Dannyhaddon
Jr. Member
Karma: +0/-0
 Offline
Posts: 25
|
 |
« Reply #2565 on: April 09, 2024, 01:58:58 PM »
|
|
|
Yes everything should be shared, it's not like there's many of these cars left to keep things secret
|
|
|
|
|
Logged
|
|
|
|
s60rawr
Full Member
Karma: +19/-1087
Offline
Posts: 170
|
|
« Reply #2566 on: April 09, 2024, 04:35:01 PM »
|
|
|
I strictly got into this for fun and my own enjoyment.
|
|
|
|
|
Logged
|
There is a free flash suite in progres
-Karma Sponge |
|
|
s60rawr
Full Member
Karma: +19/-1087
Offline
Posts: 170
|
|
« Reply #2567 on: April 09, 2024, 04:46:20 PM »
|
|
|
My daily driver for the past 7 years has been a car that does 3.3s 0-100 with stock hardware and just a remap. I don't know how "modern" it is, it's now 12 years old.Not in most civilized countries. Do some 0-100 in the city and most likely you end up without a license.
Let me put it this way, my stock car does around the same 100-200 that some of the 0-100 posted here, but does not matter.
Really now? I don't know where you're doing that shit.
Drag race starts at 0. Lights start at 0. It's like creating a new special (retarded) discipline to show off the one tiny thing you can do decently. "Fastest FWD car with stock turbo, stock intercooler and modified exhaust in the world" or some other stupid highly specific stuff like that.
Grip limit on decent powertrain 0-100 is over 1000hp, not the things you say here. I don't have any issues with grip after 20-30km/h and I have just 650hp.
Maybe if you run dogshit tyres on tiny pizza cutter wheels it's a problem...
Anyway, does not matter.
As BaxtR said very well:
Having "secrets" on 20 year old shit is completely retarded. It's like being proud of wagging your dick around at a micropenis competition.
Cooperation is much more important, because in the end it's a tiny enthusiast community with these old cars - there is no significant money to be made here.
Already 10 years ago when I was tuning these cars there was super low demand for it, because usually the people who buy a Volvo are those that don't care about any tuning. And here are a lot of Volvos on the streets.
I think P3 and some late P1 there is probably some money to be made, but P2 is completely dead from that standpoint.
Do we have any p1 / p3 SBLs? Let's blow it wide open. I have the damos for it I think I have a me9 one but idk wtf to do with it. |
|
|
|
|
Logged
|
There is a free flash suite in progres
-Karma Sponge |
|
|
keichi
Full Member
Karma: +10/-2
Offline
Posts: 73
|
|
« Reply #2568 on: April 10, 2024, 01:33:19 PM »
|
|
|
5120 hack on 2004 S60R Volvo with big ass turbo  Works very well. One interesting conclusion - it doesn't have to be exactly double of stock pressure limit. It can be done with any scale, like 1,5times stock pressure or whatever. Doubling pressure scale results in double loss of precision but no issues because of that so far |
|
|
|
|
Logged
|
|
|
|
|
prometey1982
|
|
« Reply #2569 on: April 10, 2024, 11:29:07 PM »
|
|
|
Works very well. One interesting conclusion - it doesn't have to be exactly double of stock pressure limit.
It can be done with any scale, like 1,5times stock pressure or whatever.
Ты ошибаешься. В некоторых местах, вместо умножения, используется сдвиг. Ты не можешь просто взять и сдвинуть на пол бита. В случае двойного понижения точности, мы просто увеличиваем величину сдвига на один. Либо этот код, который использует сдвиг, нужно изменять на деление. |
|
|
|
|
Logged
|
|
|
|
|
prometey1982
|
|
« Reply #2570 on: April 10, 2024, 11:40:00 PM »
|
|
|
I would love to find pointers on how to read and tune the Aisin in one of my Volvos. Is the VAG stuff related at all and can I find something there that I can use on my Volvo?
