Fuel/Spark tuning for LH 2.4/EZK with TunerPro! - Page 9

ipdown · 01-26-2011, 04:51 PM

Quote:

Originally Posted by gross polluter

This is why I suggest we start a thread or sub forum on attacking the hardware aspect of LH. Back when we picked at the DSM ECU in the early 2000's it took a community of 40-50 people to actively pick at small parts of the ASM, to report their findings, to develop the necessary documentation to modify the ECU code on the ASM level.

I agree, I can post everything I know about hardware and software. So, if there are people who wants and can help, let's do it!

ipdown · 01-26-2011, 04:54 PM

Quote:

Originally Posted by stoni

Jan Willem, can you send me this xdf via email, have still no login at ecuprojects, the board admin has still not activated my account

No, if you do it this way the mixture is too rich, and this in every area. Main reason for this is that the 5x5maps will be calculated in another way. You can change all fields to the ones from the 984 except this 6 tables.

Yes, the 962 box has obviously the same hardware like the 977/984 ECU, means with the chip besides the eprom.

Dirk, I think there is a problem with the database on ecuproject, it is slow and continuosly crashes for me when I post there. May be that's why you can't register. I will write to Dilemma (one of admins) to see if it helps..

klr142 · 01-26-2011, 05:07 PM

It took me 1-2 weeks to be accepted when I originally registered about a month ago or so... For what it's worth.

linuxman51 · 01-26-2011, 05:10 PM

hm. i registered a looong time ago, it was pretty quick back then.

orie · 01-26-2011, 05:18 PM

Quote:

Originally Posted by sbabbs

I have here a 984 ECU with the stock 984 chip and bin. It won't start or run in my car if I put it in. If I put the 984 chip in my 967 ECU it will start and run. If I put my 967 chip in my 984 ECU it will start and run. Supposedly the 983 bin is the same as a 984 it just doesn't have the immobilizer. That is the bin we need, but nobody seems to have it. It won't be in any cars here in the US.

That's great news.. I have a 984 from the UK and of course, it won't let the car start. I still need to try my TLAO chip in it, sounds like it would work.

EDIT:
Can folks recommend a good place to buy a reader/writer and rewritable chip (or should I just use my spare OE chips?)

linuxman51 · 01-26-2011, 05:26 PM

moates.net

gross polluter · 01-26-2011, 05:32 PM

Quote:

Originally Posted by ipdown

I agree, I can post everything I know about hardware and software. So, if there are people who wants and can help, let's do it!

I'm in.

orie · 01-26-2011, 05:36 PM

Quote:

Originally Posted by linuxman51

moates.net

thank you

Edit: same site as linked in the FAQ/intro. Pardon my dumb.

Also, any chip burner is also a reader, yes?

http://www.moates.net/burn2-chip-programmer-p-197.html

beepee · 01-26-2011, 06:14 PM

Quote:

Originally Posted by ipdown

Hey, I have that one. I though it is 987..

You may be correct, I am not sure myself. I can't remember where I got all these to be honest

I just polished up the 984 XDF I have and posted it, check it out.

And I am 110% down with figuring out/disassembling/understanding LH. Ipdown, if you don't mind, please send me the disassembly you have; I am trying different tools to disassemble and I am getting frustrated....

Time to drive home in the slush... ugh.

beepee · 01-26-2011, 06:15 PM

Quote:

Originally Posted by orie

Also, any chip burner is also a reader, yes?

http://www.moates.net/burn2-chip-programmer-p-197.html

Yup. Otherwise you really can't verify that it burned successfully!

orie · 01-26-2011, 06:27 PM

Quote:

Originally Posted by beepee

Yup. Otherwise you really can't verify that it burned successfully!

Just making sure. Thanks!

beepee · 01-26-2011, 08:18 PM

Just got home from a nice, long slushy drive home using the 984 bin. And, despite my skepticism, I am a convert. The 984 bin just seems... smoother. I can't put my finger on it. I think the table I labeled as fuel cut is incorrect though... It's hard to safely hit 15psi when the roads are covered in 1.5" of slush

I think further XDF development will be focused on the 984 bin... anyone else given it a shot?

klr142 · 01-26-2011, 08:22 PM

Most chips have seemed smoother after I first put them in, but maybe that isn't what's going on here... Maybe it just is that much better.

beepee · 01-26-2011, 08:26 PM

Oh I forgot to mention, with lambda on it seemed like the ECU held near 14.7 much more solidly than the 93x... I would do a datalog but all I can log is o2 right now.

orie · 01-26-2011, 09:29 PM

Got myself a programmer, chips and an ostrich. Now to have some fun.

beepee · 01-26-2011, 11:34 PM

Quote:

Originally Posted by orie

got myself a programmer, chips and an ostrich. Now to have some fun.

eta?

beepee · 01-27-2011, 03:29 AM

Ok, i've been gathering the injector constants from all of the stock BINs I have, here's what I got:

I think that the injector constants might be 16 bit, or have a multiplier of some sort. The values were obtained using ipdown's Injjector Tuner. "INJ CONST A" starts one byte before at what the program identified as constant 1 and preceeds for 4 bytes. "INJ CONST B" starts at what was identified as constant 3 and also proceeds for 4 bytes..

