89 AW-4 Trans schematic, programming, and more

This is for all the AW-4 trans guys... and a few Qs for the gurus...

I've done some detailed looking at the TCU - it seems like the kind of thing I've wished for since I was a kid! A totally programmable trans!!! Anyway, this one seems to have a problem keeping the TC locked. I'll explain.

Background: The TCU gets no input from the Renix computer, only ground speed from the trans sensor (pulses) and throttle position from the second TPS, and from the brake switch. It then decides, based on the gearshift switch, which solenoids to turn on or off, using 2 binary positions for 4 possible combinations, plus 1 for the TC lockup. According to the docs, it should lock in 3rd or fourth (poss in second too, but I've not seen this).

The TC will unlock when the brake is pressed, or the throttle closed. It will specifically lock under accel too - opposite the Ford way. However, recently, mine will lock-unlock-lock cycle at highways speeds on flat ground. Especially when it's just cruising along easily (think tail wind).

Well, the TPS gets a stable 4.65v from the TCU as a reference voltage, and it is a variable resistor with the +4.65v and ground across it, and the wiper is the input signal to the TCU. As the throttle is opened, the voltage drops. At idle, it's about a 4v level, at wide open, about 1.5v. I believe there's a series resistor inside the TPS to prevent full + or gnd at the input.

When there is nearly 4v at the input, the TCU uses it as a reason to unlock the TC (throttle presumably closed). I put a DVM on it at several points. Perhaps someone can see something in my findings below. I also have some Q's at the end...

The idle point seems to vary from 3.96v to 4.15v. Things work well when it's closer to the 3.96 level. It only seems to affect TC locking as you will see - shifting is fine. I can't find any correlation in it's variance with any other thing (temperature, speed, length of time running, etc). When the idle (closed) signal is reading at 4.15, the TC unlock happens at 3.75v. When it is in the mood to read 3.96v-4.02v at idle, the TC will unlock between 3.85 and 3.95v. I don't have precise nums at the 3.85 area because it's tough to do all this and drive too...

Anyway, normal operation at 60 mph opens the throttle to give a reading of about 3.7v +- on flat land. In normal decel and accel, it should stay locked. If closed is 4v, then 3.95v is real close, and all is well. However, when the idle voltage seems to creep up at idle, it'll then unlock at 3.75v - the unlock point creeps down! As you can see, that's close to 3.7v, and using cruise will move the throttle just enough to cycle it over and over. Accel a bit harder and it'll stay locked, but I end up at 80mph... I'll let off the pedal once in a while to see what the closed value reads and if it's 4.15, I can count on the lowered unlock point. The reference voltage is a steady 4.65 all the time. In all of this, it should stay locked, only coming out if I let off almost completely and letting the voltage move to close to 4v.

Bear in mind, it's the TCU causing this, so it's not an internal trans issue... The unlock line is driving the actions.

Any ideas why/how this is occuring? Even if the TPS was loose (it isn't), it shouldn't cause the TCU to change the voltage at which it unlocks.

After looking at sites on making this into a manual type shift electronically (all quite crude too), I have some Q's (some unrelated to the problem):

- Is there a schematic or reverse eng. of the TCU or some description of the input and output circuits? (op amp inputs, open collector outputs, diode protection, etc) so I can figure out what options I have?

- What might cause this interaction between the set point for unlock and the idle level? It's inversely related - as the idle point moves up, the TCU moves the unlock point down. One would expect they would move up or down together at least...

- What happens if the TCU turns on the lines for first gear at high speeds? Do the hydraulics protect it and only let it downshift when appropriate?

- Has anyone tried changing the voltages from the sensor, like adding some resistor network? It's only a resistive divider.

- Has anyone tried running with a fixed value, say 3.6v? It would change the shift points a bit, but a moderate value might work for most accelerations, and bypass this whole issue. I don't push it anyway with 250k miles on it It would make the TCU think light accell all the time, but actual road speed and brakes would override when used.

- Anyone see a problem with adding some external logic (another patch approach) to lock it once in 4th for a few seconds, with unlock from the brake or from leaving 4th? I'd build a simple timer/driver and a logic gate for that.

- Anyone gotten into this further than me so far?

Well, thanks in advance,
Skip

The TCU is trying to learn your driving habits but can't because your TBS is dirty? Is the Voltage in to your TPS steady?

Bad info Langer1, the TCU has no ability to learn.

PaulJ said:

Bad info Langer1, the TCU has no ability to learn.

Click to expand...

Bad info, do you know the differance between a statement and a question?

Sorry, bad question.

Ref voltage is rock steady. Not sure about dirt - never took it apart. All TCU contacts cleaned, and the TPS is smooth and constant all the way up and down. The DVM I use will flash if the signal drops or goes out of range even for an instant, and it says continuous signal. In fact, it's very consistent in use, just that up/down correlation between idle signal voltage level and the resulting unlock voltage level (which I think shouldn't change).

