B9 ZF Tranny

AdyZaky

Registered User
Joined
Jun 7, 2017
Messages
4
Reaction score
1
Points
3
Hello

Found a strange issue with the car, it always use to shift all the way to 6th quite quickly, around 30mph or so the car would shift in to 6th (Non Sport mode)

I've been driving more enthusiastically recently and have now noticed the car holds onto gear longer (not quite as long as S mode), it's very noticeable, it now will only go to 5th gear at 30 mph or so.

I assume the tranny software has learnt my driving style I suspect and is a little more aggressive in changes now, is there a way to reset this?

Cheers
 
No cars learn driving styles, if you are using more throttle the ecu is programmed to hold the gears for longer...
 
No cars learn driving styles, if you are using more throttle the ecu is programmed to hold the gears for longer...

Note sure I understand, I'm using the same amount of throttle as before, but the car just holds the gears longer, I could coast at 30mph and it won't go above 5th gear, where as before it would go to 6th. So the stuff about adaption with the ZF isn't true?
 
But you stated you are driving it more enthusiastically, now you are saying you are using the same amount of throttle. Which is it because in my book driving more enthusiastically and using more throttle is the same. It is true that a ecu will calibrate all the sensors and make adjustments where necessary but this usually happens within the first 50 miles.
 
@AdyZaky - do you have ADS? If so is it possible you have changed mode - that will certainly affect the shift points.
 
You were probably in Eco mode before, it changes up very early.

As @Dippy suggests check the Drive Select settings.
 
My car cruises at 30mph in 5th gear and shifts upto 6th at 33/34mph or something like that, this is in non sport mode.
 
No cars learn driving styles, if you are using more throttle the ecu is programmed to hold the gears for longer...

That's not quite true, google "Adaptive Learning Gearbox"

There's a lot of info out there about it.
 
  • Like
Reactions: Spaceman_RS3
Right, so if many people drive one car does it have a hissy fit because it can't adapt. No it goes about changing settings according to what it receives from sensors. Switch the car off and the whole process restarts when you set off again.
 
It doesn't reset with ignition on Jaguar Land Rover vehicles, to reset the box you could do this via the OBD or disconnect the battery, short the leads together, this would hard reset all the ecu's.
If I drove the Freelander hard it would be like a demented animal the following morning until it learnt the new driving style.
 
  • Like
Reactions: Spaceman_RS3
Ok if you think so...

Haha, this technology genuinely exists, you're correct it is computer controlled but there are algorithms that help it adapt to the way it's being driven, therefore there is a level of 'learning' capability.

Are you saying you don't believe this?
 
  • Like
Reactions: Spaceman_RS3 and Daggerit
It doesn't reset with ignition on Jaguar Land Rover vehicles, to reset the box you could do this via the OBD or disconnect the battery, short the leads together, this would hard reset all the ecu's.
If I drove the Freelander hard it would be like a demented animal the following morning until it learnt the new driving style.
Spot on and Audi is the same there were plenty guys having to have resets done on gearbox for rs3 due to jerkiness. Cuke has his own mind on this. Old dog new tricks comes to my mind.
 
  • Like
Reactions: R.A.G and Phil's Barber
Ok then and Cuke pointed this out in a previous post, how does it adapt when more than one person drives the car. Are you really saying it knows person A is driving and not person B.
 
It adapts dynamically as you drive, however, it takes a few miles, there are the initial adaptations which must be performed to 'write' the basic instruction to the gearbox, these are quite a set procedure and probably completed on a dyno I suspect? then there are the dynamic instructions as you drive, I don't feel as though they reset with the igintion.
With my previous car, as I said I could blast over the moors, arrive home and the following morning the car would take a few miles before the gear changes reverted to a more sedate pace, I believe this issue also causes some people to dislike cars when they test drive them, the box is expecting the last driver and his/her style and can be viewed as a 'crap' gearbox by unsuspecting buyers.
 
