That shows one of two things: (1) Tedious calculations shoudl always be left to computers, or (2) the code does not do what you want it to do.
We do not know what the expected results are. We do not know what the permissible ranges and combinations of input values are. We do not know what this code is supposed to do.
So. We cannot tell whether the algorithm is correct. We cannot tell whether the algorithm accounts for roundoff correctly. We cannot tell which these versions is correct or whether they are all wrong.
Here is my VI showing how I implemented my interpretation of your formula nodes in LV primitives. As you can see there are multiple versions of some parts where I tried variations in representation and algortihm. Note that where comparing floating point numbers I used In Range & Coerce with a tolerance of +/-1E-6 (arbitrarily chosen). See maxFFcheck as an example. I did not clean it up or document things. If you have questions, please post them. I also added some extra outputs and indicators in the formula nodes for troubleshooting purposes.
Lynn
