45 replies to this topic

[presslab]

This is about my paddle shifter project on my GL-10 with the 4EAT automatic. I used a 1990 Legacy TCU because it was easier to modify the code, it uses a much faster 68HC11 processor vs the stock GL-10 TCU which is an old Hitachi type.

The first thing was to graft the Legacy TCU onto the GL-10 harness. It's quite straightforward - I basically matched up the wires using the wiring diagrams. There are a few differences, like the "1 hold" switch, but these things aren't needed once the paddle shift is in place. The Legacy also uses a atmospheric sensor input which I left disconnected. The tach input to the Legacy TCU is 0-5V where as the GL-10 TCU is right off the coil. I initially hooked it straight up and it worked, however I wasn't really happy with it so I used a zener diode and a resistor to clip the signal.


Apparantly the Legacy 4EAT has a stronger high clutch, as initially the 2-3 shift was pretty slow. The transfer clutch has changed as well. To resolve the 2-3 shift problem, I modified the maps in the TCU ROM to make the 2-3 shift faster. This solved that problem.

Initially I modified the maps for the transfer clutch; this made the AWD work again, but it wasn't so great. Sure, it worked as well as it did with the GL-10 TCU but there was a lot of room for improvement. There was a delay in engagement, and it sometimes spun the front tires on takeoff. I decided to upgrade the Duty C solenoid/valve body in my 4EAT to a Legacy unit, along with the orifice plate right behind the Duty C valve body. There were obvious differences in these parts, it seems like it is designed to flow much more fluid volume to the transfer clutch unit. I also replaced the friction discs while I was in there. It's a tad grabby now, but it works much better; I'm back to using the original Legacy TCU transfer clutch maps in place. I'm sure the clutch plates just need to break in a bit, as the cluch was pretty tight when I put it back together.


The TCU was modified by soldering a socket onto the circuit board. I was going to use a different Legacy TCU but it had no holes to solder the socket! I'm not sure what year that TCU was. After the socket is in place, there is a 0-ohm resistor on the bottom of the board that needs to be moved to enable the new socket. The ROM chip I am using is a SST27SF512; this EEPROM is easily reprogrammed without a UV eraser.

The paddles are just bent metal rods that I tack welded to some switches. They work pretty good. I made a small metal frame that attached to the steering column; the switches screw down into that. The paddles use the "manual" and "power" button inputs to the TCU. The paddle code will protect against overrev on downshift, upshift at redline, and downshift to 1st when stopped.

My most recent modification is a DCCD knob. It's a six position rotary switch with FWD, 75%, 100%, 150%, 200%, 4WD settings. It still allows the TCU to calculate the transfer clutch value but it then it modifies it depending on the knob setting; 100% would be stock, 200% would be twice as aggressive as stock, etc. The DCCD knob switch has some 1N4001 diodes wired in series between each pole. The FWD switch pole is attached to ground and the common is attached to the atmospheric sensor input. FWD=0V, 75%=0.5V, 100%=1V, 150%=1.5V, 200%=2V, 4WD=2.5V. If the switch is disconnected the code defaults to 100%.

Attached Files

•  IMG_0402.JPG   29.54K   154 downloads
•  IMG_0400.JPG   39.18K   220 downloads

Edited by presslab, 13 December 2008 - 12:27 AM.

[presslab]

It seems the forum does not allow text file attachements anymore, so here is the code in this post. Unfortunately the formatting is messed up so it's hard to read.