Let's all be nice to each other (or at the very least not hostile) and leave the whole value and secret stuff discussion off this topic. The whole market in aftermarket tuning is filled with people wanting to make money, some actually do. Very few of the ones making money respond here, most read and learn and keep their mouth shut. None of them here will tell here how much time and money they invested and how much money they made eventually, so the discussion in my opinion is pointless and takes away from the good stuff happening here.
Я планировал выпустить возможность прошивки АКПП в свободный доступ, но друг разубедил меня. У него был подобный опыт. Это привело к тому, что множество людей начали править прошивки двигателей Митсубиси и убили веру людей в то, что есть нормальные тюнеры. В итоге, поток клиентов упал и новые разработки были заморожены. Рынок прошивок АКПП очень маленький. Хотя, он перешил уже несколько десятков Вольво и Пежо с коробками AW55, TF70 и TF80 в России. И несколько машин с коробками Aisin из других стран. Что бы Дмитрий (prj) тут не писал, но сам он открывает только то, что уже не имеет коммерческой ценности. Прошивки АКПП имеют коммерческую ценность. Например, позволяют не сжигать АКПП на бустапе на тех же французских и японских машинах на TF70. А они с этими АКПП выпускаются и сейчас. |
|
|
|
|
Logged
|
|
|
|
keichi
Full Member
Karma: +10/-2
Offline
Posts: 73
|
|
« Reply #2571 on: April 11, 2024, 02:13:41 AM »
|
|
|
Ты ошибаешься. В некоторых местах, вместо умножения, используется сдвиг. Ты не можешь просто взять и сдвинуть на пол бита. В случае двойного понижения точности, мы просто увеличиваем величину сдвига на один. Либо этот код, который использует сдвиг, нужно изменять на деление.
There are few 1013 constants and 1013 divisions and in DECOS cars (R models and face-lift) there is also additional dppk_w calculation. All of them can be freely scaled. Just arithmetic vars that can be change to whatever you want. Tricky part is just boost sensor scaling DLSGRAD/OFS. ECU is not doing simple Y=GRAD*X+OFS when calculating boost from voltage. If you leave stock boost senor (in case of 2bar Rmodels that may be the case) you can get on with this by just multiplying stock GRAD/OFS. But when you change sensor you have to figure out how ECU calculate boost from sensor voltage (additional multipliers and/or additions) to come up with proper GRAD/OFS. And its different in prelift and postlift models also. If you don't do it you will be off with boost like 0,1bar or something - no big deal but if you want ECU to see true boost and if you want to have true boost in logs then its tricky |
|
|
|
« Last Edit: April 11, 2024, 02:29:34 AM by keichi »
|
Logged
|
|
|
|
|
prometey1982
|
|
« Reply #2572 on: April 11, 2024, 02:32:03 AM »
|
|
|
There are few 1013 constants and 1013 divisions and in DECOS cars (R models and face-lift) there is also additional dppk_w calculation.
All of them can be freely scaled. Just arithmetic vars that can be change to whatever you want.
Tricky part is just boost sensor scaling DLSGRAD/OFS. ECU is not doing simple Y=GRAD*X+OFS when calculating boost from voltage.
If you leave stock boost senor (in case of 2bar Rmodels that may be the case) you can get on with this by just multiplying stock GRAD/OFS.
But when you change sensor you have to figure out how ECU calculate boost from sensor voltage (additional multipliers and/or additions) to come up with proper GRAD/OFS.
And its different in prelift and postlift models also.
С наклоном и смещением показаний датчика давления как раз проблем нет. Там можно сделать деление пополам или насколько нужно, хоть в 1.5 раза. Я же говорю о коде. Вот, выделенные жирным изменения, которые не делают /2 или *2. |
|
|
|
|
Logged
|
|
|
|
keichi
Full Member
Karma: +10/-2
Offline
Posts: 73
|
|
« Reply #2573 on: April 11, 2024, 03:04:35 AM »
|
|
|
С наклоном и смещением показаний датчика давления как раз проблем нет. Там можно сделать деление пополам или насколько нужно, хоть в 1.5 раза. Я же говорю о коде. Вот, выделенные жирным изменения, которые не делают /2 или *2.
