How do automatics "learn"?

Status
Not open for further replies.
^^^ concur.

I think the "adaptive learning" is overblown because the limits and details are too specific to understand. Subaru used to say the same thing about their ECU's, that they were design to compensate and adjust for engine wear. by default, closed loop operation provides the mechanism, and if the acceptable parameters is widened, then you've allowed for wear and tear. but how do you simplify that for the public to understand?

So the transmissions are designed to adjust shifts to meet a given standard. but that's a programmed standard, not your individual standard. the best it can do is adapt its parameters to your inputs to meet its programmed standard, then maintain that as clutches wear, shift fluid volumes change, etc..
 
The upshot is, the AT will train you, as much or more than you train it.

I bought my first auto in many, many years a couple of years ago. Actually, a CVT. I tried to second guess it for awhile, just as you are doing. I finally just decided to relax, and let it do it, its way.
 
Originally Posted By: supton
I have hooked up my Scanguage, and it does eek out like 2mpg better if I force a downshift into 5th rather than let it stay in 6th unlocked. Going from like 10-11 to 12-14. RPM might drop or be about the same; but getting rid of convertor loss is the big issue.
...

I am with you on this subject. I very much dislike the "unlock TCC on every little incline" when I don't need much more power, and if I really did need much more power then a downshift works much better! And yes, all the torque converter does is waste fuel as it generates lots of heat that the transmission now has to dissipate. With all the ratios we now have, the torque converter is mostly a "launch from standstill" device IMO.

For light vehicles, I also disagree that the TCC it too weak for high output. I suspect most torque converter clutches can hold WOT at 4,000rpm just fine. I know it can on my old 4-speed automatic, and at that RPM the fluid coupling slip is little anyway so you don't gain "engine power" from higher RPM.

And I believe the 'adaptive' buzzword is overused. As others wrote, the adaptive part is mostly to 'adapt' shifting parameters to compensate for gradual wear over the lifetime. Adaptations never overcome significant problems.
 
Any opinions on the "Autostick" type transmissions in some Chrysler products that allow you to manually shift the automatic tranny?
 
Originally Posted By: supton
Or more precisely, how do you "teach" them? I've read that if you disconnect the battery you can make 'em "forget" and it will start over again, and relearn.

I'm curious what one can do to basically make it shift when you want it to. Dawned on me recently to simply let off in high hopes of getting it to shift--which is fine, assuming it won't downshift when you step on the throttle again.

Would be nice if I could somehow pick a shift schedule. I still say it'd be nice to have a light that would indicate imminent downshift!


Since there are so many different designs of slushboxes and other automatically shifted transmissions there is no way to give advice until you specify the exact make and model.

Many patents are involved, for example GM has patented algorithms that control the converter clutches, and the tech used is extremely variable. Then you have the "one trans fits all" boxes like the ZF 8 speed found in Audis, Bentleys, BMW's, Jaguar, Chrysler, etc. That slushbox is made in differing variations for all the different models of vehicles it is used in, each with unique programming!

In my old Hemi you can simply pull the powertrain control fuse and reset everything as far as adaptives go, but they will quickly change as you drive. My solution was an aftermarket tuner which drastically altered the maps used to determine shift points and finally made the autostick feature work instantly like it should. But it has infinitely variable TC lockup in any gear but first and reverse with extremely high torque capacity.

Our newer 3500 vans with the 6L80E transmission almost never unlock the converter and they are rated to tow small houses! Designs vary wildly, so don't assume the TC is a weak point.

Note that most DBW vehicles do not respond to the old "shift it with your foot" method of altering the throttle to make a shift happen, in the old daze you could just let up and shift and be back in the throttle easily and quickly.

Those days are gone. But if our new Ram is any indication, slushboxes are getting a LOT better!
 
Last edited:
Originally Posted By: BearZDefect

I am with you on this subject. I very much dislike the "unlock TCC on every little incline" when I don't need much more power, and if I really did need much more power then a downshift works much better! And yes, all the torque converter does is waste fuel as it generates lots of heat that the transmission now has to dissipate. With all the ratios we now have, the torque converter is mostly a "launch from standstill" device IMO.


