So I found the output to be 70 W. Which is very high compaired to the article (3 W). So I send an email to the manufacturer, they replied:

The power of the machine will not give you the power applied to your samples, since you do not know the efficiency of your machine. This is not determined, and there is no good way to determine this. Several normalisation-institutes have been looking for years to determine this.

So now im just wondering how the authors came up with the number.