#English:
The tutorial : https://lebois-racing.fr/seat-belt-tensioner
#French:
The complete tutorial : https://lebois-racing.fr/tensionneur-harnais-simracing
The harness: 4 points are enough. At the time of tensioning, the belt goes up a little, but it's not annoying. 5 points would probably be better, but well you need to have a seat that allows it, and it's less convenient to settle down (we were doing relays).
The motor: An electric scooter motor. I took 350W, but it's too much, we used it around 50%. With more, it starts to hurt... and become dangerous in case of problem. Around 180W it's more than enough, especially since it's complicated to find a driver powerful enough for the 350W (it's usually motors used in on/off mode).
The driver is this one. We must not skimp on the amperage, especially since often, when an amperage is indicated, it is in the optimal conditions of use, namely immersed in liquid nitrogen. For example, the bts7960 burns out immediately when they try to drive this engine. With a less powerful engine it should be possible to use a less powerful engine too.
It is also necessary that the driver is of good quality, and "at ease" with the motor it drives because it must manage the pwm signal perfectly. If it takes 0.5s to get up to power, it's already too late.
On the PC side, I used simhub. We flash an arduino with the PWM signal function (which is usually used to drive fans to make a wind simulator). On the pwm pin, we then need to make an RC circuit (resistor and capacitor) to transform the pwm signal into a "continuous" signal for the motor driver.
I also mounted a potentiometer in parallel between the arduino and the driver to be able to set a pre-tension. Thus, when a new driver sits in the bucket, the belt tightens by itself. That said it consumes a little electricity, a spring system would be more economical.
Then I printed a winder for the motor. Beware the motor has a reverse pitch, so do not lose the nut, itself in left pitch, otherwise it's a shot to lose 10β¬ for a nut. I also tested a lever system, but it is less efficient. The smaller the diameter of the retractor, the more force there is, and the less quickly the belt retracts... Well, with my motor it was not a problem.
Note that the motor must be mounted on the moving part of the chassis (if you have a dynamic simulator). Otherwise, no matter how you mount it (with a thrust sleeve and so on), indirectly your motor pulls on the simulator, and intellectually it is not satisfactory.
I put a PVC bar mounted on a bearing to make sure the belt pulls the driver back. Feeling that, the rendering is a little less good. I wanted to put it on a bearing so that there is no friction that interferes with the belt (and thus the feel), and also to not damage the belt. That said, before you wear out a seatbelt...
Final rendering: frankly very nice. Then honestly the first braking, we say to ourselves "mouais bof, it was perhaps not worth to take the head like that...". After a few laps we forget it. Then it's when the belt doesn't work anymore (yes because at the beginning, there were unsuccessful tests ^^) that we can't do without it... we have again the feeling to play a video game, and not to drive a racing car...
The system is quite precise, it engages immediately, and you can feel the difference in braking between an F1 and a basic car, you can also feel the activation of the ABS.
It's probably not accurate enough to use the information to dose the brake like "ah it's not serving me right, I can brake more".
[media]https://youtu.be/egIHlUpW2mg [/media]