Is the TCU all analog?? (not to changethe previous questions) I doubt they would have 2 CPUs in an 89... and it's not that tough a problem - an analog circuit might be all that's needed. Not that it tells me anything without the schematic...

skipc said:

Ref voltage is rock steady. Not sure about dirt - never took it apart. All TCU contacts cleaned, and the TPS is smooth and constant all the way up and down. The DVM I use will flash if the signal drops or goes out of range even for an instant, and it says continuous signal. In fact, it's very consistent in use, just that up/down correlation between idle signal voltage level and the resulting unlock voltage level (which I think shouldn't change).

Is the TCU all analog?? (not to changethe previous questions) I doubt they would have 2 CPUs in an 89... and it's not that tough a problem - an analog circuit might be all that's needed. Not that it tells me anything without the schematic...

Click to expand...

It has to have a DA converter, the TPS is analog and the Speed signal is digital.
I'm not sure what you mean by 2 cpu's but the transmission controller has one and the ECU has one

The speed signal is just a pulse train - which could be used in an analog circuit (think freq-to-voltage). Perhaps shift points work out to be an easy ratio of road speed to throttle for each gear. Only 3 gears to work out (first is default - 0). 3 analog comparisons, each trigger at some ratio, each comparison turned on based on which gear is currently in use. It's easy to get an on/off for some analog level.

However, you're probably right, now that I re-read some of the docs. It seems it will talk to a DRBII tester and has some memory. Of course, I've heard or read that the datastream to a DRBII is analog - perhaps why none of the diagnostic computers have a front end module to connect to it. Either way, it can't be that complex.

If someone has info on the input and output circuits, I could probably "enhance" it... and figure out why it's doing this.

Some of what you are hypothesizing crossed my mind also. Something you may want to do, is investigate the ground circuit through the TCU. There is some kind of primitive choke (in the TCU), in the ground circuit, that seems variable.
A millivolt, standing volt test just before the TCU and just after; in the ground circuit, may provide a rough graph (or idea) of what's going on.
This is just about the voltage variances you mentioned and has little to do with the programing.
A standing volt test in the ground circuit, near the TPS may also correlate to the variances you mentioned. It doesn't take much resistance, to skew an analogue graph, at very low voltages.
The TCU is pretty primitive stuff, one reason it seems so robust, is it's lack of sophistication. But it seems to do a pretty good job, of a stable/consistent power supply.
My training isn't in electronics, so excuse the language. But I have a whole bunch of hours, in cause and affect in control circuits.

It is likely it don’t use a CPU at all.
It’s likely it’s a discreet logic circuit using NAN NOR gates and a clock.

OK, let me ask another thing: Inside the TPS, is it built in a way that one of it's failure modes might be that the resistance element could shift right and left a little or come loose? I've never taken one apart...

I ask, because it seems to have 3 idle values - 3.85v, 4.0v, and 4.15v. It works best at around the 4v one, but the other two cause locking irregularities, even to locking more than it's supposed to.

I have also noticed that it's idle value can change after one accel in an instant, or creep up after several. If I imagine a loose carbon strip as the element in the TPS, this would be consistent perhaps. I have not noticed it creeping up without at least a pedal press.

Anyone tear into a TPS?

With the TPSes working fine otherwise, my next option will be to add a 555 timer and power transistor or relay, and automatically lock a few seconds after going into OD all the time, and coming out on braking or coming out of OD on its own. Therefore -

- No lockup in third (I spend very little time there anyway)
- No lockup during a shift (this is good)
- it will unlock if the TCU downshifts
- it seems a little easier on the tranny this way too

Anyone see problems with this simple circuit after the TCU?

Skip

skipc said:

OK, let me ask another thing: Inside the TPS, is it built in a way that one of it's failure modes might be that the resistance element could shift right and left a little or come loose? I've never taken one apart...

I ask, because it seems to have 3 idle values - 3.85v, 4.0v, and 4.15v. It works best at around the 4v one, but the other two cause locking irregularities, even to locking more than it's supposed to.

I have also noticed that it's idle value can change after one accel in an instant, or creep up after several. If I imagine a loose carbon strip as the element in the TPS, this would be consistent perhaps. I have not noticed it creeping up without at least a pedal press.

Anyone tear into a TPS?

With the TPSes working fine otherwise, my next option will be to add a 555 timer and power transistor or relay, and automatically lock a few seconds after going into OD all the time, and coming out on braking or coming out of OD on its own. Therefore -

- No lockup in third (I spend very little time there anyway)
- No lockup during a shift (this is good)
- it will unlock if the TCU downshifts
- it seems a little easier on the tranny this way too

Anyone see problems with this simple circuit after the TCU?

Skip

Click to expand...

Take a good look at the TCU grounds, I think there on the dip stick bolt.
Good referance voltage don't help if the ground point is flakey.
Also the ECU for the automatic is different that the stick ECU so there is something interconnected.