Tuning Wiki

Reflashing and A/B firmware banks can change the software or calibration present in an ECU, but they are distinct from a driver-selectable calibration mode. Reflashing programs nonvolatile memory. A/B update schemes generally prepare an inactive image and later change which image boots or executes. A vehicle may support either, both or neither; a driver button does not automatically control a firmware bank.

Scope: Examples below describe update mechanisms, not instructions to defeat authentication or safety systems. Confirm the ECU part number, exact software, authorised tool, memory partition and recovery procedure before programming.

Three different operations

OperationWhat happensWhen the new behaviour starts
Single-image reflashA programming routine erases and writes the relevant flash sectors for an approved software or calibration image.Typically after programming, verification and ECU restart; exact sequence is implementation-dependent.
Two firmware banksThe active image remains available while another is prepared in a separate bank, if the hardware and software support it.The update manager validates and activates the new image according to its boot and reset design.
Live calibration selectionA running application selects predefined data or calculation paths for a requested mode.At the designed transition point, subject to operating conditions; no flash erase is implied.

Two files saved on a laptop are not two on-board ECU banks. Equally, two on-board banks created for software updates do not establish driver-controlled map switching. An ECU may accept a new file via diagnostics yet offer no live selection at all.

EDC15 and ME7: distinguish reflashing from hardware selection

On a compatible external-flash ECU, programming each image and selecting which memory device is visible are separate operations. The EDC15P and ME7 dual-firmware boards documented by Inovauto for EDC15P and for specified ME7.5/ME7.1.1 hardware are physical memory-selection products; they are not evidence that those ECUs have the modern internal A/B update mechanism described below. Exact flash device and board compatibility must be checked.

For comparison, a software patch can add a driver mode inside one programmed image. This EDC15 MapSwitch product describes three selectable modes on supported files and requires the user to correct checksums. The published product description does not establish that every ECU has the same layout, or that the mode is a boot-bank change.

Modern example: Infineon AURIX TC3xx SOTA

Infineon's TC3xx documentation describes a TC375 software-over-the-air architecture in which program flash is divided into two groups. With SOTA configured, one group can be read and executed while another group is programmed. This relies on the device's documented bank arrangement: the documentation says simultaneous read-while-write within a single physical PFLASH bank is not supported. It does not mean arbitrary flash regions are safe to erase while executing.

In the documented approach, the address mapping changes when banks are swapped; programming operations use physical flash addresses rather than simply assuming that the current CPU-visible mapping identifies the physical bank. The TC3xx family manual describes SOTA configuration and says the relevant hardware mapping is updated during a subsequent system reset. Device variants, memory sizes, configuration and the production application all matter. This is a semiconductor capability and update example, not proof that a particular engine ECU permits a tune to be selected through this mechanism.

Illustrative update lifecycle

  1. Identify the ECU, current software and approved target image; check compatibility and dependencies between engine, transmission and other controllers.
  2. Prepare the update in the supported programming session and bank or sector. Maintain the supply and communications required by the ECU and programming tool. A/B capability and diagnostic access must be confirmed, not assumed.
  3. Verify integrity and, where required, cryptographic authenticity and authorisation. A simple checksum detects some errors but is not a signature or message authentication.
  4. Activate according to the implementation's permitted vehicle state and reset sequence. Confirm the active software identity, successful boot and required initialisation.
  5. Validate function and diagnostics, including coordination with related controllers. Follow the supported recovery or rollback procedure if verification or boot fails.

NXP's automotive OTA discussion describes dual-bank read-while-write, authentication and rollback support in a particular update architecture. These are design capabilities, not a guarantee that an arbitrary production ECU provides a usable rollback path. The implementation must account for interrupted updates, power loss, authentication failure, memory wear and any anti-rollback policy.

Why update banks are not a shortcut to a switchable tune

  • Updates can involve a reboot, bootloader and state reinitialisation; live map selection can occur within the running control program.
  • Firmware and calibration may have linked version, security and hardware dependencies. A valid image in one ECU is not automatically valid in another.
  • Bank swapping can change executable code and multiple regions, not just fuel or boost targets. Physical and logical address mappings are not interchangeable without verification.
  • Learning, coding, immobiliser data and fault memory may be outside the swapped region and may persist across an update.
  • The ability to receive an OEM update is not permission or proof of technical support for installing a modified image.

For an actual driver mode, establish exactly which maps and protections are switched and test the transition. For an update, establish the approved programming, validation, activation and recovery process for the exact ECU.

Related pages