Gearshift Maps

[b3lha]

I've been comparing early and late model TCUs to find the differences. I have found a change to the "Overheat-Stick2-Gear2-Lock" map for the Torque Converter. This map governs TC lockup when the stick is in position 2, the transmission is in 2nd gear, and the ATF is too hot.

When the throttle is less than 3%: Early models will not lock the TC. Later models will lock it above 50km/h.

This applies to JDM and UK models. I haven't investigated whether the same applies to US models.

It's a very specific and uncommon set of circumstances, I can't really imagine what Subaru were trying to fix.

[longassname]

Hey Phil,

Speaking of JDM and UK TCUs I think you forgot to post the rev limit locations for them. If you did I just didn't see them or remember seeing them. Here are the locations and what I've been changing them to. Thanks for the basecode by the way. I've shipped a UK ROM to a guy in Australia so we should get feedback on that soon. Do you have your car back on the road yet? I can send you a bin to save you the copy and paste if you want to try my maps.

JDM
D95F=68 : increase in rpm limit for 1-2 shift
D960=62 : increase in rpm limit for 2-3 shift
D961=62 : increase in rpm limit for 3-4 shift
D963=68 : increase in rpm limit for 1-2 shift
D964=62 : increase in rpm limit for 2-3 shift
D965=70 : increase in rpm limit for 3-4 shift

UK
d7C0=68 : increase in rpm limit for 1-2 shift
d7C1=62 : increase in rpm limit for 2-3 shift
d7C2=62 : increase in rpm limit for 3-4 shift
d7C4=68 : increase in rpm limit for 1-2 shift
d7C5=62 : increase in rpm limit for 2-3 shift
d7C6=70 : increase in rpm limit for 3-4 shift

[b3lha]

Thanks for posting that Michael. I don't have my car back on the road yet, but I some good leads on the parts I need.

I tried to install one of my modified TCUs for somebody last weekend and it wasn't working right. I think the TCU software crashed. No power light, wouldn't change out of 2nd, sol c binding, and occasional flashes on the AT OIL TEMP and DIFF LOCK lights.

Rather embarrassing for me, but he was very understanding and took it in good humour. I'm going to check my soldering carefully, reflash the chip and try again.

He let me take a copy of his late model UK TCU, which I've posted on my website. There are some code differences to the early model, presumably bug fixes and improvements. It might be a better base code for future work.

[longassname]

Does your programmer verify after it writes? I've seen some friends who do their own tuning who use cheap programmers have terrible problems. In fact I know a lot of people who use cheap programmers in addition to having problems with the timing routines which are handled in software make a practice out of writing all their ROMs twice to make sure their hardware doesn't leave any of the bits borderline.

My BP microsystems device programmer has never failed a write but if it did I would know about it. It verifies twice after writing (writing and verifying together takes 3 or 4 seconds total for the devices we are using). They recently dropped their support of the parallel port models which just means they aren't updating the software with new devices. If you keep your eyes open on ebay you may be able to find one for a good price and be way better off than with a cheapy that just doesn't have the real goods inside.


I'll switch to the new UK base code before I send any more out. Thanks

Originally Posted by b3lha

Thanks for posting that Michael. I don't have my car back on the road yet, but I some good leads on the parts I need.

I tried to install one of my modified TCUs for somebody last weekend and it wasn't working right. I think the TCU software crashed. No power light, wouldn't change out of 2nd, sol c binding, and occasional flashes on the AT OIL TEMP and DIFF LOCK lights.

Rather embarrassing for me, but he was very understanding and took it in good humour. I'm going to check my soldering carefully, reflash the chip and try again.

He let me take a copy of his late model UK TCU, which I've posted on my website. There are some code differences to the early model, presumably bug fixes and improvements. It might be a better base code for future work.

[b3lha]

Originally Posted by longassname

Does your programmer verify after it writes? I've seen some friends who do their own tuning who use cheap programmers have terrible problems. In fact I know a lot of people who use cheap programmers in addition to having problems with the timing routines which are handled in software make a practice out of writing all their ROMs twice to make sure their hardware doesn't leave any of the bits borderline.

Yeah, i verified the chip. The EPROM is fine.

Looks like I damaged a track on the board when I removed the original ROM. One of the address lines. I think next time I'll try moving the resistor instead of unsoldering the chip.

[longassname]

Sounds like a plan. You should be able to fix that tcu with a piece of wrapping wire or solid core phone wire depending on the size of the track broken. Just remove some solder mask on both sides and solder a piece of wire over the breach.

[b3lha]

Originally Posted by TomsSVX

I have a ton of TCU's here if anyone needs a spare, pay shipping and its yours.

Tom

Tom,

If this offer is still open, I'm looking for two TCUs. If you let me know the shipping cost to London, UK then I'll send you a paypal.

Phil.

[longassname]

If you want to route them through me I'll socket them for you.

