Hello,
I think that the difference comes from the calculation of the circle of confusion. It depends of the sensor size/resolution, and the human eye visual acuity. This is a subjective value but some standard rules exist.
In the tool that you are talking about, I think that "d" represents the circle of confusion. It is automatically set to a very different value than in WinMoDof. When you set it at the same value a in WinMoDof camera database then the results are quite the same (the difference is about 3-4%). I don't know why the d value is so different, and also why there is a 3-4% difference in the calculation. Need to investigate.
I tested WinMoDof with this web site:
http://www.dofmaster.com/dofjs.html and the results are coherent.