  ; Paddle shift TCU mod for '90 Legacy AWD TCU 31711AA911, v705005

  ; Ryan Press

  ; version 1.1

prevPaddleFlags  EQU $D0

paddleFlags  EQU $D1

paddleGear  EQU $D2

ignorePaddleCtr  EQU $D3

signedSlipRPM  EQU $D4

tempVar   EQU $D6

paddleFlagOn  EQU $1

paddleFlagUp  EQU $2

paddleFlagDown  EQU $4

paddleFlagStopped EQU $8

vehicleSpeed  EQU $14

inShaftRPM  EQU $19

engineRPM  EQU $32

currentGear  EQU $45

shiftFlags1  EQU $4F

gearParm  EQU $BC

powerMode  EQU $51

paddleUpPort  EQU $08

paddleDownPort  EQU $07

paddleUpBit  EQU $40

paddleDownBit  EQU $40

redlineTable  EQU $CDFF

LookupShift  EQU $D99C

CalcVSS1  EQU $D6ED

CalcSol3  EQU $DF46

CalcPowerMode  EQU $DA84

IGNORECT  EQU 8      ; Number of cycles to ignore input from paddles

STOPPEDKPH  EQU 3      ; Vehicle speed <= STOPPEDKPH then we are stopped

 org $F000

 ldaa gearParm

 staa paddleGear

 ldaa paddleFlags

 anda #(~(paddleFlagUp | paddleFlagDown) & $FF)  ; Clear paddle flags

 staa paddleFlags

 ldaa paddleUpPort

 bita #paddleUpBit

 bne GetDownPaddle

 bset paddleFlags, paddleFlagUp

GetDownPaddle

 ldaa paddleDownPort

 bita #paddleDownBit

 bne CheckOnOff

 bset paddleFlags, paddleFlagDown

CheckOnOff

 brclr prevPaddleFlags, paddleFlagUp, CheckOnOff1  ; Check if both paddles pressed, but not previously

 brclr prevPaddleFlags, paddleFlagDown, CheckOnOff1

 bra CheckPaddles

CheckOnOff1

 brclr paddleFlags, paddleFlagUp, CheckPaddles

 brclr paddleFlags, paddleFlagDown, CheckPaddles

 ldaa #IGNORECT      ; Reset paddle ignore

 staa ignorePaddleCtr

        brset paddleFlags, paddleFlagOn, CheckOnOff2

        bset paddleFlags, paddleFlagOn    ; Paddle mode off, so turn it on

 jmp Exit

CheckOnOff2

 bclr paddleFlags, paddleFlagOn    ; Paddle mode on, so turn it off

 jmp Exit

 

CheckPaddles

 tst ignorePaddleCtr

 beq CheckPaddles1

 dec ignorePaddleCtr

 jmp Exit

CheckPaddles1

        brset paddleFlags, paddleFlagOn, CheckRedline   ; Exit if not paddle shift mode

 jmp Exit

CheckRedline

 ldx #redlineTable      ; Get redline rpm from table

 ldb paddleGear

 abx

 ldd engineRPM

 lsld

 lsld

 tab

 ldaa 0,x

 sba        ; compare engineRPM to redline

 bcc NotRedline

 inc paddleGear      ; Redlined, shift up

 ldaa #IGNORECT      ; Reset paddle ignore

 staa ignorePaddleCtr

 bra Exit

NotRedline

 ldaa vehicleSpeed

 cmpa STOPPEDKPH      ; If we just stopped, set gear to 0

 bhi NotStopped

 brset paddleFlags, paddleFlagStopped, CheckUpPaddle

 bset paddleFlags, paddleFlagStopped

 clra

 staa paddleGear

 bra Exit

NotStopped

 bclr paddleFlags, paddleFlagStopped

CheckUpPaddle

 brset prevPaddleFlags, paddleFlagUp, CheckDownPaddle  ; Check if up paddle pressed, but not previously

 brclr paddleFlags, paddleFlagUp, CheckDownPaddle

 ldaa #IGNORECT      ; Reset paddle ignore

 staa ignorePaddleCtr

 ldaa    paddleGear      ; Increment gear but only if gear < 3

 cmpa    #3

 bhs     Exit

 inc paddleGear

 bra Exit

CheckDownPaddle

 brset prevPaddleFlags, paddleFlagDown, Exit   ; Check if down paddle pressed, but not previously

 brclr paddleFlags, paddleFlagDown, Exit

 

 ldaa #IGNORECT      ; Reset paddle ignore

 staa ignorePaddleCtr

 tst paddleGear      ; Decrement gear but only if gear not zero

 beq     Exit

 dec paddleGear

  ; Downshift overrev protection

  ; Check for over-rev, multiply RPM by ratio difference

  ; newRatio / curRatio * inShaftRPM + (engineRPM - inShaftRPM)