[b3lha]

Originally Posted by b3lha

If you've followed me so far then you will realise that at low speed and throttle, it takes a faster kick to engage power mode. At higher speeds, and with more throttle, a slower kick will do it.

Just reading back through my notes. I'm wondering if I got this the wrong way around?

It would make more sense that the power mode is engaged by a slower kick at low speed/throttle or a faster kick at high speed/throttle? That's why it's easier to trigger power mode if you lift off first.

At high speed+high throttle, it takes a very fast kick to engage power mode. If you lift off first, (high speed+low throttle) a slower kick will do the job.

[longassname]

I don't know. You're way out ahead of me on the tcu now. In fact a couple days ago I sat down to change the temp that the atf temp warning light comes on and apparantly got it wrong. It looked to me like pj2 is the output to the temp warning light and comparing c025 to the output from the temp sensor (looked like ad2) in the calculate atft routine set or cleared it (bit two of bitmask stored at 0010 then moved to 1031). Changing that value didn't work though so I'm going to wait till I get a chance to interogate the tcu in operation before I go back and look for my mistake.

[b3lha]

Originally Posted by longassname

I don't know. You're way out ahead of me on the tcu now. In fact a couple days ago I sat down to change the temp that the atf temp warning light comes on and apparantly got it wrong. It looked to me like pj2 is the output to the temp warning light and comparing c025 to the output from the temp sensor (looked like ad2) in the calculate atft routine set or cleared it (bit two of bitmask stored at 0010 then moved to 1031). Changing that value didn't work though so I'm going to wait till I get a chance to interogate the tcu in operation before I go back and look for my mistake.

I agree with your analysis.

I think C024 is the "on above" temperature, C025 is the "off below" temperature.

On above 244 degrees (0xF4), off below 238 degress (0xEE).

I don't know much about transmissions, but it would make sense to me if the "warning light" limits matched the "overheat mode" limits: 233 degrees and 230 degrees (hardcoded at 0xE200). That way the light tells you when the trans is using the overheat shift mode.

[longassname]

ah, someone else finding my mistake..that's even better thanks

Someone asked for 215 which sounds reasonable to me. I was thinking maybe even a bit lower. It is a warning after all. If we set it to say 205 then we'll actually be warned if we are getting into temps we don't want to be in. 230 is more like omg quick stop the transmission from destroying itself. Ideally you want to run at 176 all day long.

[longassname]

are you sure the temp is in F?

I'm looking at it now and it's it's just the ones complement of the da 2 register unless the value is below 14h in which case it defaults it to the value stored at c062.

The MC68HC11G5 datasheet says the da operates from 0 to 5v (0 to 5v range taken from sensor data from fsm) with 5v resulting in ffc0 in the register and 0v resulting in 0000.

The FSM says:
20c=68f=3 to 3.5v
80c=176f=1 to 1.3v
(note 2 volts change = 60c = 1 volt per 30c in normal operating range of sensor, theoretical range then is 0 to 150c)

case 0v = max temp of sensor:result ff is stored for atft
00 in ad register
ones complement of 00 is stored to atft = ff

case 5v = min temp of sensor:result default value e2 is stored for atft
ff in ad register
ones complement of ff =00
E1D5 ldaa LC062 (e2 loaded into acc a from c062)
E1D8 bra LE1E5
E1E5 LE1E5: staa *ATFT (e2 stored as atft)

case 3v = known to be about 20c = 68f from fsm: 66h=102 is stored for atft
99 in ad register
ones complement of 99 = 66h = 102 decimal
E1DD suba *ATFT (66-66= 0)
E1DF adda #0x02 (0 + 2 = 2)
E1E1 asra (00000010 becomes 00000001) (same as halving)
E1E2 asra (00000001 becomes 00000000) (halving again)
E1E3 adda *ATFT (0 + 66 = 66)
E1E5 LE1E5: staa *ATFT

case 1v = known to be about 80c = 176f from fsm: cc = 204 is stored for atft
33 in ad register
ones complement of 33 = cc = 204 decimal
E1DD suba *ATFT (33-33= 0)
E1DF adda #0x02 (0 + 2 = 2)
E1E1 asra (00000010 becomes 00000001) (same as halving)
E1E2 asra (00000001 becomes 00000000) (halving again)
E1E3 adda *ATFT (0 + 33 = 33)
E1E5 LE1E5: staa *ATFT

[b3lha]

Originally Posted by longassname

are you sure the temp is in F? I wasn't about to figure out the math resulting from the odds complement and all those compares but I was thinking it was value - 127 = degrees C.

No. I don't know for sure. Degrees C would certainly make more sense to me.

I originally estimated it by reading the value when the car had been parked for a couple of days. Obviously the ATF temp should approximately match the engine coolant temp in that situation.

[longassname]

woops, you were responding while I was editing..I went ahead and traced out what sensor readings result in what being stored. It doesn't look like either of our first guesses would pan out.