricky with 4 port solenoid
I didn't add any timing to the timing maps in the stock ranges, just extended. I think with the large turbo and front mount I could add some timing in the moderate load ranges

Engine bay
http://freon.shackspace.com/car/perring ... tdone1.jpg
Solenoid Image
Last edited by Freon on Sun Oct 01, 2006 6:45 pm, edited 2 times in total.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby fasterthanurwrx » Wed Sep 20, 2006 2:13 pm

How much different would a Map for a GT35R be?
fasterthanurwrx
 
Posts: 19
Joined: Sat Sep 16, 2006 4:28 pm

Postby Freon » Fri Sep 22, 2006 8:27 am

Again, I don't suggest loading this map on any other car, let alone use it on another car, different turbo, etc.. However comparing it to the stock map might be of interest.

You can at least see what sort of direction I'm taking things, how I deal with expanding the maps, how I deal with my custom solenoid, etc.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby Freon » Sun Sep 24, 2006 3:43 pm

Map and XML updated today. There are a few bugs in the map defs from XMLwrite causing floating point errors when some maps are opened. Fixed. Started writing comments on guesses to what some of the unknown maps are.

Map updated after more logging.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby Freon » Thu Sep 28, 2006 9:43 pm

Drag logs on the rev4 map. Still having boost spike issues. I made changes and will be running again next week.

12.6@107 and 12.7@106 runs, my exhaust silencer was still in, kinda killed top end for me.

http://freon.shackspace.com/car/ecu%20a ... -50-31.csv
http://freon.shackspace.com/car/ecu%20a ... -29-07.csv

Not wonderful examples, just want to post so you can follow what I'm doing. Mainly I am aiming to get rid of the boost spike for now.

Here is a post-drag revision of the ROM:
http://freon.shackspace.com/car/ecu%20a ... 3-rev5.hex
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby Freon » Fri Sep 29, 2006 10:44 pm

Latest rev didn't help much with spike. Created a worksheet to compare the stock turbo PID process control maps to what I was messing with. I had hacked them up pretty bad in rev5. Reformed the P and I maps.

Rev6 of map, not much changed but boost control stuff. Lowered stationary RPM open loop 1D map as well, wonder if this would allow me to run the open loop fuel map when launching at drag... Will have to log to see.

http://freon.shackspace.com/car/ecu%20a ... 3-rev6.hex

http://freon.shackspace.com/car/ecu%20a ... omparo.xls
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby Freon » Sun Oct 01, 2006 6:44 pm

Still no dice. I'm starting to search for some sort of derivative map. I have some guesses but I'm not sure I want to just do a trial by fire to find out if one of them really is a derivative map.

Here's a log snippet.

Image

You can see to maintain 18psi I require only ~20% WGDC! The problem is getting it to go from 80-100% down to 20% quickly. This is a log with rev6 of my ROM. What I really want that to do is peg at 100% until I reach about 10psi, then drop to 35% until 15psi, then drop to 20%.

My next rev uses the Max WG maps to cut the duty down, but that's not what I really want to have to do. On throttle tip-in or gearshifts at high RPM I would still like the WGDC to spike to near 100 for a brief moment. I may be able to make the proportional map more violent.

New XML, much improved, same link as above. I started placing lots of guesses on the maps based on where the definitions are. It seems in general all the maps that control certain aspects of the ECU are grouped together in either the definition area or actual location in the ROM, or both. Not sure what I can really do to identify them, though.

I borrowed a rather thick (1200page) control theory book. I'm going read it when I get the chance to see if I can gleam anything from it to help me out.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby Freon » Sun Oct 01, 2006 6:47 pm

Here is a functional diagram of the 4 port solenoid vs what I believe people typically do with a 3port solenoid and a dual port external wastegate:

Image
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby merchgod » Sun Oct 01, 2006 7:44 pm

The 32-bit roms tend to group the references to the tables at the end of each block of code. I'm not talking about the map structure offsets, as you are obviously aware of, but the actual literal addresses in the rom that point to the map structure offset (2d or 3d) or the individual 1d values. So, there will usually be a relatively small block of code that references this data which is immediately after the code. Then after that another chunk of code and its data so on. You'll see table offsets, ram locations, and subroutine locations in nice little chunks after the code. So, even if you are not disassembling the entire rom, you can find the literal reference in the rom to the tables of interest and then back track to the last data point and mark what's in-between as code. Also, look at the ram locations because some of them are already known for Ecuexplorer for IAM and engine load and once you know it for one revision it is relatively easy to find it in other revisions.