I think I am on to something; I don't know.

Doesn't really shed any light, does it? I can't tell, I keep nodding off at the PC.

aaaaaaaahhhhhhhhh goodnight!

linuxman51 · 01-27-2011, 08:57 AM

Hm. usually the number itself is a multiplier used elsewhere (for instance, req fuel in megasquirt is derived from injector size and displacement, other ecu's use a set value that's ~ the size of the injectors to factor into the fueling equation). Do you (or ipdown) know where these constants come in to play?

ipdown · 01-27-2011, 09:43 AM

I'll explain how I found these constants.
I have built a "stimulator" for LH 2.4, which provides it
with all sensors input and rpm signal, and monitored the injector pulsewidth
on oscilloscope. Then, having the "source" assembly of 950 ECU I focused on
data section and code references to different values from there.
Feeding the ECU with constant rpm, load, etc. the above constants changed
the resulting pulsewitdth. Reg. bitsize of them, the #4 and #5 (as numbered
in InjConst utility) could be 16 bit value, while the others are most likely
alone bytes. They could statically scale temp., voltage or load for example,
thus changing the pulse width, but still at least one of them is the
real injector constant, since I literally tried every one suspicious
referenced byte in the data section and these are that affect the resulting pulsewidth.
There are another two which also changes injection frequency, but they are probably
dangerous to try on a running engine.

ipdown · 01-27-2011, 10:14 AM

This is disassembly listing of 950 ecu:
http://dl.dropbox.com/u/9914656/0280000950.s

Browse it in you favourite text editor, preferably with
syntax hi-lighting for .asm/.s files. I am using VIM editor.
It is commented, and cross references can be searched quickly with vim.
It would be nice if there are new findings and to update them somehow..
May be put it under revision control? Or I am wanting too much..

Anyway, it compiles with as31 assembler to the same file as original binary -
the only difference are empty gaps of unused memory space, which in the stock bin are
filled with 0xFF, and here are zeroes. You can compile it and compare in hex to the
original binary file.

AS31 assembler can be downloaded from here:
http://www.pjrc.com/tech/8051/tools/...-pj3-win32.zip

Use the following command line to compile the file:



    C:\>as31.exe -Fbin 0280000950.s

This will create file 0280000950.s.bin in the current directory

beepee · 01-27-2011, 01:02 PM

Quote:

Originally Posted by ipdown

I'll explain how I found these constants.
I have built a "stimulator" for LH 2.4, which provides it
with all sensors input and rpm signal, and monitored the injector pulsewidth
on oscilloscope. Then, having the "source" assembly of 950 ECU I focused on
data section and code references to different values from there.
Feeding the ECU with constant rpm, load, etc. the above constants changed
the resulting pulsewitdth. Reg. bitsize of them, the #4 and #5 (as numbered
in InjConst utility) could be 16 bit value, while the others are most likely
alone bytes. They could statically scale temp., voltage or load for example,
thus changing the pulse width, but still at least one of them is the
real injector constant, since I literally tried every one suspicious
referenced byte in the data section and these are that affect the resulting pulsewidth.
There are another two which also changes injection frequency, but they are probably
dangerous to try on a running engine.

I am so jealous of your hardware hacking abilities. Can we see a picture of the setup?

I'm going to download the ASM stuff today. These things that you have done are pretty damn impressive.

orie · 01-27-2011, 02:33 PM

Quote:

Originally Posted by beepee

eta?

According to UPS tracking, next wednesday.

Quote:

Originally Posted by beepee

I am so jealous of your hardware hacking abilities. Can we see a picture of the setup?

I'm going to download the ASM stuff today. These things that you have done are pretty damn impressive.

Me too. I've always loved electronics, but i never got that far into them. Do show!

ipdown · 01-27-2011, 02:57 PM

The first one is nothing fancy, basically proper power supply, some pots, and an ATtiny MCU for RPM signal. I don't have camera right now, but here are two older pics from the testbench:

blabla · 01-27-2011, 03:31 PM

Lord_Athlon · 01-27-2011, 03:33 PM

What kind of rpm signal is Lh looking for? A 60-2, or a standard 2 pulses per revolution.

01-26-2011, 05:07 PM	#203
klr142 Turbo, what? Join Date: May 2004 Location: OR	It took me 1-2 weeks to be accepted when I originally registered about a month ago or so... For what it's worth. __________________ Kyle - NLMGG: '91 244 NA DD/Track - General Leif: '71 142 Endurance Racecar - The General's FB page - Oregon Volvo Tuners - Died - 2014 S60 T5 - 2006 Elantra DD

01-26-2011, 05:10 PM	#204
linuxman51 BRANDSCHUTZVORSCHRIFTEN! Join Date: Dec 2002 Location: mont, AL	hm. i registered a looong time ago, it was pretty quick back then. __________________ "They bum rushed them in their own crib, they drank all their beer, they partied with their ladies and they left with the trophy" Now with in-house Dyno tuning! Megasquirt Tuning! Plug and play LH 2.4 Megasquirt, now with stealth mode!