  ; use CalcVSS1 instead which calculates inShaftRPM with VSS and currentGear

 ldd engineRPM      ; Calculate TC slip

 subd inShaftRPM

 std signedSlipRPM

 ldaa currentGear      ; Save currentGear

 staa tempVar

 ldaa paddleGear

 staa currentGear

 jsr CalcVSS1      ; Use CalcVSS1 function to get inShaftRPM

 ldaa tempVar

 staa currentGear      ; Restore currentGear

 ldd inShaftRPM

 addd signedSlipRPM

 lsld

 lsld

 sta tempVar

 jsr CalcVSS1      ; Put VSS1 variables back to normal

 ldx #redlineTable      ; Get redline rpm from table

 ldab paddleGear

 abx

 ldaa 0,x

 ldab tempVar

 sba

 bcc Exit

 inc paddleGear      ; Would be over redline, so put it back

Exit

 ldaa paddleFlags

 staa prevPaddleFlags

 ldaa paddleGear

 staa gearParm

        brset paddleFlags, paddleFlagOn, Return   ; Exit if not paddle shift mode

 jmp LookupShift      ; Execute automatic shift lookup

Return

 rts

 

 org $F600

        brset paddleFlags, paddleFlagOn, SetSS3   ; If in paddle shift mode, set solenoid 3

 jmp CalcSol3      ; Execute automatic shift solenoid 3 calc

SetSS3

 bclr shiftFlags1, $10     ; Set shift solenoid 3 off for engine braking

 rts

 org $F650

        brset paddleFlags, paddleFlagOn, SetPwr   ; If in paddle shift mode, set power mode

 jmp CalcPowerMode

SetPwr

 ldaa #1

 staa powerMode

 rts

  ; Driver controlled center diff mod for '90 Legacy AWD TCU 31711AA911 v705005

  ; Ryan Press

  ; version 1.0

dccdInput  EQU $11      ; Previously used as atmospheric pressure input

CalcSolCDuty  EQU $E062

CalcSolCReturn  EQU $E23C

solCtargetDuty  EQU $72

solPeriod  EQU $C005

 org $F800

 jsr CalcSolCDuty

 

 ldx #dccdMap

 ldaa dccdInput

mapLoop:

 cmpa 0,x

 bls mapLoopEnd

 inx

 inx

 bra mapLoop

mapLoopEnd:

 ldd solPeriod ; $4E20 = 100%

 subd solCtargetDuty ; solC = period - ((period - solCduty) * dccdMap / 64)

 ldab 1,x  

 cmpb #$FF  ; if dccdMap is $FF then just lock it up

 beq fourwd

 mul

 lsld

 bcs fourwd

 lsld

 bcs fourwd

 jmp CalcSolCReturn

 

fourwd:

 ldd solPeriod ; Set duty to max (4WD)

 jmp CalcSolCReturn

 

dccdMap:

 fcb $B, $0  ; FWD

 fcb $21, $30 ; 75%

 fcb $37, $40 ; 100% (no change)

 fcb $4D, $60 ; 150%

 fcb $61, $80 ; 200%

 fcb $75, $FF ; 4WD

 fcb $FF, $40 ; DCCD input disconnected (no change)

[Numbchux]

Ho-Ly Crap


I'm officially beyond impressed. I would absolutely LOVE to do this on my XT6. but I definitely wouldn't have a clue on where to start in the reprogramming of the TCU.

could it be set up to have a switch/button for a full-auto mode?

Edited by Numbchux, 13 December 2008 - 05:41 AM.

[ausubaru92]

This is awesome!

I'd like to know more about modifying the TCU as i will be putting an early legacy 4sp auto in my 83 leone sedan in the near future.

Is it possible to modify the shift points and the aggressiveness of the gear changes? I really wanna learn more.

How exactly did you hack it?

[presslab]

Ho-Ly Crap


I'm officially beyond impressed. I would absolutely LOVE to do this on my XT6. but I definitely wouldn't have a clue on where to start in the reprogramming of the TCU.

could it be set up to have a switch/button for a full-auto mode?

Thanks.

If you get serious about this, I could email you my ROM file, although you'd need an EEPROM burner. You'd need up upgrade your transfer clutch if it's the same as mine was.