Its categorically NOT true that "all the torque convertor does is waste fuel." It may have a relative slip (speed differential) between its input and output shafts when its it its torque multiplication mode, but it is not just "slipping" the way a dry clutch would be. It is pumping fluid through a stator and onto a turbine, creating a hydraulic reduction/torque multiplication system that is actually pretty efficient most of the time. Is there some loss? Sure, but its not necessarily a large loss. Even a gearset doing a similar reduction/multiplication has SOME loss. You have to realize that the programming in automatic transmissions has, as one of its primary goals, efficiency. Modern automatics don't unlock the convertor to deliberately waste fuel- they usually do it to *save* fuel, or at least maximize performance without excessively hurting efficiency. They're pretty darn good at deciding whether dropping a gear or unlocking will do the job best, in contrast to automatics of 15 years ago.

Quote:

For light vehicles, I also disagree that the TCC it too weak for high output.


Actually its heavy-duty vehicles that typically have the strongest TCC's designed for holding against a much higher percentage of the engine's torque than lightweight front-drive automatic transaxles do.

Quote:
I suspect most torque converter clutches can hold WOT at 4,000rpm just fine.


Some can. I doubt that "most" can. "Most" TCC clutches are very narrow 1-disc clutches at the rim of the torque convertor, often with a very thin layer of friction material compared to the discs in the main clutch packs. A few are actually implemented as multi-disc clutch packs, though, and some of the "rim of the convertor" type do have far more holding power and are made much beefier than others.

Quote:
And I believe the 'adaptive' buzzword is overused. As others wrote, the adaptive part is mostly to 'adapt' shifting parameters to compensate for gradual wear over the lifetime.


Agreed 100%- as well as to reduce the effects of manufacturing tolerances, compensate for servo leakage rates, pump wear, and so on. However, adaptives DO adjust things like the sensitivity to downshifting. If a driver consistently demands more power on every uphill to the point of forcing a downshift, the transmission will downshift more readily than if the driver consistently applies minimal pressure and avoids a downshift. But that isn't really the primary purpose of adaptives.
 
Originally Posted By: Olas
If you want to control your gearbox to shift when you want it to, buy a manual.


Nah. I will stick with the manual mode with the 6 speed auto in my pickup. It is a great feature for the rural rolling hills in my neck of the woods. And on the commercial side, with manual mode on the 10, 12, 13, and 18 speed automated transmissions in the heavy trucks. Just getting older. I would much rather have the trans do all the work and I just tell it when to do it with a button. I have operated manual non synchronized 13 and 18 speed transmissions for several decades and have no desire to push a clutch any more. Let technology do the hard work now. It does it much better, it doesn't get an attitude or road rage, it doesn't get tired after going down the road for hours on end, etc. It just shifts smoothly, every time.

Those that have a fascination with a manual trans have not had to operate a non synchronized variety, over 100,000 miles a year, had to do it in major metro rush hour traffic, crawling mountain passes with gross weights of 80,000 lb plus, etc. Moving thru manual gears in a auto or pickup is easy by comparison. Pure child's play. The OEM's have those trans synchronized to take up a lot of the slack. It allows even the incompetent drivers to get by. A heavy truck manual trans is a very unforgiving critter by comparison.

Those that are "purists" about manual transmissons need to get in a class 8 truck and try to shift thru those gears under a load. It will make the best auto driver look like a total idiot.
 
Originally Posted By: 440Magnum
Originally Posted By: BearZDefect

I am with you on this subject. I very much dislike the "unlock TCC on every little incline" when I don't need much more power, and if I really did need much more power then a downshift works much better! And yes, all the torque converter does is waste fuel as it generates lots of heat that the transmission now has to dissipate. With all the ratios we now have, the torque converter is mostly a "launch from standstill" device IMO.


