DIY TCM Ideas

[robertisaar]

so, me being a bored and poor bastard and potentially wanting the ability to run an automatic transmission with either a PCS or more than 4 gears or DCT or CVT or whatever, i've decided that i'm going to build my own TCM. i haven't decided if i'm going to modify code in an existing PCM(knowing me, a 16149396) to deal only as a TCM, or to use something like an Arduino.

basically the PCM would only be used as an ECM, but send signals to the TCM for shift points(or maybe the TCM would send signals to the ECM to let it know it's shifting/locking the TCC???). that way both computers would know what's going on at all times so that they can act appropriately(ECM applying TCC locked spark retard when the TCC is locked by the TCM)

kind of like a slightly less sophisticated 93-95 flash PCM, except without all of the unnecessary and confusing crap added in.

on the arduino side, there are a few different versions of it floating around, i think the least capable unit that will work as a full fledged TCM would be a Mega 2560... the Uno ALMOST fits my requirements, but the 2560 more than exceeds it. too bad there's nothing in between them?

anyways, the mega has a 16 channel 10-bit A/D converter(6 on the uno), 15 8-bit PWM outputs(6 on the uno), 33 digital i/o pins (8 on the uno), 6 interrupts (2 on the uno), 256KB of program space (32 on uno), 8KB of RAM(2 on uno), and 4KB of EEPROM(1 on uno).

the uno costs ~$30, while the mega is a little under $60.... double the price, a lot more than double the hardware. of course, stuff like transistors would need to be added to drive solenoids and stuff, but it's still looking uber-cheap for a capable MCU.



anyways, automatically the uno was disqualified for 1 thing: lack of interrupts. at least 3 will be needed for what i have in mind..... engine speed, trans input speed and VSS.








double anyways: anybody else at least interested in this concept?

[SappySE107]

I would like to see it happen. I don't have an auto trans to test it with right now, but have given serious thought about an auto in the MGB.

[robertisaar]

luckily, a lot of the GM stuff runs on the same principle.... 2 shift solenoids, 1 input speed sensor(for torque converter and clutch slip tracking), 1 VSS, 1 PCS, 1 TCC(some have 2, but at least i understand why they use 2), a temp sensor and that's about it.

wouldn't know about non-GM stuff though, i think the 4 speed taurus transmissions use 3 shift solenoids, no PCS, weird stuff.... but it's not like those are a priority. i think the big ones that anyone here would be interested in would be 4T65, 4L60/4L65, 4L80, 4T40/4T45, 4T80..... and probably in that order as well.

[bszopi]

This is like my first post in probably 2-3 weeks. Finally getting caught back up on reading the forum (moved, or well, moving, and got a new job). Anyway, I'm interested. And if you want to know, I actually had something like this in the works many years ago on the Beretta. Basically had the stock flash PCM for the tranny and was using a '9396 (or whatever you said, been too long and can't scroll on my phone) for the engine only.

Brad via Tapatalk

[robertisaar]

at least the 4T60E is easy to control, just (up to) 4 solenoids, none of which that should cause low-mileage failure if commanded wrong. easily done though, since they share so much with the 4T65, just need to not worry about the PCS, though TCC slip (actually that's a mislabel from GM, BTW, is really TCC/clutch slip in the case of the 4T60E(and early 4L60E) since they lack an input speed sensor) could still be calculated and possibly disable the TCC should it start to become an issue.

[pocket-rocket]

Didn't brian89gp do something like this year's ago? I'm not sure what he used to control it but I remember something about a manual gear select mode and a programmable microcontroller. This should be a cakewalk for you, Robert.

Sent from my HTC One X using Tapatalk 2

[robertisaar]

i seem to remember microswitches on the shifter? or maybe that was someone else's project?

anyways, it should be relatively simple, the arduino programming language is REALLY user friendly, unlike a few other microcontrollers i've tried to play with. i should be able to just setup a 160Hz loop in the program and do something similar to what GM did with major and minor loops. since this is controlling just a transmission, a lot of stuff can be done a lot more often than if trying to control an engine and transmission. instead of shift points being checked every .1 seconds, that could be dropped down to every .05 or .025, PCS could be updated 80 times per second.... just in general, LOTS of possibilities.



so what's the most advanced transmission anyone would be interesting in playing with? the GM 6T70 or something like it? figure if i generate code for the most advanced/complex unit, the less complex units can probably run directly off of the same concepts, just with less gears/etc.

then we get to stuff like paddle shifters/performance shift modes/etc..... the fun stuff.

