Home

References from More Than Three Decades of Embedded Development

Clients in automotive and industry have trusted my embedded expertise since the founding of SCHMITT CONSULTING. Two projects exemplify the way I work.

As an independent embedded consultant, I develop for clients who need hardware-near software, real-time behaviour, or the combination of both. My work ranges from fixed-price contracts with firm delivery commitment, through hourly budgets, to accompanying engineering departments over several months.

For confidentiality reasons, not all projects can be presented publicly. The two examples below describe projects whose key facts I am allowed to mention — and which exemplify how I work.

Stihl — Complete Redevelopment of a Chainsaw Engine Controller

For Stihl in Waiblingen, I completely redeveloped the engine control software of a chainsaw. It started with an analysis task: the existing control software had come from a Japanese supplier and was written in assembler. Stihl wanted to understand what this software was doing in detail. I analysed the assembler code and documented the functions in such a way that the Stihl developers could fully follow the logic.

On this basis I received the engagement for a redevelopment — this time in C, with the goal of delivering more functionality than the original software. The constraints were hard: the chainsaw market is extremely cost-competitive, so only a very small, low-cost microcontroller was available. Memory was so tight that a real-time operating system was out of the question. The software had to run bare-metal and reliably process real-time signals from the hardware at the same time.

Throughout the entire development, I had only the PCB with the hardware and my code on the desk — no test environment, no engine. The first practical test took place on the Stihl test bench in Waiblingen. After months of analysis and development, it would not have been unusual if the engine had not started initially. It started immediately and ran fundamentally smoothly. The remaining work was fine-tuning of the ignition timing, which we iteratively optimised on the test bench — a process normally handled by adaptive learning software, but here done manually for memory reasons.

In the end, I delivered a control software to Stihl that did more than the previous Japanese version, ran on the same cost-saving microcontroller and was handed over fully documented. Notable was the commercial frame: at the final contract negotiation in Stihl purchasing, the previously agreed price was not reduced. The buyer simply made clear that every Euro had to be traceable — but a price reduction was not the goal. This experience — a client who pays traceable prices for traceable quality — is rare in the market and shapes my understanding of good cooperation to this day.

The project was handled as a fixed-price contract. Development took place entirely outside Stihl's premises; coordination with the Stihl contact happened only on fundamental decisions. This form of cooperation has proven to be remarkably effective.

Audi — MOST Simulation and Proprietary Development of a CAN Bus Hardware

For Audi in Ingolstadt, my actual engagement was the creation of a MOST bus simulation software for an infotainment system — a real-time simulation that allowed control units to be tested against a virtual MOST bus.

On site, it turned out that Audi had a parallel second problem: an existing CAN bus simulation board, supplied by an external provider, did not work reliably under certain operating conditions — it simply stopped sending data at decisive moments. Since my MOST work had given me detailed knowledge of the interactions between CAN and MOST bus, the Audi contact asked whether I could rebuild this CAN simulation independently.

I agreed, with the note that the first step would be a working prototype — not pretty, but functional. In less than five working days, a self-designed electronics PCB with an ATmega128 was on the table, with simulation software written by me. The prototype was hand-wired but it worked — even under the operating conditions where the previous supplier's board had failed.

Audi then asked whether I could manufacture and supply units — but in industrial quality, fully machine-fabricated, not as handwork. I designed the layout for the production PCB and worked with a PCB manufacturer in the Audi environment who handled the machine fabrication. After resolving an issue in the manufacturer's machine test, the board delivered reliable results — exactly where the original supplier board had failed.

This project shows a case in which an engagement grew through concrete technical substance: the original software task became, in addition, an independent hardware development with series production — because I knew the bus architecture deeply enough to propose a working solution path on demand.

Other Clients

Beyond the projects described in detail above, I have worked for further clients in automotive, industry and medical technology. The selection above follows the criterion that the key facts are publicly representable. More concrete references for comparable tasks in your environment are best discussed in the initial consultation — under confidentiality if desired.

What These Projects Have in Common

Three properties run through both projects and through most of my work:

Independent end-to-end responsibility
Both projects started with an analysis task and ended with a finished, delivered solution. Specification, design, implementation, test and handover were in my hands — at Audi even including the hardware series production via a partner.
Hardware-near embedded work under real constraints
Stihl: bare-metal on a microcontroller with minimal memory. Audi: real-time MOST simulation plus proprietary CAN hardware. In both cases, this was not about standard applications but about tasks where the constraints shape the solution.
Cooperation as fixed-price contract with firm delivery commitment
Both projects ran outside the client's premises, with selective coordination on fundamental decisions. This form of cooperation is my preferred model — it protects the client from unnecessary steering work and gives me the focus that hardware-near development demands.

Color Scheme

Language