Overriding an Electronic Automatic Transmission

[rfdesigner]

Help...

I should point out I'm a professional electronics enginner.. which gives me enough knowledge to be scared of what I want to do.

I have a 3.2l common rail diesel merc with an electronic auto box (computer controlled).

The auto box is set up to be both 'squishy' and 'revvy'

To run the engine efficiently I want to 'gear up and foot down' But with the auto if you put your foot down you get whiplash and end up in the middle of next week as the car launches.

In a perfect world I'd like to chuck the transmission control module in the bin and start again.. sadly the shear complexity of the the whole car means this would be a truely gargantuan project as I'd need to convince every other system that the gear box was origional.

I could probably get about 50% of potential gains by just over-riding the torque converter clutch (TCC), which won't lock until about 70mph in 5th.

Now I've done quite a bit of research on this , I have all the wiring diagrams and some of the operational docs on the 722.6 autobox. It seems that the system has checks, cross-checks and double cross checks for this.. I have found that it will throw a 'limp-home' code if I just over-ride the signal and make it engage the TCC.. meaning it would then fail to make any further gear changes until the car is turned off i.e. it would be stuck in 5th from the point I hit my over-ride switch if I press the button on the motorway.

(probably.. I haven't yet tried this, but the MB documentation is saying this will happen if it detects the TCC is locked when it thinks it shoudn't be.. I'm not 100% convinced it will know, but that's a whole other story)..

So... has anyone else here tried to convince an electronic auto box to do what they want it to?.. even if you failed.. it would add to the sum of knowledge.

Derek

[saand]

Hi RFdesigner, I am an elec eng as well.
Good to see another person trying this one out. I have gotten control of my TC (Torque converter) and my gear selection.
The methods i have used are written up in the wiki if you haven't come across it yet the links are here
Torque Converter Lockup - EcoModder
Manumatic - EcoModder

my car sounds like its a bit older than yours, the ECU did throw errors however it has not caused any issues.
I now have full gear selection and TC lock up control with just switches and i highly recommend the modification if your after more control of your efficiency and driving.

When i was looking into it, other people fooled their ECU by using a representative load on the ECU lines which went to the transmission. I believe this just involved a resistor on each line which simulated the same resistance as the solenoid coil.

you could test this out just using a switch and resistor on the TC control, cut the TC control line put the resistor on the ECU side and put the switch on the transmission side so you get control. Drive around the block and see if it throws any errors or starts limping home.

[Phantom]

Assuming that you have a newer car the PCM might have to be fooled several ways to get it to work. For TCC it most likely will look for the lock signal and determine how it is functioning by looking at the output shaft RPM vs TCC slip. If you command the TCC to lock below 70MPH but mimic the unlocked signal and the converter slip is read as 0-50RPM it should throw a code.

What I would try to do first is locate a company that sells a tuner solution for your car (I use HPtuners but they only support GM, Ford, and Dodge.) That is how I tune the shift points, TCC lock, shift Pressure and time.

If there is not a tuning solution that you can find for car you could pay for a chip or build a solution like you are thinking. In building a solution it would be best to build a controller to go between sensors, trans, and command wires. In that solution you would need to be able to log the current conditions so you can trick it into thinking it is running normally.

IE. Manually command TCC follow this logic "If MPH<70 then interrupt TCC slip and send represented value of 500RPM slip, other wise pass original signal."

You will need to know the pulses per mile for MPH and probably the same for TCC slip and build a frequency generator.

Hope this helps and does not discourage you.

[rfdesigner]

Thanks.. I found the thread and it's given me encouragement to give it a try.

What I already knew:

TC is unlocked until 70mph in 5th or 65mph in 4th, if you change from 4th to 5th at 65mph the TC remains locked through the change. I've used this to engage lock sub 70mph in 5th.. I then use the cruise control to prevent it unlocking... very tricky on the driving though as you are driving on a bit of a knife edge speed wise.

I've now established the following.

Solenoid DC impedance is 3.0 ohms

When 'locked' the solenoid is pulse at 100Hz, the voltage across it reads as ~0.8V which in a 14V system means 100Hz pulses with a 1:20 mark to space ratio.

The TC is not even partially locked below 65mph.

I already know the TC never fully locks, being MB I don't know if this is because the TC is effectively 2 TCs, use one at low speeds and both at high speeds, or if it is scrubbing the TCC when 'engaged'.. I do know the TCCs wear out, mine did about 50,000 miles ago.

What I do know is that if you put 14V across a 3ohm load you will dissipate 50W, so I'm 99.9% certain I'd blow the solenoid if I just pull the control wire with DC.

Oh.. and with these mercs, the Solenoids are held at 12/14V and then grounded to activate.

Making a 100Hz PWM controller driving a FET to pull the TCC control line will not be hard for me... I'll probably use a PIC micro to do it.

Derek

[rfdesigner]

Originally Posted by Phantom

Assuming that you have a newer car the PCM might have to be fooled several ways to get it to work.

I'm afraid of that.

Originally Posted by Phantom

For TCC it most likely will look for the lock signal and determine how it is functioning by looking at the output shaft RPM vs TCC slip.

I've identified both signals, so that should be do-able.

Originally Posted by Phantom

If you command the TCC to lock below 70MPH but mimic the unlocked signal and the converter slip is read as 0-50RPM it should throw a code.