Yes, it's currently set up to enable/disable paddle mode by pulling both paddles at once. Once out of paddle mode it works automatic just like before.

[presslab]

This is awesome!

I'd like to know more about modifying the TCU as i will be putting an early legacy 4sp auto in my 83 leone sedan in the near future.

Is it possible to modify the shift points and the aggressiveness of the gear changes? I really wanna learn more.

How exactly did you hack it?

Yes, it's possible to modify the shift points, torque converter lockup, line pressure, transfer clutch, etc. Phil over at www.alcyone.org.uk has been doing some work with SVX TCUs, which are similar to the Legacy ones. I used his tcuscan software to interface to the select monitor port on the Legacy TCU. The tcuscan software had all the wrong memory locations though, so I had to figure this out; while the Legacy and the SVX TCUs use the same type microprocessor, the circuit board and the ROM are similar but not the same.

To hack it I downloaded the ROM through the select monitor interface. I then used IDA Pro software to disassemble the ROM to the assembly source code. This is the lengthy part, where I had to figure out what the code meant and what it was doing.

Once I figured out where to hook into the code, I made my paddle and dccd assembly code and compiled it. I then used a hex editor to modify the ROM file so that it would execute my code before it executed the original code. I stuck my code in some unused ROM locations.

Then, the ROM file was burned into a chip and stuck on the TCU. Okay, it wasn't that simple, more like crash and burn. But after a while I got it to work and with some more trial and error the paddle and dccd was working just the way I wanted.

[grossgary]

that is totally awesome. had you ever done this before for a vehicle?

i think it should be mentioned that this is an extremely intense project, beyond the means of most folks. i like that you had to do some welding and assembly code to do this, that's great!

probably one of the most technical and difficult projects ever posted on here. not many people have the skill or time to put into something that in depth on the software side (particularly here). others have done it, but they're usually unwilling to talk about it.

Edited by grossgary, 13 December 2008 - 10:11 AM.

[presslab]

that is totally awesome. had you ever done this before for a vehicle?

i think it should be mentioned that this is an extremely intense project, beyond the means of most folks. i like that you had to do some welding and assembly code to do this, that's great!

probably one of the most technical and difficult projects ever posted on here. not many people have the skill or time to put into something that in depth on the software side (particularly here). others have done it, but they're usually unwilling to talk about it.

Thanks Gary, this is my first time modding any ROM on a car but I'm pretty good at this kind of stuff. I used to do the hardware/software for motion control, servo loops, that kind of stuff.

About 6 months back I quit my job because I was about to have knee surgery and I hated who I worked for; that gave me some time to tinker with stuff. I have a job again but just doing computer repair, it pays the bills for now. I'm thinking of moving to Canada soon but this isn't the best time of year. My citizenship has been approved and I should get my certificate any day now.

[ausubaru92]

So if you can do this to a TCU, is it possible to do the same to an EA82T ECU and modify the fuel maps?

[presslab]

So if you can do this to a TCU, is it possible to do the same to an EA82T ECU and modify the fuel maps?

I'm sure it's possible, but I believe it uses a different CPU; I haven't cracked the cover open on my ECU. It's a lot different, like Macintosh vs Windows. (Although those use the same CPU now.)

[grossgary]

sweetness, what did you do previously? with those skill sets i can't imagine it would be that hard to get a job....if you're willing to relocate of course.

moving to Canada, oh man is that a huge move! what prompted that?

the EA82T ECU can be reprogrammed. i mean really ANYTHING can be done if you have the resources. it might cost $32,000 but it can be done. years ago there used to be a company that did modify EA/ER ECU's. i actually talked to them a really long time ago but don't recall anything about them. though it seems they were out of Australia.

actually there might even be some info on the subaruxt boards about that. someone was playing with the EA chips not too long ago i believe.

[Numbchux]

Thanks.

If you get serious about this, I could email you my ROM file, although you'd need an EEPROM burner. You'd need up upgrade your transfer clutch if it's the same as mine was.

Yes, it's currently set up to enable/disable paddle mode by pulling both paddles at once. Once out of paddle mode it works automatic just like before.

