Arduino Powered TCU Replacement

[wildBill83]

I am considering taking on an arduino powered TCU replacement project. It will not be fast or easy and I was just curious if anyone else had a leg up on open source hardware and software and would be willing to give it a run. This would be a programmable technical part that controls shift solenoids and reads speed and temp sensors and controls up shifts and down shifts, and with the touch of a button you can change your shift points to be more efficient or more powerful. I want to start only on the basics of it and start gathering components over the next few weeks, I'll have a lot of studying to do between now and then to be sure to cover all functions of the TCU and inputs/outputs based on schematic diagrams and shift control programs.

This will only work on OBD I cars as the TCU is separate from the ECU.

Just seeing who is interested in giving up more of their life to the CB7 and who can help with this type of deal. Also if the 16mhz arduino is too slow I can also use a teensy board or Exspresso ARM board.

Also can add an LCD screen for MPG and current gear and paddle shifting capabilities fairly cheap once the boards are working correctly.

Some extra reading:
http://www.yotatech.com/f131/custom-...-shift-247091/
http://www.yotatech.com/f2/paddle-sh...roject-245425/
https://forum.sparkfun.com/viewtopic.php?t=18756
http://arduino.cc/forum/index.php/topic,84997.0.html
http://arduino.cc/forum/index.php/to...html#msg729425
http://arduino.cc/forum/index.php?ac...;topic=69166.0

Basically I could buy a MegaShift from MegaSquirt and mod it too I guess.

[Jarrett]

Here are some threads you might find interesting:

http://www.cb7tuner.com/vbb/showthread.php?t=177881
http://www.cb7tuner.com/vbb/showthread.php?t=179566
http://www.cb7tuner.com/vbb/showthread.php?t=190463

We have somewhat of an OEM upgrade in that a TCU from a '92-'96 Prelude Si with an automatic will work and provide higher shift points. Nothing quite on the level you're discussing, though.

[wildBill83]

Thanks, I read on the KDI product prior but I wanted more options and a lower cost. Not wanting to be in H-Town gridlock revving the RPMS way too high or have to tinker with line pressure or have a rigged box fail on me. I want a computer to do the right things when told to. Also being open source and using open source hardware means it will go from one transmission to another seamlessly (or easily modded) as long as you adapt the harness, just open the programmer and change some variables and drive to test.

I actually used the second thread you posted to get a leg up on my TCU repair, because my Haynes manual doesn't do circuit board repair. Was a good post and very helpful for stock repair. It also contains good info on what I need to program and what kind of inputs and outputs I will need.

[drew12000]

interesting. Please keep android useres in mind haha

[wildBill83]

interesting. Please keep android useres in mind haha

Someone already did an Android to Ardunio kit:
http://www.amarino-toolkit.net/

and an andriod to bluetooth arduino kit:
http://www.instructables.com/id/Andr...ks-to-Arduino/

I can work on that after the basics are done. I plan to start this project in the next few weeks. If anyone is interested in crowd sourcing this idea let me know. I'll share all source and schematics either way but crowd sourcing makes it go faster. Could use extra programming help, hardware help, testing help, and funding help. If you feel drawn to contribute in any way that would be awesome. A dev kit is going to run around $60 so far from what I have priced and that is just a basic controller with no bells and whistles that does what the TCU does, including diagnostic code output to OEM lights.

It is going to use this board for the heart:
http://www.ebay.com/itm/261065366999

and a relay shield similar to this one:
http://www.amazon.com/hacktronics-Re.../dp/B006G0L8QI

along with a custom I/O shield I am starting to lay out on paper and will bread board once I have the $$ to do it and not spend it at the pick-n-pull lot.

I also own and operate a reprap 3D printer and will print the housing once I have the boards assembled and tested, the plan is to use OEM connectors from junk TCUs so that may be the hardest part to come by, but will make it plug-n-play.

Any transmission techs in the house? I'm going to have questions for you.

[Skidd]

Wow. I was just thinking about doing something like this; I was just thinking if there is a way to test the ecu out of the car, this this is close enough that its worth a try! I've been looking to get into some sort of computer controlled electronic stuff, I've got the tools and I have enough spare computer parts that I could get a lot of stuff off of those. I guess I'd be willing to hardware test if you can provide the necessary schematics and software, that is if I still have my car at that point :P

if possible, you'll need a pin-out of the tcu, and what voltages they are, so you don't over or under power anything. good luck!

[wildBill83]

Pinout of the TCU is in the schematic in the FSM, Voltages are all 12V, the relay shield will handle the 4 solenoid high amp items, the custom I/O board will hand things like dummy lights, VSS, gear selector location, and a few other items.

The programming is the part I'm trying to understand yet. I'll start the layout and share it here, then all that is left is programming the tranny logic, and different logic selection. Then add on time. Can anyone describe the logic for our TCU in simple charts? That will make my life easier.

