Some ECUs can select a different calibration region by changing an address used by their software. A second region can contain corresponding maps, scalars, axes and metadata in the same relative arrangement. This method is fundamentally different from switching a physical flash chip, and from switching only a handful of individual tables. It requires matching executable code and sufficient verified, addressable storage; extra memory alone does not provide a switch.
Applicability: The addressing scheme varies by processor and software version. The MEMS3 example below is documented for a specific third-party patch; it must not be assumed to apply to EDC15, ME7, EDC17, MED17 or MG1 generally.
How base-address selection works
A controller may load a calibration base into a register or variable, then address parameters by offsets from that base. Conceptually, effective address = selected base + parameter offset. If two regions have the same layout and the readers use this base, selecting the second base redirects those readers as a group. Other implementations select a pointer table or per-table pointers, or use hardware address translation; the switch interface does not reveal the internal method. Vector's CANape information lists several pointer- and overlay-based calibration concepts.
Illustration only: suppose two 64 KiB regions begin at byte addresses 0x10000 and 0x20000, and a parameter is at offset 0x20A0. The effective addresses are 0x120A0 and 0x220A0 respectively. Four complete 64 KiB regions use 256 KiB before code, metadata, alignment or flash-sector overhead. These numbers are invented to explain the arithmetic, not addresses for any production ECU.
Data copied to another location will have no effect if the executable still reads an absolute address. Check every consumer: direct references, constants embedded in code, pointers and pointer widths, indexes, axis arrays, cached copies, data loaded at startup and accesses by other tasks or cores. A region may omit unrelated calibration, learned values, coding or code. Even a coherent pointer change cannot guarantee that stored adaptations or control integrators are appropriate for both modes.
Documented example: Rover MEMS3
Andrew Revill's MEMS3 investigation documents a petrol MEMS3 dual-map patch in which a RAM-held map base is loaded into register a5 at the start of a calculation loop; map data is then accessed relative to that base. The patch places another complete map in unused EEPROM space and adds switching logic. The author requires both calibrations to match the firmware and lists examples where available space is insufficient. His physical-switch option switches at a loop boundary; his throttle option selects at ignition-on and requires the ECU to shut down and restart before changing. Pin availability depends on ECU and vehicle equipment.
Fuel and idle adaptations may be shared between the modes, and variables initialised at start-up may not immediately match a changed map. The author reports limited personal testing; this is not a guarantee of safe switching in every operating condition. Check exact hardware, firmware, switch configuration and safety validation instead of copying addresses or pin functions into another vehicle.
EDC15, ME7 and modern multi-file products
One published EDC15 software map-switching tool and a separate ME7.x solution demonstrate software selection on supported files. Their public descriptions do not prove that either uses one relocatable complete region. Some EDC15P hardware instead selects entire external flash images; see the external memory method.
For a newer contrasting example, the beta MG1/MED17 MultiFile tool accepts an original and up to seven additional matching-software files. The vendor says it extracts differences, stores alternative maps in unused flash and adds switching logic. It describes certain VAG and Mercedes applications, cruise or pedal selection, tachometer indication and optional EEPROM persistence. Despite the multiple complete input files, this description does not prove that the complete calibration address or entire program is switched. Confirm the precise ECU, software, changed areas and protections with the supplier.
Space and layout checks
- Establish actual physical and logical memory layout for the exact image, including bootloader, firmware, calibration, NVM variables, diagnostic and security data. A run of
FFbytes or a difference between file and chip size is not evidence of safe, accessible spare memory. - Check whether the processor can address each bank through the existing instruction sequence, including segment registers, alignment, addressing modes and potential overflow. Check flash erase and program granularity, ECC, write permissions and valid power-up behaviour.
- Maintain identical expected parameter layouts and formats where base-relative selection is used; check signedness, scaling, axes, dimensions and interpolation. Determine whether axes are shared or copied and verify all indirect references.
- Identify checksum and authenticated ranges, and use the authorised supplier's programming and validation workflow. A simple checksum is not the same as a cryptographic signature. Do not assume unused-looking space is accepted by secure boot or runtime integrity checks.
Transition and verification
Specify a valid default and what happens on invalid input, reset or interrupted nonvolatile selection writes. Latch a validated mode at a defined calculation boundary; coordinate all affected tasks and cores. Consider gating changes until a low-load state and applying validated rate limits where needed. Avoid combining values from different modes within one calculation, and verify the safety limits remain active in every mode.
- Identify hardware and software IDs, memory map, tool version, calibration definitions and a known-good recovery route.
- Record exactly which parameters, axes, code and adaptations are shared or switched. Validate mode-specific fuel, ignition or injection, torque, air, boost, limits and drivetrain interactions.
- Check selected addresses or actual lookup results, not only an indicator. Test each mode at initial start, after mode changes, reset, bad input and loss of selection storage.
- Validate steady-state and transition behaviour at progressively higher loads in a controlled environment, recording actual operating parameters and retaining protective controls.
Related: Switchable maps overview. Technical details above are examples and design checks, not a claim that each ECU supports a particular patch.