  • Like
Reactions: Spaceman_RS3 and R.A.G
It's an argument as old as the hills and unfortunately, like using more expensive fuel in a 1.4 engine, there is no real objective scientific evidence to back it up only anecdotal. If you really think the tiny ecu with it's small memory can 'learn' inside a gearbox then fine. Yes algorithms can appear to learn but all is happening is that it is following programed instructions. In the true sense of the word that is not learning.
 
Ok, if you think so ...


Here's food for thought, ZF's own manual on how to 'teach' the gearbox once it has been reset, through this procedure the gearbox learns when it should change gear, these adaptions are not pre programmed and are unique to each car, do this procedure differently and the gearbox will behave differently.
As you say it is an age old argument and semantics can skew the interpretation.


IMG 0067
 
  • Like
Reactions: R.A.G
It's an argument as old as the hills and unfortunately, like using more expensive fuel in a 1.4 engine, there is no real objective scientific evidence to back it up only anecdotal. If you really think the tiny ecu with it's small memory can 'learn' inside a gearbox then fine. Yes algorithms can appear to learn but all is happening is that it is following programed instructions. In the true sense of the word that is not learning.

Well these forums would be boring places if we all agreed

.....but you're completely WRONG!!

:yahoo:
 
  • Like
Reactions: Spaceman_RS3
Can you actually imagine getting the tech to do that 9 times? It's hard enough to get them to unplug it from the computer and take it on the road. :D
 
I wonder if the car can be controlled on a rolling toad to perform this procedure automatically? It is an onerous task to complete the above and probably nigh on impossible on public roads?
 
*** ... Fat Finger Syndrome :whistle2:
 
  • Like
Reactions: scotty76
Pretty graph but meaningless as adaption does not necessarily mean learning. All that graph is demostrating is probable calibration of the sensors.
 
The disagreements in this thread may be down to interpretation* of the word "learning". In light of current developments in AI and the move towards self-driving cars, it would be easy to dismiss adaption as not being true learning. However a system which simply changes a single parameter in response to user action, and as a result changes its own future functionality in order to perform more closely to the user's probable preference, has, in my mind, learnt.

*Is society getting worse at this? I used to think that the word "phobia" meant "fear". But it seems these days it means "hate".
 
There is definitely capacity in the DSG / S-Tronic gearbox systems to 'adapt' due to wear of the clutch packs and other components. It does not seem impossible that the engine / gearbox management system can identify different driving styles though monitoring inputs in throttle / braking behavior and steering inputs and then moving to different pre-programmed maps.

After all this is at the heart of dieselgate where the ECU's were programmed to monitor for driving behavior indicative of testing scenario's and then adapt the engine parameters to suit.
 
Of course they change their calibration, and in modern ecu's and sensors they perform this task instantaneously and thousands of time in a second, but not identify because to identify you have to have a human like qualities. They do not need to because they can make theses changes the instant you get into the car and start the engine. Gone are the days when a car's driven conditions could be determined by its driving characteristics and only reset by disconnecting the battery. At least in that sense they have become cleverer.
 
Last edited:
The disagreements in this thread may be down to interpretation* of the word "learning". In light of current developments in AI and the move towards self-driving cars, it would be easy to dismiss adaption as not being true learning. However a system which simply changes a single parameter in response to user action, and as a result changes its own future functionality in order to perform more closely to the user's probable preference, has, in my mind, learnt.

*Is society getting worse at this? I used to think that the word "phobia" meant "fear". But it seems these days it means "hate".
You are correct to a point, algorithms can be written that would, say, if the throttle was operated at 65% for a given time then the gear can be held for a higher rpm before it changes. But that isn't learning the drivers style of driving. Input - output...
 
We're in agreement there :)
 
  • Like
Reactions: cuke2u

Similar threads

Replies
18
Views
2K
Replies
153
Views
13K
Replies
4
Views
925
Replies
3
Views
1K
Replies
43
Views
9K