41TE "Bullet Proofing"

[RoadWarrior222]

It should be the same as the engine computers though, figuring out one sensor is off by the other sensors, if you feed them all a matching set of crap it shouldn't know the difference.

[Vigo]

The computer learning is primarily concerned with not killing the trans because improper apply and release timing on the holding elements on a 604 will result in a flare or bind which both wear the clutch elements very quickly. The computer times how long from solenoid application to actual change registered on the ISS and OSS so that it can start its applies and releases so that they all finish in synchronicity.

As much as it has been talked about, i dont think the stock controller 'learns' anything at all about 'driving style'.

[bakes]

The computer learning is primarily concerned with not killing the trans because improper apply and release timing on the holding elements on a 604 will result in a flare or bind which both wear the clutch elements very quickly. The computer times how long from solenoid application to actual change registered on the ISS and OSS so that it can start its applies and releases so that they all finish in synchronicity.

As much as it has been talked about, i dont think the stock controller 'learns' anything at all about 'driving style'.

I would have to disagree as my my experience a Chrysler transmission tech there was fuzzy logic at play most notice on the early TCM's software i have seen where two different drive sharing would screw the shift schedules and hardness of the shift maybe it it was looking at different tables for each drivers habits and it was getting lost but that was then this is now

[RoadWarrior222]

As much as it has been talked about, i dont think the stock controller 'learns' anything at all about 'driving style'.

Maybe not beyond initial "relearn", and I think that's probably just figuring out fluid and friction state.

- - - Updated - - -

My Escort's transmission was rather annoying for that, I'd drive for economy and it would learn it's way into long draggy granny shifts, annnd the mileage would drop, so I'd have to beat on it once a week to make the shifts firm again.

[5DIGITS]

I would have to disagree as my my experience a Chrysler transmission tech there was fuzzy logic at play most notice on the early TCM's software i have seen where two different drive sharing would screw the shift schedules and hardness of the shift maybe it it was looking at different tables for each drivers habits and it was getting lost but that was then this is now

Hello.
The TCM does not learn driving habits.
The TCM's primary goal is to avoid over-lap and correctly adapt the handshaking events of the apply clutch with the release clutch while adjusting for fill volume.
The fill volume portion of learning is whats adjusted during a Quick Learn while the duty cycle tables for torque phase and speed change are simply returned to their nominal default values.
The shift schedules are predefined and are not adaptable in any way.
When the clutch friction coefficients are changed due to upgraded material or machining of the reaction plates, the initial default values will be incorrect for the application.
This typically causes over-lap (shift tug) until the TCM adapts the learned parameters.
ALTHOUGH, these tables for the torque and speed change duty cycles have limits and are only permitted to learn to the 'rail' and will stop.
Once this occurs, the fill volumes will be adjusted to attempt to bring the DC tables back within a reasonable range.
If the clutch capacity has been significantly increased, the TCM will be unable to compensate and poor shift quality is the imminent result.

In short, the adaptives within the TCM target clutch capacity and adjust for wear rather than driver habits.

[Reaper1]

Huh...well...I can say for certain that my '90 Daytona with the A604...it would shift nice and crisp as long as I got in it a bit a few days a week. If I didn't the shift quality would suffer and there was a noticeable difference in the way the car drove. I have several friends that will attest to this as well.

So, knowing that...how is the TCM using wear adjusting strategies in such a way where it makes this kind of driving experience difference?

[RoadWarrior222]

It could be the actual frictions themselves, glazed vs scuffed up.

[Vigo]

The TCM does not learn driving habits.
...
In short, the adaptives within the TCM target clutch capacity and adjust for wear rather than driver habits.

Thanks very much for posting. It's been too long that butt-dyno impressions and unfounded opinions have dominated the discussion about a604s.

I for one am ready to see real info and progress!

[Aries_Turbo]

I have been trying to put together a package that would benefit the over-all community.
There are items like clutch volume, DCt and DCa controls that are waaaaaay off target for what the intended use is for these transmissions.

yes please.

i now have a 07 town and country with the 4 spd for family use and a slightly damaged (just the case) 01 4-speed (that was originally mated to a 3.8l) sitting in the garage and would love to put a 4speed behind the 2.4L in my garage.

the controls are definitely slowing those plans down.

Brian

[t vago]

I have emailed vigo with a zipped copy of my current draft of the 1989 TCM disassembly. I have not been able to analyze the code itself in any great detail, as I have had to create my own disassembler from scratch. I figure this was quicker than trying to adapt an existing disassembler out there to what I needed it to do.

