Hi, the trigger wheel looks a bit thin (what is the thickness?) and there is steel quite close to the sensor on the side.

Are you able to move the sensor and/or use a thicker wheel?

The point you make about the curves being smooth in one video and not in another is just because of the RPM noise filtering. There is a clever adaptive noise filtering algorithm, but it is just too much for it sometimes. That is what is causing the jumps too. The log file you sent shows noise on all RPMs. You may not see it on the oscilloscope unless you trigger on abnormal pulse length. 

Hi Jostein@admin. Please find attached a pic of the second trigger we tried. This used the bolts on the brake as a trigger, same result with the noise.

Hi Jostein@admin. We tried a basic sensor test, please find video results attached. Sensor was bench mounted in a vice and triggered by touching with a pair of pliers. When the pliers were removed from the sensor, the speed reading intermittently went off the scale.

You cannot use a "wheel" like that (some screws) and you cannot test the RPM with the pliers like that, other than seeing that it triggers. 

Try what I proposed: A thicker wheel and further away from the heavy steel on the side. If that does not work I propose you try the YourDyno RPM sensor!