01-27-2011, 10:14 AM	#220
ipdown Член на форума Join Date: Jul 2008 Location: Bulgaria, Sofia	950 assembly listing This is disassembly listing of 950 ecu: http://dl.dropbox.com/u/9914656/0280000950.s Browse it in you favourite text editor, preferably with syntax hi-lighting for .asm/.s files. I am using VIM editor. It is commented, and cross references can be searched quickly with vim. It would be nice if there are new findings and to update them somehow.. May be put it under revision control? Or I am wanting too much.. Anyway, it compiles with as31 assembler to the same file as original binary - the only difference are empty gaps of unused memory space, which in the stock bin are filled with 0xFF, and here are zeroes. You can compile it and compare in hex to the original binary file. AS31 assembler can be downloaded from here: http://www.pjrc.com/tech/8051/tools/...-pj3-win32.zip Use the following command line to compile the file: C:\>as31.exe -Fbin 0280000950.s This will create file 0280000950.s.bin in the current directory

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

01-26-2011, 05:26 PM	#206
linuxman51 BRANDSCHUTZVORSCHRIFTEN! Join Date: Dec 2002 Location: mont, AL	moates.net

01-26-2011, 08:18 PM	#212
beepee how hard can it be? Join Date: Sep 2008 Location: Southern MD	Just got home from a nice, long slushy drive home using the 984 bin. And, despite my skepticism, I am a convert. The 984 bin just seems... smoother. I can't put my finger on it. I think the table I labeled as fuel cut is incorrect though... It's hard to safely hit 15psi when the roads are covered in 1.5" of slush I think further XDF development will be focused on the 984 bin... anyone else given it a shot?

01-26-2011, 08:22 PM	#213
klr142 Turbo, what? Join Date: May 2004 Location: OR	Most chips have seemed smoother after I first put them in, but maybe that isn't what's going on here... Maybe it just is that much better.

01-26-2011, 08:26 PM	#214
beepee how hard can it be? Join Date: Sep 2008 Location: Southern MD	Oh I forgot to mention, with lambda on it seemed like the ECU held near 14.7 much more solidly than the 93x... I would do a datalog but all I can log is o2 right now.

01-26-2011, 09:29 PM	#215
orie Doctor Who Join Date: Dec 2004 Location: Folsom, CA	Got myself a programmer, chips and an ostrich. Now to have some fun.

01-27-2011, 03:29 AM	#217
beepee how hard can it be? Join Date: Sep 2008 Location: Southern MD	Ok, i've been gathering the injector constants from all of the stock BINs I have, here's what I got: I think that the injector constants might be 16 bit, or have a multiplier of some sort. The values were obtained using ipdown's Injjector Tuner. "INJ CONST A" starts one byte before at what the program identified as constant 1 and preceeds for 4 bytes. "INJ CONST B" starts at what was identified as constant 3 and also proceeds for 4 bytes.. I think I am on to something; I don't know. Doesn't really shed any light, does it? I can't tell, I keep nodding off at the PC. aaaaaaaahhhhhhhhh goodnight!

01-27-2011, 08:57 AM	#218
linuxman51 BRANDSCHUTZVORSCHRIFTEN! Join Date: Dec 2002 Location: mont, AL	Hm. usually the number itself is a multiplier used elsewhere (for instance, req fuel in megasquirt is derived from injector size and displacement, other ecu's use a set value that's ~ the size of the injectors to factor into the fueling equation). Do you (or ipdown) know where these constants come in to play?

01-27-2011, 09:43 AM	#219
ipdown Член на форума Join Date: Jul 2008 Location: Bulgaria, Sofia	I'll explain how I found these constants. I have built a "stimulator" for LH 2.4, which provides it with all sensors input and rpm signal, and monitored the injector pulsewidth on oscilloscope. Then, having the "source" assembly of 950 ECU I focused on data section and code references to different values from there. Feeding the ECU with constant rpm, load, etc. the above constants changed the resulting pulsewitdth. Reg. bitsize of them, the #4 and #5 (as numbered in InjConst utility) could be 16 bit value, while the others are most likely alone bytes. They could statically scale temp., voltage or load for example, thus changing the pulse width, but still at least one of them is the real injector constant, since I literally tried every one suspicious referenced byte in the data section and these are that affect the resulting pulsewidth. There are another two which also changes injection frequency, but they are probably dangerous to try on a running engine.

01-27-2011, 03:31 PM	#224
blabla Board Member Join Date: Feb 2006 Location: The Netherlands