I will probably take you up on that offer eventually. but no way am I delving into anything this serious until the loyale has it's new motor and the 4runner is back on the road (so not until summer sometime, most likely).

I think you've hit the nail on the head with what you've done. full auto shifting, as well as full manual (I'm assuming when you're using the paddles, it doesn't downshift on it's own like it does with the stock selecters). and the DCCD knob is genious!!

did you make the DCCD knob, or use an STi one?


replacing/upgrading the transfer clutches will probably happen next summer, regardless. as there's quite a bit of slip right now.

[presslab]

sweetness, what did you do previously? with those skill sets i can't imagine it would be that hard to get a job....if you're willing to relocate of course.

moving to Canada, oh man is that a huge move! what prompted that?

I was an EE for a company that made automated measuring machines. It was a crappy place, I'm much happier now that I'm not there any more. Canada has less people, beautiful scenery, and lots of outdoor stuff to do; perfect for me. I'm done with the "rat race," it's not worth it.

I'm checking out Vancouver Island, and I've been up there twice. Lately I've been looking for jobs and there aren't many. Maybe I'll go back to school.

did you make the DCCD knob, or use an STi one?

My DCCD control is this switch; a few dollars from the surplus store:
 6327_1.jpg   22.58K   38 downloads

I still need to make a decal with the switch settings. Maybe an Avery label printed on color inkjet would look good. A quick spray of matte clear coat to finish it off.

[Numbchux]

so, I've been thinking about doing this for the last few days, and a problem occurred to me....My XT6 (being a 6-cyl) will have a different rpm signal calibration. so the legacy TCU would not work properly.

any chance this would work with an SVX TCU? I know an SVX TCU can control a legacy 4EAT....and I happen to know someone parting out an SVX....so getting my hands on the TCU and plugs would be quite easy.

[presslab]

so, I've been thinking about doing this for the last few days, and a problem occurred to me....My XT6 (being a 6-cyl) will have a different rpm signal calibration. so the legacy TCU would not work properly.

any chance this would work with an SVX TCU? I know an SVX TCU can control a legacy 4EAT....and I happen to know someone parting out an SVX....so getting my hands on the TCU and plugs would be quite easy.

Hmm, yes you are right. I'm sure it's possible to modify the /4 to /6 in a Legacy TCU. OTOH I don't see any reason why a SVX TCU would not work. There are some SVX ROMs on the www.alcyone.org.uk site, so you wouldn't necessarily need to download your TCU ROM through the select monitor.

It would be worthwile to investigate the XT6 4EAT vs a SVX 4EAT to see if the number of clutch discs is similar, etc. Differences could be worked out by modifying the maps but it adds more work, of course.

[Numbchux]

boy, I know so little about Autos....I think I'm in over my head


will have to do some research. and pull a bunch of parts off that SVX (duty C and stuff....since the TC is shot anyway). maybe this summer the XT6 will get paddle shifters....

[nipper]

techncially it may work. In the real world i dont know. Differnt torque curves, stall speeds, power curves. Will be interesting to find out.

nipper

[grossgary]

since a legacy trans can run just fine in an XT6 (takes mechanical work to make it work of course), i don't see any reason it wouldn't "work".

the RPM signal as you said should be carefully looked at, but might be a simple work around.

[nipper]

i love grand experiments

[Numbchux]

since a legacy trans can run just fine in an XT6 (takes mechanical work to make it work of course), i don't see any reason it wouldn't "work".

the RPM signal as you said should be carefully looked at, but might be a simple work around.

you've seen a legacy 4EAT run with an ER27? Which TCU? I know GLoyale did it in his EA82, but again, 4-cyl rpm signal.

[grossgary]

you've seen a legacy 4EAT run with an ER27? Which TCU? I know GLoyale did it in his EA82, but again, 4-cyl rpm signal.

i do know someone that did it, though i don't know all the details. i'm fairly certain you can swap the harness and front diff and bolt it right up.

i was planning on doing this a year ago along with a front LSD, still have the EJ 4EAT i was going to use, but then i bought a manual trans XT6 so i abandoned the idea.

wouldn't surprise me if there's something over at XT forum about this.