[Force Fed Mopar]

IIRC, Shelgame had a disassembly of the TCM. What he ran into was some routine that isn't with the rest of the programming that is stopping him from being to fully figure out how it works. Some hidden bit of code.

- - - Updated - - -

Hello.
The TCM does not learn driving habits.
The TCM's primary goal is to avoid over-lap and correctly adapt the handshaking events of the apply clutch with the release clutch while adjusting for fill volume.
The fill volume portion of learning is whats adjusted during a Quick Learn while the duty cycle tables for torque phase and speed change are simply returned to their nominal default values.
The shift schedules are predefined and are not adaptable in any way.
When the clutch friction coefficients are changed due to upgraded material or machining of the reaction plates, the initial default values will be incorrect for the application.
This typically causes over-lap (shift tug) until the TCM adapts the learned parameters.
ALTHOUGH, these tables for the torque and speed change duty cycles have limits and are only permitted to learn to the 'rail' and will stop.
Once this occurs, the fill volumes will be adjusted to attempt to bring the DC tables back within a reasonable range.
If the clutch capacity has been significantly increased, the TCM will be unable to compensate and poor shift quality is the imminent result.

In short, the adaptives within the TCM target clutch capacity and adjust for wear rather than driver habits.

So if we can figure out how to change the shift timing for firmness and shift points, we should be able to leave the logic part alone and it will still adjust for wear?

[RoadWarrior222]

a slightly damaged (just the case) 01 4-speed

BTW, JB Weld holds up on a case... as long as you're anal about cleaning oil/grease off. Tested on A670, exposed to ATF, TDH, Lucas, oil and probably a splash of brake fluid, windshield wash and coolant from on top, hosed off with brakleen, hit with WD-40... normal crap that happens to things in engine bay.

[Vigo]

20150313 - 89 TCM.zip

[5DIGITS]

Huh...well...I can say for certain that my '90 Daytona with the A604...it would shift nice and crisp as long as I got in it a bit a few days a week. If I didn't the shift quality would suffer and there was a noticeable difference in the way the car drove. I have several friends that will attest to this as well.

So, knowing that...how is the TCM using wear adjusting strategies in such a way where it makes this kind of driving experience difference?

The learning strategy isn't 100% robust when it comes to fill volume vs torque phase vs speed change related controls.
In other words, it wanders a bit due to non-linearity in the tables.
When heavier throttles are used, it has a tendency to slightly drive up the fill volume.
When you consider this is the 'full-on' portion of the solenoid control to fill the clutch before duty cycle control takes over, the elevated fill volume will slightly over-shoot the clutch timing target.
This produces a harsher shift until it learns back down for lighter throttles.
This is the wandering mentioned above and likewise works the other way around when learning low throttle shifts and proceeding to heavy throttle shifts... shifts get extended and drawn out.
Eventually it does find somewhat of an over-all 'comfort zone' but there's always some level of adaptive hunting going on.

[Vigo]

The more juvenile auto enthusiasts of the internet (namely the 18-second 2.0/auto non-turbo DSM crowd, but probably others as well) figured out long ago that if you quickly let off and then slammed open the throttle you could get the 604 to 'bang' shifts. That and what 5DIGITS just described would be examples not of the TCM learning driver behavior but of failing to adapt to it quickly enough.

I think that the main needs of a would-be 604 enthusiast don't actually start with power. In my opinion it would go something like this:
1. Full shift control including being able to move full throttle shift points and not forcing an early 3-4 upshift.
2. Being able to brake-boost the torque converter to a high stall rpm without having the TCM set ratio error codes and go into limp mode.
3. Optimistically, being able to move or mess with a 'torque management' output from the TCM to limit torque during full throttle shifts.
4. Being able to speed up the shifts.

So to me, speeding up the shifts is on the list but its not #1. Lots of people here are ok driving 3spds that in their current condition are shifting slower/worse than a 604. Only the high hp cars really need faster shifts to keep the trans alive. Stock shifts are fine behind ~250tq in stock cars and probably substantially over if you arent realistically looking for 200k-mile life expectancy.

