You don't need two tests, just data that contains both a 5 min and a 20 min max. When you ride with power and collect data over time, these numbers are automatically tracked for you by many data analysis programs.

The important thing to realize is that all of these formulae are going to assume a power declination curve for a "typical" cyclist. But truthfully there's enough variation amongst cyclists that there's a good chance your power declination curve is going to be different than the formulae assume- I know mine is.

The way around this- sort of- is to just ride a one hour test. But most people aren't used to one hour max efforts, they're painful and they take practice to get right. Also note that I say "sort of". Because the definition of FTP is based on an assumption that there is one power value that everyone can hold for exactly one hour. But even this isn't true for everybody. Not to mention that there can be differences in FTP depending on what muscles you engage- different muscles use oxygen and clear lactate more or less efficiently- which means you might have a different FTP climbing or on a TT bike vs road bike.

Bottom line is that FTP is something of a squishy concept. Don't take any number you generate as sacrosanct. Because of all that, my opinion is to use the easiest formula based on the longest test you can complete well. For most people, this is 0.95 x ave power on a 20 min effort.