No need to touch them. All those fxxx are just multipliers (unit less scalers) and you just need to trace from where they come from and if its related to pressure just scale there. For example: fpvdk = fkmsdk x fpvdkds fkmsdk - not pressure related, ignore fpvdkds - pressure related = (udsl * DSLGRAD + DSLOFS) / 1013 - just scale 1013 division here. 1013 division in ASM look like this: mov r4, pvdkdsl_w
mov r2, #647 ; 1013 division
mulu r4, r2
mov r5, MDH
mov r4, MDL
mov r2, r5
shr r4, #10
shr r5, #10
shl r2, #6
or r4, r2 When you analyze this code (just throw it into Keil and debug line by line and look at whats going on in registers) you will notice that you can just scale #647 variable - x2 for 5120 hack or any other scale you want. EDIT: And this part: udsl * DSLGRAD + DSLOFS its not computed exactly that. In reality ECU does something like (udsl * DSLGRAD + DSLOFS *2) in postlift models. In pre facelift its even more complicated. So when you come up with DSLGRAD/OFF doing simple linear function using sensor datasheet (P(U) for two points) you can leave DSLGRAD but need to divide DSLOFS by 2 - thats the trick here And in pre-lift models there is different "trick" |
|
|
|
« Last Edit: April 11, 2024, 03:10:23 AM by keichi »
|
Logged
|
|
|
|
|
prometey1982
|
|
« Reply #2574 on: April 11, 2024, 03:14:00 AM »
|
|
|
Вот так выглядит код в моем случае: ROM:000B07BA sub_B07BA: ; DATA XREF: ROM:00010FA0↑o
ROM:000B07BA mov r4, fpvdkds_w
ROM:000B07BE mov r5, fkmsdk_w
ROM:000B07C2 mulu r5, r4
ROM:000B07C4 mov r2, MDL
ROM:000B07C8 mov r3, MDH
ROM:000B07CC jmpr cc_N, loc_B07D8
ROM:000B07CE shl r3, #2
ROM:000B07D0 jmpr cc_C, loc_B07D8
ROM:000B07D2 shr r2, #14
ROM:000B07D4 or r3, r2
ROM:000B07D6 jmpr cc_UC, loc_B07DC
ROM:000B07D8 ; ---------------------------------------------------------------------------
ROM:000B07D8
ROM:000B07D8 loc_B07D8: ; CODE XREF: sub_B07BA+12↑j
ROM:000B07D8 ; sub_B07BA+16↑j
ROM:000B07D8 mov r3, #0FFFFh
ROM:000B07DC
ROM:000B07DC loc_B07DC: ; CODE XREF: sub_B07BA+1C↑j
ROM:000B07DC mov fpvdk_w, r3
ROM:000B07E0 shr r3, #8
ROM:000B07E2 movb fpvdk, rl3
Может, конечно, я что-то лишнее изменил, но мне пришлось восстанавливать fpvkd_w, после того, как fpvdkds_w был понижен в 2 раза. Тут нет деления, тут операции битовых сдвигов. Точнее деление есть и оно реализовано через битовые сдвиги. |
|
|
|
|
Logged
|
|
|
|
|
prometey1982
|
|
« Reply #2575 on: April 11, 2024, 03:21:35 AM »
|
|
|