[RoadWarrior222]

The learning strategy isn't 100% robust when it comes to fill volume vs torque phase vs speed change related controls.
In other words, it wanders a bit due to non-linearity in the tables.
When heavier throttles are used, it has a tendency to slightly drive up the fill volume.
When you consider this is the 'full-on' portion of the solenoid control to fill the clutch before duty cycle control takes over, the elevated fill volume will slightly over-shoot the clutch timing target.
This produces a harsher shift until it learns back down for lighter throttles.
This is the wandering mentioned above and likewise works the other way around when learning low throttle shifts and proceeding to heavy throttle shifts... shifts get extended and drawn out.
Eventually it does find somewhat of an over-all 'comfort zone' but there's always some level of adaptive hunting going on.

Heh, Heh, what I'm getting from that is... "If you overfill it, it shifts better."

[Reaper1]

Very interesting information regarding the TCM adaptives. Kinda interesting that Chrysler advertised that it adapts to the driver. I'm not saying you are wrong in any way. Advertising and hype have a way of skewing facts.

[5DIGITS]

The computer learning is primarily concerned with not killing the trans because improper apply and release timing on the holding elements on a 604 will result in a flare or bind which both wear the clutch elements very quickly.
The computer times how long from solenoid application to actual change registered on the ISS and OSS so that it can start its applies and releases so that they all finish in synchronicity.
As much as it has been talked about, i dont think the stock controller 'learns' anything at all about 'driving style'.

This is true.
If you pay very close attention while repeating a particular shift at light throttle (1-2 for example) and repeat the same shifts several times at a higher throttle; you'll skew the learning and feel a difference when you return to the original low throttle test point.
This is why the perception of driving style vs TCM learning is somewhat misleading.

- - - Updated - - -

Heh, Heh, what I'm getting from that is... "If you overfill it, it shifts better."

True but it comes with a consequence.
This exists as minor over-lap of clutch release and apply times and has long term negative effects on premature clutch wear.

[5DIGITS]

Very interesting information regarding the TCM adaptives.
Kinda interesting that Chrysler advertised that it adapts to the driver.
I'm not saying you are wrong in any way.
Advertising and hype have a way of skewing facts.

Very true.

For what its worth and food for thought, a critical variable in the trans architecture is the conversion factor used to properly scale the engine RPM.
This is used to determine much of the TC lock up control parameters and will fault out terribly if scaled improperly.
The best EATX module to use within our applications is the early Prowler EATX controller.
This is largely due to the performance attributes of rapid build up of clutch capacity when compared to the standard applications and the increased rate of speed change speeds.
The shift speeds were set so high during development that the accessory belt would fall off due to the inertial force of the alternator during the 6600RPM WOT up-shift and why ORC's were incorporated on later alternators.
While supportive of the Auto-stick shifter, the CCD version of that controller could be used if the engine pulse train from the crank signal duplicated/produces the signal as received from the 3.5L V6.

[5DIGITS]

The more juvenile auto enthusiasts of the internet (namely the 18-second 2.0/auto non-turbo DSM crowd, but probably others as well) figured out long ago that if you quickly let off and then slammed open the throttle you could get the 604 to 'bang' shifts. That and what 5DIGITS just described would be examples not of the TCM learning driver behavior but of failing to adapt to it quickly enough.

I think that the main needs of a would-be 604 enthusiast don't actually start with power. In my opinion it would go something like this:
1. Full shift control including being able to move full throttle shift points and not forcing an early 3-4 upshift.
2. Being able to brake-boost the torque converter to a high stall rpm without having the TCM set ratio error codes and go into limp mode.
3. Optimistically, being able to move or mess with a 'torque management' output from the TCM to limit torque during full throttle shifts.
4. Being able to speed up the shifts.

So to me, speeding up the shifts is on the list but its not #1. Lots of people here are ok driving 3spds that in their current condition are shifting slower/worse than a 604. Only the high hp cars really need faster shifts to keep the trans alive. Stock shifts are fine behind ~250tq in stock cars and probably substantially over if you arent realistically looking for 200k-mile life expectancy.

Additional...
The 604 (41TE) was offered in Mexico behind the 2.2.
It was also evaluated behind the T-III as a potential option but never gain supportive backing.
With this, EATX controllers exist for Mexican 2.2 packages along with transmission cases that have the starter bell on the 2.2 side, rather than the 2.4 side.
If a TC supplier can be found that's willing to weld the TC to 'min-spec' (front cover to rear impeller housing) then the ~100psi pressure used in 1st and 2nd and the ~80psi used in 3rd and 4th can be increased by 10-25psi while avoiding enough front cover deflection and over-all TC ballooning that typically results in front pump damage.
This is why the lower pressure exist in gears that permit LU.