I want to mention that this will be a DIY, not a for sale product, as such there is going to be little support, but I will help you as much as possible if you decide to build one yourself. Nothing is more satisfying than doing it yourself anyway right?

[Skidd]

well.... I know absolutely nothing about arduino programming. I DO however, know a bit of c++, if that helps at all.

This does ring a bell of something that has been done with older cartridge video games, by reading to programming from the, um, well, its a cartridge. I don't know if that technique could be modified to get the programming off of the TCU or not, because then you would need to convert that program to arduino code and then modify all of the data points concerning where you want the better performance... and to top it off, it would be honda specific then. so that idea may not have any merit, but its out there... good luck!

[wildBill83]

well.... I know absolutely nothing about arduino programming. I DO however, know a bit of c++, if that helps at all.

This does ring a bell of something that has been done with older cartridge video games, by reading to programming from the, um, well, its a cartridge. I don't know if that technique could be modified to get the programming off of the TCU or not, because then you would need to convert that program to arduino code and then modify all of the data points concerning where you want the better performance... and to top it off, it would be honda specific then. so that idea may not have any merit, but its out there... good luck!

Arduino is a mix of C/C++ and ASM, you can code in so many ways, I'm looking at other coding from the MegaSquirt project (unfortunately not Honda specific) to convert the logic this week, I'm still waiting on a bit more $$ so I can buy the hardware needed. I have already desoldered a set of TCU plugs.

My current roadblock is how the TCU communicates with the ECU, I'm assuming it is a serial connection, but I'm reviewing schematics to make sure, and I hope it is not a unique form of serial communication.

Here is a programming reference, it is super simple: (I'm not even a programmer but I understand it)
http://arduino.cc/it/Reference/HomePage

So the free parts of this project will be the software and schematic layout, the only part that is not DIY is the 3D printed housing for the electronics, but if given time to perfect my design I will share the 3D model of the housing for free on thingiverse.

Arduinos are stackable electronics so future upgrades are just added to the stack if expansion is needed, here is an extreme example:

And a list of shield suppliers:
http://shieldlist.org/

Initial goal is not for performance shifting, those values can be stored as fixed variables and modified later, the initial goal is just to have a more reliable transmission controller that is upgradable.

PS. You're from Brainerd? Know anyone from Pequot lakes? A goofy guy named Tyler Bacon? I just moved back to TX from MN a few months ago, so I know a few folks out your way.

[wildBill83]

Just found this cool OBD1 Honda ECU tuner Ardunio based of course:
http://crome.software.informer.com/1.5/

I may tinker with it in the future just placing a note here for my future reference.

[cloudasc]

Did you know that the factory honda tcu, has a blank spot on it for an external prom? Did you know that all the OKI processor's used in our ecu's and tcu's, have a "hack" that allows you to pull the original coding that stored within the cpu of the ecu/tcu itself?

Pull the code from the cpu, load it onto a prom, and figure out how to get the tcu to read from the external prom. Once you've gotten that far, its just a matter of knowing assembly language and from there you should be able to find the appropriate sections of the code that control the shift points. Change the values, re-burn to a chip, and voila you should have a "tuned" or "modified" TCU.

Instead of re-inventing the wheel from the ground up, why not just make the existing one better, and save yourself a TON of work. In order to pull off your idea, your most likely going to need to dig into the ecu and tcu communication code anyways. I wish you good luck with this project.


The biggest reason why I have not moved forward with my pt3/pt6 ecu tuning software (besides everything life throws your way) is I do not understand assembly. If I knew where in the file/rom to edit, and any formula's of how the value's are interpretted by the code, I can definately throw something together. My missing link is an assembly programmer. The best my software can offer at this time is adjustable timing maps because of that. One of these days, maybe you, Jarrett, and I should have a local pow-wow. If you would like a place to store/structure/host your notes, your welcome to use my wiki.

[wildBill83]

cloudasc:
I greatly admire your tuning work, but your method only works if you have a fully functional TCU and are willing to risk modding it. Buying a new TCU from Honda is over $400, buying a re-manufactured/repaired TCU runs about $200 and an eBay unit that may already be broken or may fail in a few years runs around $100. I think I can make a more reliable, easier to program, and feature upgradable (paddle shifting mode/economy mode/sport mode selectable with LCD readout) device for much cheaper, and it may be simple enough that others can do it too. I am in no means trying to use the existing TCU at all, and I wish to replace it with a USB programmable device that nearly anyone can tune with just a few steps, even while the car is driving. Piggy backing off of failed or weak TCU electronics is a point of failure in my opinion. I just want to have a cheaper stock/tuner solution that is a one size fits all, and works on any ODB1 Honda transmission and ECU. I appreciate the tips about the OKI chips and will definitely look into it, but can't you only pull it of in HEX format? I'm not sure I can translate the HEX to usable ASM, and that is where it would be faster and easier to recreate the wheel, but I'm not the first to do this and have been doing a TON of research on it. I feel it is feasible and I hope to be able to prototype it soon and get to programming debugging once I assemble the prototype hardware. I hope to have the initial device be the Arduino + 2 other shields (I/O and solenoid control relays) that I will eventually reduce to a single PCB solution and if there is a need I can release the PCB files so others can have them manufactured at will for very cheap. The end goal is to use a single PCB, the stack is just for development.