Yes this is a concern.. The code it throws though is a 'stuck TCC' which is resetable on by turn the car off and on again.

Originally Posted by Phantom

What I would try to do first is locate a company that sells a tuner solution for your car (I use HPtuners but they only support GM, Ford, and Dodge.) That is how I tune the shift points, TCC lock, shift Pressure and time.

Tried that.. no takers.

Originally Posted by Phantom

If there is not a tuning solution that you can find for car you could pay for a chip or build a solution like you are thinking. In building a solution it would be best to build a controller to go between sensors, trans, and command wires. In that solution you would need to be able to log the current conditions so you can trick it into thinking it is running normally.

IE. Manually command TCC follow this logic "If MPH<70 then interrupt TCC slip and send represented value of 500RPM slip, other wise pass original signal."

You will need to know the pulses per mile for MPH and probably the same for TCC slip and build a frequency generator.

Hope this helps and does not discourage you.

This is where this may well end up.. although I'm hoping not. What I can't afford to do is to try and replace the TCM, as that has a serial 'CAM-BUS' input that chats to everything in the car. I don't want to have to fake that.

I can fake TC slip easily, that's just some lines of code in a PIC... I can bench test it no problem and have access to an enviromental chamber.. important for automobile applications. (I run the test departement in a small electronics R&D company.. and usually I'm the one called in to fix the problems on the new chips... the chips that don't yet have datasheets)

Derek

[rfdesigner]

And the latest addition to this little project

I measured the PWM ratio of the TCC signal under 'harsh' operating conditions this morning on the way to work.. i.e. lock the TC, set the cruise control to ~80, then approach the bottom of a long hill at 70, engage cruise.. Use the multimeter set to min/max capture mode, and see what the extremes are that the TCM is prepared to go to.

Turns out 2.2V average, or ~15% on, is as hard as it pushes.

So that's what I'll set my TCC override to, to get the TCC as tight as I dare.

Derek

EDIT: oh additional: I found that the gear changes are made by engaging the appropriate solenoid.. note: changes, not selection, UNDER NORMAL OPERATING CONDITONS ALL GEAR CHANGE SOLENOIDS ARE OFF.

Thus If I want to have a PIC work out if the car is in an approprate gear to have the TCC locked I will need it to constantly monitor all gear change solenoid signals and deduce what gear has now been selected.

Solenoids:

1-2/4-5 (one solenoid does both gears).. disengages TCC during change only.
2-3 disengages TCC (fail safe for 3-4 solenoid confusion)
3-4 toggles TCC

Thus: start in 1st or 2nd, then accelerate:

(optional)
1-2 activates.. TCC disengage during change, reengages if ON.. this time it isn't so it remains disengaged.

(started in 2nd if in Winter mode)
2-3 activates.. disengage TCC (fail safe)
3-4 activates.. TCC toggled to ON
4-5 activates.. TCC disengage during change, reengages if ON

decelerating:

4-5 activates.. TCC disengage during change, reengages if ON
3-4 activates.. TCC toggled to OFF
2-3 activates.. disengage TCC (fail safe)

(if in Summer Mode)
1-2 activates.. TCC disengage during change, reengages if ON.. this time it isn't so it remains disengaged.

TCC modulation from TCM is modulated to control slip to a specific amount, so will need to allow TCM to modulate faked slip via it's TCC signal. Sounds complicated, but in fact it's easy.

Oh, and the TC is a single fluid clutch + mechanical clutch with friction plates that it allows to slip under 'lock' hence my inclination to push the control signal as far as I dare and hopefully reduce slip to zero... the TC has worn out once already.. that's with the MB setup.

[Tygen1]

I set up some switches to control my shift solenoids. If I start to use them, I need to keep using them until I turn off the car because the PCM forces limp mode. Essentially I am cutting off the PCM's comunication and just sending twelve volts to what ever solenoids need to be active for each gear change. What I found is that the pcm's shifting was pretty much the most effcient for my car. Now if I could lock the TCC in second gear then I could probably improve over the stock pcm shifting, however my TCC is not very robust and it's too much of a risk because I'd have to accelerate pretty briskly to take advantage of the locked TCC which could kill it.

[Thymeclock]

Originally Posted by rfdesigner

TC is unlocked until 70mph in 5th or 65mph in 4th, if you change from 4th to 5th at 65mph the TC remains locked through the change. I've used this to engage lock sub 70mph in 5th...

Did you mean to say 70 or 65 Kph rather than MPH?

On most cars the TC locks up at around 40 MPH. The ones that lock up at speeds of higher than 45 MPH are a PITA. But at 70 MPH? I don't think so.

[comptiger5000]

My grandmother has an older (88) Jag with an extremely delayed lockup. It doesn't lock the TC until 4th gear and almost 60 mph! However, most lock up much, much earlier. Jeeps like mine lock up between about 36 and 50, depending on the engine and gearing.

[rfdesigner]

Originally Posted by Thymeclock

Did you mean to say 70 or 65 Kph rather than MPH?

On most cars the TC locks up at around 40 MPH. The ones that lock up at speeds of higher than 45 MPH are a PITA. But at 70 MPH? I don't think so.

It's 70MPH, in 5th.. Or ~2000rpm... this is the UK, we may know kph from touring Europe, but we use MPH here.

No mistake.. the Control signal is abscent below 65MPH where is drops out.

Derek