References

What we have built.

A selection from our projects. We name customers only with their consent. That is why most projects are described without a name, but with what matters: task, approach and result.

Energy management

One meter emulator for inverters from ten vendors

inverter vendors covered
10
transports: Modbus RTU, Modbus TCP, Speedwire
3
  • Modbus RTU/TCP
  • SunSpec
  • SMA Speedwire
  • Node.js
  • Edge
Challenge
PV and battery inverters regulate on the readings of a grid meter and usually accept only the model of their own vendor. Anyone who wants to use readings from a different metering system needs a separate meter for each device family.
Approach
We developed an emulator that presents itself to the inverter as the meter it expects. Profiles are selected by configuration, several run at the same time, and each one was commissioned on real inverters.
Result
One software service replaces vendor-specific meters. New inverter families are added as a profile, secured by automated tests.

Test Environments & Simulation

Energy management · § 14a EnWG

One data contract for all assets, § 14a as a tested building block

data points in the contract
318
source of truth for all services
1
  • § 14a EnWG
  • EEBUS
  • Modbus
  • JSON Schema
  • CI/CD
Challenge
Heat pumps, wallboxes, storage and PV systems from different vendors had to be connected and controlled uniformly across several edge services, including power limitation under § 14a EnWG at the grid connection point.
Approach
We introduced a versioned data contract as the single source of truth, with validators and a release pipeline. A reference implementation for asset integration and the limitation logic build on it.
Result
All services speak the same, automatically validated language. New asset types are developed and tested against the contract.

Software & IoT Engineering

Charging infrastructure

Virtual charge points: integration tests without a charging park

virtual charge points
OCPP 1.6
emulated meters and chargers
Modbus
  • OCPP 1.6
  • Modbus
  • Python
  • WireGuard
  • CI/CD
Challenge
A controller for charging and energy management has to work with charging stations and meters from many vendors. Real charging parks are rarely available for testing.
Approach
We built a web application that connects several virtual OCPP charge points to the controller and emulates meters and chargers over Modbus. It runs in the cloud and reaches the real on-site controller through an encrypted tunnel.
Result
Integration tests run against an environment that is operated from the browser and can be redeployed at any time.

Test Environments & Simulation

Metering point operation

Connecting smart meter gateways securely to network management

transport security for SNMPv3
TLS 1.3
pre-standard implemented
DIN VDE V 0418-63-12
  • SNMPv3
  • TLS 1.3
  • Docker
  • AWS
  • VPN
Challenge
Smart meter gateways report operating states via SNMPv3 with TLS transport security as defined in DIN VDE V 0418-63-12. Common tools do not support this profile out of the box.
Approach
We built a receiver for SNMPv3 over TLS, packaged it as a container and put it into operation in the cloud.
Result
The receiver was tested successfully with a real smart meter gateway over TLS 1.3.

Software & IoT Engineering

E-mobility · software platform

Separating a software platform: from findings to a running target environment

repositories assessed
90+
phases: assessment, steering, execution
3
  • Assessment
  • Terraform
  • AWS
  • GitHub
Challenge
A platform for charging and energy management was to be carved out of an existing corporate structure. It was unclear which repositories, dependencies and services were affected and where the risks were.
Approach
An inventory of all repositories with a findings report and an executive summary, tooling to take over and synchronise the code into the new organisation, and the development environment as infrastructure as code.
Result
The development environment can be built reproducibly. It was handed to the customer’s team together with the agents that deploy and verify it.

Software & IoT Engineering

Energy retail · market communication

Ordering load-profile data through German market communication

quote request and order
EDIFACT
gateway for message exchange
AS4
  • EDIFACT
  • AS4
  • S/MIME
  • Python
  • Docker
Challenge
An energy company needed load-profile data from the distribution system operator. The route there runs through the standardised market processes: EDIFACT messages, signed and encrypted.
Approach
We implemented quote request and order as a tool, including signing and encryption, and set up an AS4 gateway for message exchange.
Result
Quote request and order run as a repeatable, scripted procedure.

Software & IoT Engineering

Battery electronics · test systems

Test systems from one team: electronics, firmware, fixtures and test software

HiL systems built
20+
to the first system
3 months
  • HiL
  • PCB
  • Firmware
  • Python
  • CAN
Challenge
A manufacturer of battery management systems needed test equipment that could not be bought: available quickly, tailored to the product and extensible by its own team.
Approach
We developed hardware-in-the-loop systems and functional testers: our own boards for cell emulation and fault injection, firmware, mechanical fixtures and an open Python test framework. Test cases from development also run in production.
Result
The first system was ready after three months. It grew into more than 20 HiL systems and more than 7 functional testers for several production sites.

Hardware & Embedded Engineering

In-house project · demonstrator

The agent checks its own work: firmware development with camera and oscilloscope

feedback paths: camera and oscilloscope
2
instrument controlled directly
SCPI
  • Firmware
  • SCPI
  • PlatformIO
  • Teensy
Challenge
AI agents write firmware quickly. Whether it does the right thing on the board only shows in a measurement. Without feedback from the real world, the agent works blind.
Approach
We gave a coding agent tools: build and flash firmware, send commands to the microcontroller, observe the result through a camera and measure signals with a networked oscilloscope.
Result
The agent closes the loop itself: change, flash, measure, correct. We use the same principle in customer projects for tests on real hardware.

AI Development & Consulting

Next step

Let’s talk about your project.

A short conversation usually settles faster than a specification whether and how we can help.