EDIT: And this part: udsl * DSLGRAD + DSLOFS its not computed exactly that. In reality ECU does something like (udsl * DSLGRAD + DSLOFS *2) in postlift models. In pre facelift its even more complicated. So when you come up with DSLGRAD/OFF doing simple linear function using sensor datasheet (P(U) for two points) you can leave DSLGRAD but need to divide DSLOFS by 2 - thats the trick here And in pre-lift models there is different "trick" Ты что-то путаешь. Сколько я смотрел прошивок, нигде такой дичи не видел. Везде были прямые вычисления и дальнейшее использование полученных переменных. Вот тебе пример из прошивки моей машины, как раз рестайлинг 50WRHJ с DECOS: ROM:000D7092 mov r12, DSLOFS
ROM:000D7096 mov r13, DSLGRAD
ROM:000D709A mov r14, udslsum_l
ROM:000D709E mulu r14, r13
ROM:000D70A0 mov r15, MDH
ROM:000D70A4 cmp r12, #0
ROM:000D70A6 jmpr cc_N, loc_D70B2
ROM:000D70A8 add r12, r15
ROM:000D70AA jmpr cc_NC, loc_D70B8
ROM:000D70AC mov r12, #0FFFFh
ROM:000D70B0 jmpr cc_UC, loc_D70B8
ROM:000D70B2 ; ---------------------------------------------------------------------------
ROM:000D70B2
ROM:000D70B2 loc_D70B2: ; CODE XREF: sub_D704A+5C↑j
ROM:000D70B2 add r12, r15
ROM:000D70B4 jmpr cc_C, loc_D70B8
ROM:000D70B6 mov r12, #0
ROM:000D70B8
ROM:000D70B8 loc_D70B8: ; CODE XREF: sub_D704A+60↑j
ROM:000D70B8 ; sub_D704A+66↑j ...
ROM:000D70B8 mov pdsl_w, r12
ROM:000D70BC rets
|
|
|
|
|
Logged
|
|
|
|
keichi
Full Member
Karma: +10/-2
Offline
Posts: 73
|
|
« Reply #2576 on: April 11, 2024, 03:25:16 AM »
|
|
|
Just trace fpvdkds_w where it comes from. You have it not scaled at the beginning but try to "fix" things in the middle of the code flow.
|
|
|
|
|
Logged
|
|
|
|
|
prometey1982
|
|
« Reply #2577 on: April 11, 2024, 03:42:12 AM »
|
|
|
А вот и причина, почему я его не умножил ранее: ROM:000D58C2 mov r7, pvdkdsl_w
ROM:000D58C6 mov r4, #0A1BCh
ROM:000D58CA mulu r7, r4
ROM:000D58CC mov fpvd_w, MDH
ROM:000D58D0 mov r5, MDH
ROM:000D58D4 mov r8, r5
ROM:000D58D6 mov r9, #0
ROM:000D58D8 add r8, r8
ROM:000D58DA addc r9, r9
ROM:000D58DC cmp r9, #0
ROM:000D58DE jmpr cc_NZ, loc_D58E4
ROM:000D58E0 cmp r8, #0FFFFh
ROM:000D58E4
ROM:000D58E4 loc_D58E4: ; CODE XREF: sub_D5840+9E↑j
ROM:000D58E4 jmpr cc_ULE, loc_D58EC
ROM:000D58E6 mov r4, #0FFFFh
ROM:000D58EA jmpr cc_UC, loc_D58EE
ROM:000D58EC ; ---------------------------------------------------------------------------
ROM:000D58EC
ROM:000D58EC loc_D58EC: ; CODE XREF: sub_D5840:loc_D58E4↑j
ROM:000D58EC mov r4, r8
ROM:000D58EE
ROM:000D58EE loc_D58EE: ; CODE XREF: sub_D5840+AA↑j
ROM:000D58EE mov fpvdkds_w, r4
Тут идет умножение на 41404. Поскольку числа там 2 байта, то максимум можно умножить на 65536. Либо добавлять инструкции для умножения fpvdkds_w на 2. В общем версия 50WRHJ сложнее, чем более ранние для хака. Поэтому, пришлось потратить много времени на реализацию. Но все в прошлом, сейчас стабильно, уже на двух своих машинах, езжу на ней. |
|
|
|
« Last Edit: April 11, 2024, 03:44:07 AM by prometey1982 »
|
Logged
|
|
|
|
keichi
Full Member
Karma: +10/-2
Offline
Posts: 73
|
|
« Reply #2578 on: April 11, 2024, 04:14:21 AM »
|
|
