George Payne wrote a bit about this but its a little brain busting. The returned signal (to the coil) has two parts, X and R. The polarity of X tells the machine if the target is ferrous or non ferrous. The R signal has only one polarity. The ratio of X to R gives the targets phase. Then add that the signal strength/sensitivity of both X and R are a function of the frequency used
For all metal use you only need R but for discrimination X has to be measured. They find the -3db frequency, which is the frequency where the R signal is strongest. This is different for all metals. The example he gave was that a silver dollar has a -3db of 800 Hz while a nickel has one of 17khz.
Add discrimination and you have to use a frequency above the targets -3db for best performance.
To sum up the ideal frequency for each metal is different so you have to compromise. (Thanks to George,Reg etc who have tried to educate me on this over the years)
But in general, high frequencies work best for low conductivity targets and low for high conductivity. In general because you have to consider how the machine is set up. The DFX works best for our thin U.K. hammered silver at 15 khz but on larger silver on very bad ground its lower frequency should be used as the effect of the mineralisation may shove the LCD reading into the reject figures so you would think its rubbish and ignore it.