I will definitely scour your wiki and I find it very kind that you are offering it up to me as a storage site.

If this project sounds over anyone's head and they want to be involved or recieve a unit pre-assembled I will be starting a Kickstarter project to allocate funding to build around 10 units (more if the Kickstarter brings in more interest). Rewards would be in three levels, a DIY kit so interested parties can jump in with a pre-programmed Arduino and all that you need to assemble the entire device, an assembled kit ready to install with sport/stock/economy modes selectable with a switch, and a paddle shift option that includes all previously stated plus the ability to use paddle shifting. Prices are still being worked on but I assure you I am not trying to make a profit so I am still doing what I can to reduce hardware cost.

I will be opening a github for the code and other pertinent information soon.

I think it would be great to meet up with you and Jarrett, looks like I'm not going to work for Haliburton so I may be staying in the Temple TX area for work and not moving to Houston, I have just received a new job offer in the area that pays less, but it is what it is and I will get to work ASAP.

Is it ok if I contact you for some info from time to time?

[cloudasc]

^^ Feel free to contact me from time to time as needed. I'm pretty sure Jarrett can aquire used tcu's from the junkyard for cheap/free, and the cost to replace the caps if your really concerned is less then $5 (all thats really required to "refurbish" a tcu). One of the reasons I suggested the option.

When dealing with assembly, its all stored in binary, however you want to translate that binary byte into to work with is your choice, it doesn't have to be hex, it could be integer.

[wildBill83]

I rebuilt one TCU already that had many problems, I fixed it with about $30 worth of RadioShack components. The problem was more than just the electrolytic caps, I also had to replace the thin film caps, ceramic caps, diodes, a boat load of resistors and one voltage regulator. It was a pain in the ass. I got it to work but had to diagnose the whole board and spent a couple of days making sure all circuits were good. The PCB was even badly damaged, I had to rebuild traces. I should post a photo of it up because it was nasty even after I repaired it. The previous owner kept driving the car after the resistors fried, taking 1/4 of the board. He thought the transmission was stuck, that's why I got such a great deal on the car. Plus I think there was water damage too.

I got another TCU already for spare and a broken one from the Belton Budget-Wrench-A-Part so I can use the connectors for my little tinker project.

No worries there. I am just not a fan of the stock TCU and want options it cannot provide. Plus I'm a tinkerer, DIYer, and a bored engineer. Projects like this and solving problems keep me from feeling like I'm wasting my time/talent.

I'm working on simulating Arduino code tonight because I can't sleep and don't have all of the circuitry figured out, but want to get the logic in code that I do understand for shift/lockup solenoid control. I may run into a limit with tach sample rates but may have found a work around on the Arduino forums.

Yes, most can fix the "S" light or "D4" light with electrolytic caps and some 15ohm resistors, but what if that doesn't fix the problem like my case and they don't want to do major circuit board repair? If they decide to de-solder the plugs from the bad TCU and invest in something more than caps/resistors/time for a more useful solution (and they can solder anyway) why not? Had I been given the choice I would have gone with an Arduino solution myself over extensive repair for a stock solution.

I'm not trying to take away your market at all, I believe there is a market for chipped stock ECUs, and your prices are very fair. But I want more flexibility with ease of programming and maybe others might too.

To change the shift programs it will take all of 3 minutes to hook up USB to the TCU open the Arduino IDE (or Visual Studio or AVRStudio your choice) change a few variables, flash and go. Park and reverse are the only two things I'm not going to be able to control and they are cable operated anyway.

NOTE: I believe the ECU/TCU communicate via TTL, can anyone confirm or deny? I think I may use the TTL access in the ECU on CN2 for communication to the TCU. I see a TX and RX there so send and receive is supported. Happy day. But still looking for the TTL on the stock plug so I don't have to modify the ECU and have a plug and play solution.

[wildBill83]

Parts are on order, some code is simulated, will open what I have up to the public soon.

Still a long way off, Should I update first post with links or just keep going until I get to alpha release and then update the first post?

Current project status:
Electronic component selection: 90%
Automatic transmission basic programming (shift solenoid control): 90%
Automatic transmission logic (stock): 10% (will copy and edit for economy and performance modes)
Paddle shift logic: Not started yet but will be simple solenoid control logic
ECU interface: Still evaluating TTL from ECU or through connector communication
Clean 12VDC power board isolation and decoupling: Researching http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html
Relay board evaluation: In progress
I/O sensor board: Still on paper only, but should be VERY simple once I get it all mapped out

EDIT: also need a name for it yet, was thinking ArduShift, or something more OBD1 Honda specific.