|
I didn't spent much time in 50WRHJ but i i just noticed some crazy things Looks like it is very similar to newer MED17 codebase. But what is most interesting, looks like your pdsl_w (actually its called pvdr_w) is already with 5120 hack from factory That way your pvdkdsl_w (called pvd_w)=pvdr_w is also scaled up to 5120 from factory and that means no need to touch 0A1BCh constant. So look like for 5120hack you just don't need to scale DLSGRAD/OFS at all Maybe thats why you must "fix" fpvdkds_w later on in the code because of unnecessary scaling of DSLGRAD/OFS and/or 0A1BCh const. Look like Volvo did some very interesting things on those latest software and 5120 mod should be done different way than all earlier sw. versions. EDIT: I just checked what pvd_w and pvdkds_w are at atmospheric pressure: pvd_w = 12765 (x 0.078125 - 5120 factor) = 997mbar pvdkds_w = 25530 = 12765 x 2 (x 0.0390625 - stock factor) = 997mbar So pvd_w it scaled at 5120mbar from factory. Then some variables like fpvdkds_w also use 5120 scale and other "normal" 2560 scale like pvdkds_w. So more digging is needed probably to figure out what exactly is scaled at 5120 and what at 2560 Here is where 5120 is scaled down to 2560 for pvdkds_w: ROM:000D589C F2 F8 66 F4 mov r8, pvd_w
ROM:000D58A0 E0 09 mov r9, #0
ROM:000D58A2 00 88 add r8, r8
ROM:000D58A4 10 99 addc r9, r9
ROM:000D58A6 48 90 cmp r9, #0
ROM:000D58A8 3D 02 jmpr cc_NZ, loc_D58AE
ROM:000D58AA 46 F8 FF FF cmp r8, #0FFFFh
ROM:000D58AE
ROM:000D58AE loc_D58AE: ; CODE XREF: sub_D5840+68↑j
ROM:000D58AE FD 03 jmpr cc_ULE, loc_D58B6
ROM:000D58B0 E6 F4 FF FF mov r4, #0FFFFh
ROM:000D58B4 0D 01 jmpr cc_UC, loc_D58B8
ROM:000D58B6 ; ---------------------------------------------------------------------------
ROM:000D58B6
ROM:000D58B6 loc_D58B6: ; CODE XREF: sub_D5840:loc_D58AE↑j
ROM:000D58B6 F0 48 mov r4, r8
ROM:000D58B8
ROM:000D58B8 loc_D58B8: ; CODE XREF: sub_D5840+74↑j
ROM:000D58B8 F6 F4 64 F4 mov pvdkds_w, r4
Line 000D58A2 - here is pressure multiplied by 2 (by adding to itself - those compilers sometimes do things in weird way from human pov but we have to belive its more efficient this way So Volvo did 5120 hack (in some part at least) before even it was trendy  |
|
|
|
« Last Edit: April 11, 2024, 04:57:26 AM by keichi »
|
Logged
|
|
|
|
s60rawr
Full Member
Karma: +19/-1087
Offline
Posts: 170
|
|
« Reply #2579 on: April 11, 2024, 07:08:20 AM »
|
|
|
I have 5120 for 10ERHJ ( 00R QHHJ ( EU 04 M66) QKHJ ( US 04 M66 ) GPHJ 05+ m66 Wra72 variant of wrhj for the 06+ TF80 cars wra70 wrhj prometey shared before but don't have logging parameters for and it's a eastern Europe variant I'll never touch Don't need that many variants tbh |
|
|
|
|
Logged
|
There is a free flash suite in progres
-Karma Sponge |
|
|
|