[robertisaar]

a quick run through wikipedia turns up this:

4L30
4L60/65
4L80/85
5L40/50
6L50/45
6L80/90

4T40/45
4T60/65
4T80
6T40/45/30
6T70/75



am i missing any potentially popular GM applications?

[robertisaar]

and dug up more info......

4L30: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 PCS, 1 temp sensor (integral TCM)
4L60/65: 2 shift solenoids, 1 VSS, 1 TCC, 1 TCC PWM(95-up), 1 PCS, 1 temp sensor, 1 3-2 shift solenoid
4L80/85: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS, 1 PCS, 1 temp sensor
5L40/50: 3 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS, 1 PCS, 1 temp sensor
6L50/45: ???
6L80/90: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS. 5 PCS, 1 temp sensor (integral TCM)

4T40/45: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS, 1 PCS, 1 temp sensor, 1 brake band solenoid
4T60: 2 shift solenoids, 1 VSS, 1 TCC, 1 TCC PWM(optional), 1 temp sensor (optional)
4T65: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS, 1 PCS, 1 temp sensor
4T80: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS, 1 PCS, 1 temp sensor
6T40/45/30: ???
6T70/75: 2 shift solenoids, 1 VSS, 1 TCC PWM, 1 ISS, 5 PCS, 1 temp sensor (integral TCM)



the integral TCM ones are an issue... since the TCM itself seems to be a significant portion of the valvebody as well. so i'm not really sure what to do about those. may have to google around and see if anybody has just opened it up and replaced parts.

[robertisaar]

http://www.gmmodernmuscle.com/forums/viewtopic.php?p=10803

hmm.... based on the first pic in the thread, if the circuits are THAT easy to get to, it wouldn't take much at all to just completely depower the stock TCM via cutting wires/traces. however, it only uses a 16 pin connector, so that would either need to be replaced somehow or additional wiring ran through somewhere else, then all of the circuits could be controlled via an external TCM.

then bash head against wall trying to figure out why GM used 5 pressure control solenoids and 2 shift solenoids to make 6 gears, instead of 3 shift solenoids and 1 PCS. :/

i get the feeling it's due to the clutch/clutch shifting, but still. anyways, i imagine it wouldn't be too difficult to figure out what does what and roughly how it should do it with a scope and either a simulated engine/trans or somehow running a bunch of extra wires through it's case with a real one.

[TurboGTU]

http://code.google.com/p/opengauge/w...DuinoInterfaceWhat about using Audino and elm to communicate with the 6t70/75 TCM rather than wiring the Audino directly to the Trans valvbody. Maybe wire the Audino Inputs to the engines TPS, RPM signal and MAp and translate that to the elm and make the elm send the right signals to the TCM . This would be for Pre-07 cars that don't have CAN.

[robertisaar]

IIRC, GM started implimenting CAN around 2002, and i think everything that used the 6T70/75 had an implimentation of it.

as much as i'd like to not have to open it up and remove internal stuff, i have little to no information about the signals the internal TCM would want to see, i would need a logic analyzer and a real car that has one in a factory application to test it in.

or pretty much every module from a vehicle, then simulate it on a bench.


either way, it's going to be complex.

[TurboGTU]

search for 6t70/6t75 the future is now. Theres 5 parts to it from Gears magazine by Steve Garrett. That will give you some insite to this trans. Lots of reading.

[robertisaar]

just read them, parts 1 and 2 are pretty uninformative as to how it functions, but parts 3, 4 and 5 pretty much did all of the guesswork for me.... now i know ~90% of what i need to to control it manually, but actually commanding the internal TCM is a mystery.

i imagine the 6L stuff would be setup very similarly.....

[robertisaar]

after some PMs with Brian, i'm thinking it will probably be easier to do this with a 7730 and reuse a bit of the GM existing code, but otherwise, use it as a dedicated TCM. i had considered using an Arduino as an engine controller as well, but there is at least one significant issue i've come across that will prevent it: the way the arduino is setup, there aren't any internal interrupt timers left to schedule things like sequential injector timing and COP/CNP ignition... so that's kind of depressing, i was hoping i would be able to control a LFX or something similar with it since it would be nice to utilize the benefit of really high compression, direct injection, 4 cam VVT..... all kinds of fun stuff. anyways, all of the timers are basically used up by the PWM output system....



so, while arduino ECM seems to be relatively pointless, and arduino TCM is certainly possible... the 7730/similar should be more than capable of playing with even the 6 speed autos. can even set it up to do some extra stuff that a TCM wouldn't normally do.