[Numbchux]

I know an EA bellhousing can be put an on EJ 4EAT (Gloyale did it). but I'm mostly concerned about the 4-cyl TCU + 6-cyl relationship.

I've been hunting around on subaruxt.com and subaru-svx. as well as that alcyones site that presslabs linked to (HO-LY-CRAP!!!!).

I think what I'm going to do, is wire EJ TCU plugs (I've got at least 2 sets at my house from harnesses I've stripped). in the XT6 in addition to the existing plugs. This way I can try a few different TCUs and see which matches the shift points of the XT6 best. And, if I find one that works well, then I can start modifying code, plug it in, drive it around the block, and if it doesn't work well, plug the ER27 TCU in there, and drive it to work.



I've been incredibly inspired by this project!! thanks presslabs

[grossgary]

that sounds like a great way to attack this chux, good luck!

i guess my thinking is that if an XT6 TCU ran an EJ transmission, then opposite should work as well?

[presslab]

I think what I'm going to do, is wire EJ TCU plugs (I've got at least 2 sets at my house from harnesses I've stripped). in the XT6 in addition to the existing plugs. This way I can try a few different TCUs and see which matches the shift points of the XT6 best. And, if I find one that works well, then I can start modifying code, plug it in, drive it around the block, and if it doesn't work well, plug the ER27 TCU in there, and drive it to work.

The two plugs sound like a great idea! Did you mean EG33 (SVX) TCU plugs?

I've been looking into the ROM on how the TCU gets the engine RPM. I'm nearly certain I found it. On the Legacy $C00B = 4. On the SVX $C00B = 6. This number is used to determine how many times the RPM interrupt service routine is called before calculating the engine RPM; essentially a divisor. So it looks like it's as easy as changing one byte in the Legacy ROM.

With a 6 cyl engine on the EJ TCU the most noticable thing would be the forced shift at "redline" which would be too low. Also the engine RPM is used for TC lockup and Duty Solenoid A & C calcs.

Edited by presslab, 26 December 2008 - 12:15 AM.

[Numbchux]

The two plugs sound like a great idea! Did you mean EG33 (SVX) TCU plugs?

I've been looking into the ROM on how the TCU gets the engine RPM. I'm nearly certain I found it. On the Legacy $C00B = 4. On the SVX $C00B = 6. This number is used to determine how many times the RPM interrupt service routine is called before calculating the engine RPM; essentially a divisor. So it looks like it's as easy as changing one byte in the Legacy ROM.

With a 6 cyl engine on the EJ TCU the most noticable thing would be the forced shift at "redline" which would be too low. Also the engine RPM is used for TC lockup and Duty Solenoid A & C calcs.

from what I remember (don't have anything in front of me, but have worked on many of both harnesses), the EG33 and EJ TCU plugs are the same. so swapping between those TCUs would be easy. if not, getting my hands on the plugs will be quite easy, so 3 sets of plugs would also be doable

although, if the rpm signal is just the one byte. it might not matter!

and yes, I've been reading up on things, since the shift points themselves are determined based on TPS signal, and speed, for the most part, it wouldn't matter. except when you do romp on it a bit, it'll think it's hit redline a third earlier.

I also want to dig through some FSMs, and compare gear ratios, otherwise when the TCU compares the speed to what gear it should be in, things will work out. (my buddy had that problem when he put a jdm wrx 4EAT in his impreza LX. it worked fine, except the gear ratios were all off, and then when he put the jdm TCU in there, it was expecting a 7k redline, but the ECU would only allow 6200.)

that sounds like a great way to attack this chux, good luck!

i guess my thinking is that if an XT6 TCU ran an EJ transmission, then opposite should work as well?

yea, it's a plan....I'll try to document things as best I can. might even contribute on the alcyone boards, even though most of that is over my head.

Perfect logic, only problem is, that rpm signal. the TCU (as mentioned above) uses rpm for a couple things, and if the TCU is looking for the rpm signal from a 4-cyl, but getting it from a 6, it'll be thrown off. ever seen an EG33 swap hooked up to the stock tach? it'll bury long before it actually hits redline. but again, if that's just one byte that needs to be changed to adjust how it calculates the rpm, than that's easy.