Its categorically NOT true that "all the torque convertor does is waste fuel." It may have a relative slip (speed differential) between its input and output shafts when its it its torque multiplication mode, but it is not just "slipping" the way a dry clutch would be. It is pumping fluid through a stator and onto a turbine, creating a hydraulic reduction/torque multiplication system that is actually pretty efficient most of the time. Is there some loss? Sure, but its not necessarily a large loss. Even a gearset doing a similar reduction/multiplication has SOME loss. You have to realize that the programming in automatic transmissions has, as one of its primary goals, efficiency. Modern automatics don't unlock the convertor to deliberately waste fuel- they usually do it to *save* fuel, or at least maximize performance without excessively hurting efficiency. They're pretty darn good at deciding whether dropping a gear or unlocking will do the job best, in contrast to automatics of 15 years ago.

...

I defer to you on the hard facts, thank you for the corrections.

I will use your logic to maintain my opinion that while *some* modern automatics may intelligently unlock the TCC, I believe many still do it simply to minimize shifts. Perhaps the manufacturers find that the TCC can withstand more engagements than the transmission can withstand actual shifts?

And why do many modern automatics unlock the TCC as soon as the vehicle encounters an incline? The driver did *not* ask for more power using input go pedal. I maintain that unlocking the TCC in this case does nothing but waste fuel. It should have stayed locked in the same gear.

Another irritating behaviour is holding third (or higher) gear when launching from a fast creep after slowing down. This is blatantly relying on the torque converter when it should have downshifted. Force a downshift (with the auto-manual control) and the RPM ends up nearly identical, but the power to the wheels is clearly higher. The rest of the power goes to heat in the ATF from the torque converter when in the higher gear.

Modern six speed automatics continue to do what I am criticizing. They don't need to, they have enough ratios. They are not the old three speed automatics that needed the torque converter so much. It makes me suspect lack of programming refinement. After all, if you launch from a standstill on flat ground they seem to do everthing right. But face an incline, or resume after slowing down, and all the good logic seems to go out the window.
 
Yep.

My truck has around a 2k stall convertor. Pretty standard I think. But it's a heavy truck. I find I can sometimes get lower rpm's by forcing a downshift--the convertor will lock up in a lower gear. For example, I might be climbing a hill at 1500rpm in sixth, it unlocks and will be at 1900rpm. Force 5th and it'll be at 1800.

That's not as bad as when I'm going slower, and still find 1,900rpm in unlocked--and would get say 1,500 after a downshift&lockup. That's a lot of slippage.

I'm not worried about the heat buildup, it's the pure inefficency that bugs me. I still don't think the engine needed a downshift in the first place, most of the time (I think it's perfectly capable of hill climbing at 1,500rpm); and the rubber band effect is noticable (it'll swing a few hundred rpm on the hill in the too-tall gear if I modulate the throttle).
 
Back when automatic transmission were simple (i.e., vacuum modulated shifting) you would make it up-shift by quickly letting up on the accelerator. Still it was a shabby way to shift, but the modern electronically controlled transmissions are hopeless. Best auto tranny I ever drove is the 3 speed vacuum modulated tranny in my wife's 1999 Plymouth Voyager. But even then, driving is a chore. Put a manual in it and driving is a good time.
 
The 5-speed auto in my mother's '04 diesel BMW has the annoying feature of downshifting on downhills at the second application of the brakes. It shifts back up if you just touch the gas pedal briefly.

My new VW Polo with 7-speed DSG and Bluemotion not only does that but (according to the manual) it will coast down hills at no throttle application (both clutches open,) only re-engaging top gear clutch on first brake application. On each subsequent brake applications it will downshift one gear as long as RPM limits allow. The coasting feature has not particularly obvious to me yet, but I've only done a few hundred kms.
 
My brother lives in a hilly area and bought an automatic. He was lamenting that on slippery uphills it will upshift when he wants it to stay in lower gear. He got what he asked for. Never, never buy an automatic with the thought that it will do what you want it to do. Kind of like Microsoft operating systems.
 
Many of us need to get out more and experience a new car with a well programmed automatic.

They're out there.

Blanket statements about all slushboxes simply do not apply once you consider the number of differing designs and then factor in programming.

I have driven two slightly different vehicles with exactly the same transmission that exhibit completely different characteristics due to programming...
 
Status
Not open for further replies.
Back
Top