Configuration and calibration, what’s the real difference?

Hello everyone, I realized that I haven’t posted anything for quite some time. Recently, I have been working with ISO26262–6 Software Calibration (C5.4) and Configuration (C5.3) data, so I wanted to write a short piece about it — because why not?

Configuration Data:

Configuration data is defined in our glossary (ISO26262–1) as follows: ISO26262–1:3.22; “data that is assigned during element build and that controls the element build process”, with pre-processor variables often given as an example. From a software developer’s perspective, one might consider every define/macro as a configuration parameter. However, the important question is: how is the system designed?

And based on this, we should ask: Is there any variant information coming from the customer? If so, are we going to apply this variant at runtime, or will we generate multiple binary files?

Once these questions are answered, we can identify the first common type of configuration data. If variant creation will not be performed at runtime (which we will later classify as calibration data), and if you embed a different binary in production for each customer-defined variant using macros during build time, then these macros should be considered configuration data.

For example, imagine you have a headlamp control ECU that also handles brake lights (a fictional scenario). If you develop a shared codebase and close/open certain features with macros to create different binary files (and possibly update CAN messages with additional macros), these macros qualify as configuration data. Pre-compile-time macros that change during build are also configuration data.

What if your project has no direct configuration parameters, but you purchased a library from another internal team or a Tier-2 supplier? Will you not configure it? Will you not enable/disable specific features? If you do, every parameter you adjust becomes configuration data. According to clause C4.1.d, you must also explain the relationships between these parameters.

But what if you receive this information at runtime (over UDS or another communication protocol using a specific tool)? What do we call that?

Calibration Data:

The definition of calibration data is simpler. First, we check the glossary: ISO26262–1: “Data that will be applied as software parameter values after the software build in the development process” And the example is clear: “Variant coding, vehicle-specific parameters.”

As you may notice, both data types mention variants. However, the fundamental distinction is that one applies at pre-process/build time, and the other at runtime.

For example, imagine you have a door control unit where the right door has one switch, while the left side has 5–6 switches. If production volume is low, you may not want to generate separate binaries. If you also have no memory constraints (likely in a small ECU), you can design this as calibration data — unless the customer specifically demands otherwise.

In summary, this is the essential difference.

If you believe any information is missing or incorrect, please feel free to contact me.

See you in the next article!

linkedin : www.linkedin.com/in/yiğit-rahim-k-a22974195