RAM calibration pages and memory overlays allow suitably equipped development ECUs to use working calibration data without reprogramming flash after every change. They are often used with a calibration tool, rather than as a permanent dashboard-selected tune. Hardware, firmware, tool configuration and access permissions determine whether a particular ECU actually supports them.
Reference and working pages
In a common two-page arrangement the reference calibration resides in nonvolatile flash and a working copy resides in RAM. The tool can change the working page while the ECU uses the reference page, verify the result, and then activate the working page. The logical parameter address can stay the same while its physical backing changes. RAM ordinarily loses its contents when its supply is lost; a system must detect invalid data and reinitialise or fall back to a valid page.
ASAM's XCP description defines optional page-switching services such as Set Calibration Page, Get Calibration Page and Copy Page. The ECU's and the calibration tool's page access can be controlled independently when implemented. Preparing an inactive page and switching the ECU's access atomically avoids exposing a partly written large map. Segment/page permissions and descriptions are supplied through the ECU implementation and A2L; XCP support, its page services and accessible transport are not guaranteed simply because the vehicle has a diagnostic socket.
Hardware overlay on AURIX TC3xx
Infineon's TC3xx family manual, data access overlay section describes redirecting selected TriCore data accesses from flash or another supported source range into a supported overlay memory. It specifies up to 32 individually configured blocks per TriCore, block sizes from 32 bytes to 128 KiB, and coordinated activation of selected blocks. The exact device variant, available memories and configuration determine practical capacity.
The documented translation acts on the participating TriCore's data accesses: it does not automatically redirect that core's instruction fetches or accesses from other bus masters, including a debug interface. Each core's configuration, inter-core timing, memory protection, cache aliases and invalidation matter. A debugger viewing the original flash and application code reading an overlaid RAM value can legitimately show different data. Infineon's overlay usage note discusses cache effects, alignment and block configuration. A hardware facility in a microcontroller is not proof that a particular locked production EDC17, MED17, MG1 or other ECU exposes it.
Memory demand depends on block coverage, not just changed bytes. As a purely illustrative example, if a particular setup used 4 KiB blocks and two 100-byte maps required separate blocks, the changes could occupy 8 KiB of overlay memory. Real permitted block sizes, alignment, overlap rules and availability come from the exact MCU, software and tool; see the ETAS memory layout editor documentation.
Real development example: ETAS XETK-S20.1B
The ETAS XETK-S20.1B user guide describes a reference page in flash and a working page in reserved emulation-device RAM. The working page is initialised from flash. In its protocol-based method, the XETK requests a switch using a mailbox in ECU RAM and the ECU software changes overlay registers. For supported TC3xx variants, the guide also describes direct register access by the XETK.
That specific hardware monitors standby RAM voltage. If voltage falls below its threshold, retained working-page data is no longer assured; the system returns to the reference page and requires reinitialisation before working-page use. These documented behaviours belong to this development tool and its supported configurations, not all XCP interfaces or road ECUs.
Practical workflow
- Identify the exact controller, ECU software, development hardware, memory availability, calibration tool and matching A2L. Confirm page switching and authorisation are actually implemented.
- Reserve working memory, determine the block and cache rules, initialise it from the known reference, and verify that other software or tools cannot inadvertently reuse it. Define the startup and loss-of-power state.
- Prepare a change on the permitted inactive page, verify dimensions, axes, units and values, and activate it using the system's supported coherent page-switch mechanism. Log the selected page and actual values read by the controlling core.
- Check data visibility across the relevant cores, DMA and debugger, plus resets, faults and fallback. Validate related torque, air, fuel and protection maps together and assess transients before increasing load.
- Use an authorised, supported flash and integrity workflow if results need to survive removal of the development equipment. Do not assume the working RAM page automatically becomes a durable production tune.
Comparison with a driver-selected mode
Overlay and XCP page switching are memory and development mechanisms. A flash-resident driver mode may select only certain preprogrammed tables or larger verified regions without an external tool. The MG1/MED17 multi-file product description says it stores differences as separate maps in free flash and adds mode-selection logic; it does not establish the use of XCP or a RAM overlay. Diagnose the actual technique on the exact controller rather than treating all live-switching products as equivalent.
Related: Switchable maps overview.