Calibration information
I have created a number of videos that explain what is going on when a calibration is being performed. Go to the resources page (link above) to watch these.
But here are the basics.
A calibration is done to gather the characteristics of the transmission so it can operate afterwards. There is a lot of information that needs to be gathered about sensors (distance, pressure sensors), control devices (solenoids) and how these affect the transmission components like pressure, clutches and shift rods.
Each sensor is an output device and nothing more. There is no programming of any sensor. The TCU simply goes through many functions of the transmission and gathers the information required so it can operate later. More information is stored afterwards when driving, and adaptations will be constantly gathered by the TCU during all operations during the lifetime of the vehicle.
There are two types of calibration.
1. Calibration after parts replacement. I call this the “long cal”.
2. Calibration without previous parts replacement. I call this the “short cal”.
The long cal has 5 separate routines:
A. Programming of the TCU. This is where the previous calibration data and adaptation data is deleted.
B. Pressure sensor calibration.
C. Shift rod calibration. This is where the distance sensor data is collected. It is also where all the valve body information related to gear shifting is collected.
D. Hydraulic teach in. This mainly deals with emergency functions of the valve body. It also gathers some information about clutch valve operation. Unfortunately this routine is very prone to failure.
E. Clutch calibration.
The short cal only carries out routines C and E.
Important information to understand.
The calibration routines are in the TCU, not PIWIS or other diagnostic program. PIWIS just turns them on and off.
If you attempt a short cal and it fails, attempting the long cal will still fail on the same routine as the individual routines are identical.
BUT, here is the problem.
If you attempt a short cal only and it fails, you can delete the fault codes and it’s like you never attempted it.
If you attempt a long cal and it fails, you have deleted all the previous calibration data, so you must now complete the full long cal successfully to get the car going. Part D can be very problematic and take many attempts to pass.
If you attempt a long cal and it fails, and you consequently attempt a short cal, this will be the most likely result. The shift rod calibration will pass (if all is OK) and then the clutch cal will fail immediately it starts. This is because the clutch cal (part E) requires some date from the hydraulic teach in (part D), and this was deleted by part A of the long cal that had been previously attempted.
So unless you definitely need to conduct a long cal due to the work being done, I would suggest you never attempt it.
It has been reported that aftermarket diagnostic tools can brick the TCU during calibration. For this reason I suggest only using PIWIS for the calibration. I personally use the PIWIS 3 that is described on the PDK Pirates website. Link below. I have used this extensively for sensor testing and general diagnostic work. It has worked a charm, cost peanuts to purchase, and is relatively simple to set up.
Calibration failure indications.
Sometimes the calibration fails due to a car network connection problem. The image below shows what you see when this happens.
If the calibration fails due to incorrect sensor outputs, clutch problem, etc, you will see something like below. It will give a reason/reasons for the failure.
I’ve only had one reported case of a JUMPS sensor not calibrating. This was from a large workshop in Europe who fixed a lot of PDKs. The car was a 987 with an early software load. When the software was updated it was reported that transmission calibrated without issue. I have no idea why, but it would seem that the calibration routines have changed over the years with updated software and this does affect calibration performance.