I've been developing a program that helps me to create defs. It gives you all the 2d and 3d tables, as does xmlwrite, but it sorts it by the multiplier/additive, dimension, and storagetype and can be used on any 32bit rom. This helps a lot since tables with the same multiplier/additive will often have a similar function. It also places the size in the map name so if you are looking for a particular sized table, you can find it easily. It will read your ecuedit xml file and add the known names to the tables. It will also determine the correct datasize for most tables with no data size value, which are not all float.

If you send me your latest xml, I can send you a new xml in ecuedit created by my program which might help you in your search.
merchgod
 
Posts: 243
Joined: Fri Sep 29, 2006 10:17 am

Postby Freon » Sun Oct 01, 2006 8:31 pm

Hmm, sounds interesting. I'm just *dying* to know if there is a derivative table in addition to the proportional and integral maps. If there is and I can tune it, I'm certain I can fix my boost control issues and this 4 port solenoid is *really* going to pay off big. I'll see if I can make any sense of that advice. There are just so many tables that look like error/correction tables ala the PI tables for boost, but there are several other closed loop controls in the ECU, like for fuel and the intake cam advance.

I'd love to be able to disassemble the ROM, but I have no access to the needed software, not to mention it is a bit over my head. I'm a database and systems geek, not a hardcore coder. But I think I might know enough to at least follow the references around, spot some of the logic, and make halfway intelligent guesses.

Here's the latest XML:

http://freon.shackspace.com/car/ecu%20a ... ZJ710J.XML
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby Freon » Sun Oct 01, 2006 8:46 pm

These references you talk about, what do they look like? I don't see any references to the maps (rows, data, etc) besides the "map definition area" of the ROM.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby merchgod » Sun Oct 01, 2006 9:16 pm

Freon wrote:These references you talk about, what do they look like? I don't see any references to the maps (rows, data, etc) besides the "map definition area" of the ROM.

The references are to the offsets of each map structure in the "map definition area". So, for example, the offset to the map strcuture for your rom for target boost A is 0x538F8. So, if you search your rom for "000538F8" you will find it at 0x135e4. Surrounding this value you will see:

0001F822 00056648 FFFFA318 FFFFA314 00002100 000538F8 00053914 ETC.

These are all data elements that are used by the preceeding block of code. 1F822 is a location for a subroutine. 56648 is an offset for an unknown map (looks like a single value in float - 1d). The FFFFA318 and FFFFA314 are references to ram. 2100 is another location for a subroutine. 538F8 is target boost A map structure offset and 53914 is target boost B. The data continues on until the next block of code and then after that block of code is more data. While you might not know the exact code by looking at this with a hex editor, you will get an inkling that these tables are related and this might point you in the right direction by looking at the unknown tables that are grouped with the known tables.

I don't know much about machine language and the 32bit roms can be very cryptic when I look at the code (compared to 16bit roms which are a little more straighforward), but I can occasionally gain some more insight by taking a peek with IDA. But even if you don't want to mess with that or don't have the software, using the above method might help. Note, for 1d values, the address is the direct location of the data and if the 1d values make up a table of more than 1 value then each value's location will be listed individually depending on how they are used by the preceeding code.

I'll get you the xml file we talked about tomorrow. Hope this helps.
merchgod
 
Posts: 243
Joined: Fri Sep 29, 2006 10:17 am

Postby Freon » Wed Oct 11, 2006 7:37 pm

More goodies. Ran a 12.3 @ 109 and 12.3 @ 110 last night. Here's the map:

http://freon.shackspace.com/car/ecu%20a ... 3-rev7.hex
(I overwrote old version)

Timeslip:
http://freon.shackspace.com/car/ecu%20a ... 09.03a.jpg

Getting closer on boost control. My average duty cycle on the entire drag run from launch to finish was around 15% and pressure 17.5 psiG! I'm not giving up on the 4port solenoid yet, despite difficulties getting the factory boost control logic to control it.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Postby STIBungy » Mon Oct 23, 2006 12:48 pm

What injectors are you running on your setup? My car and mods are fairly similar to yours other than my turbo is an SBR GT-12 and I'm using the GM boost solenoid. I should be picking up a Perrin big maf sometime this week as well. I've maxed out the stock maf housing at 306 grams.
STIBungy
 
Posts: 6
Joined: Sat Sep 23, 2006 6:45 am

Postby Freon » Tue Oct 24, 2006 3:35 pm

Perrin modded STI sidefeed injectors

I wouldn't consider a GT30 rotated kit very similar to a bolt-on. Again, don't try to just load my map up on ANY other car. I suggest you compare it to a stock map to see what kinds of things I did to deal with what I have.
Freon
 
Posts: 700
Joined: Thu Nov 17, 2005 5:50 pm
Location: Indianapolis, IN

Next

Return to Subaru (all models)

Who is online

Users browsing this forum: No registered users and 53 guests