Car testing#

First test#

The engine was fired up for the first time in June 2024. After that, a lot of work remained on the hydraulics, steering and brakes. To make things more interesting, the slave clutch cylinder failed — despite being completely new — and the union between the engine and the gearbox was leaking. As a result, the engine and gearbox had to come back out and the slave clutch cylinder had to be replaced. Plenty of extra work, but we used the opportunity to make sure everything was properly assembled and avoid further surprises.

After a lot of work on the hydraulics, the electrics and the gearbox problem, the car was mechanically complete by the beginning of October 2024. Once the steering and gearbox had been tested with the front wheels off the ground and everything checked out, it was time for the first test: driving forward and backward inside the garage.

Keep in mind that the car has not gone through the MOT-like inspection (ITV in Spain) for the last three years, and the engine has been modified, so for the moment the car is not road legal and cannot be tested on the street. Because of that, we had to run the first test inside the garage.

Once the body shop finishes the car in the following days, the idea is to take it to a closed road for testing and see how it behaves. For now, we are just excited to see the car move.

Further tests#

The first road tests did not go well: the engine was running rough, and we soon discovered a misalignment between the ECU software and the engine modifications. This was somewhat expected, since we had introduced custom engine modifications — a K04 turbo compressor, a 3-bar MAP sensor and S3 injectors.

ECU tuning session while diagnosing the rough-running K04 conversion

Our solution was to grab an original BWA flash image (aligned with our ECU hardware) and use ecufiles.com to create a Stage 3 tune featuring the K04 turbo compressor, MAP scaling, and fuel injector scaling. But it did not work—the engine continued running rough. Almost a year passed. During that time, we tested multiple combinations, but nothing worked. Then, by luck, a Golf Mk5 EOS appeared with a BWA engine—the same engine code we are using in our swap—so we decided to try its ECU and software.

It worked. Suddenly, the engine was running fine. It still needed tuning, but at least it was running properly. Finally, we had a solid foundation to build on.

This time, we did not take chances. We sent the car to a professional tuner for the Stage 3 calibration.

The tuning process involved more than twenty dyno runs while progressively adjusting and validating the ECU calibration. During the runs, knock, lambda, boost pressure, engine load and temperatures were monitored to make sure that the engine was not only producing the expected power, but doing so safely and consistently.

The graph below shows the final dyno run.

Dyno graph of the final power and torque run: 273.2 hp and 445 Nm

The final result was 273.2 hp and 445 Nm.

The objective was not to extract the maximum possible power from the K04. This is a road car built around a 1996 Golf Mk3 chassis, so we deliberately chose a conservative calibration, prioritising reliability, predictable power delivery and thermal stability over the highest possible dyno figure.

The shape of the curves is probably more important than the peak numbers. Torque builds quickly and remains strong through the useful part of the rev range, while power continues to increase progressively towards the top end. For a road car, this should provide a much more useful result than simply chasing a larger peak-power number.

The dyno sessions also provided an unexpectedly useful test of the rest of the conversion. After more than twenty high-load runs, the cooling system remained completely stable, giving us considerably more confidence in the oversized radiator, intercooler and fan arrangement.