each beach is going to be different. What I am going to offer is not guaranteed to resolve the issue, but to provide possible clues toward some solutions.
1. GB on a perfectly smooth, target free area. Put machine in AM and bobb coil from 1 inch to 10 inches high, like you were GB'ing, and note if detector reacts with a (-)9 or +45. If yes, reduce sensitivity until you get no response in AM. You may end up quite surprised about how low you need to go, possibly down around +10.
2. Now dig an empty hole about 8 inches deep, 6 inches round, and bobb coil in AM, and once again note at what sensitivity you get no response at.
3. Now this part is more experimental, scrape away about 6 inches of surface sand in a 2ft circle till you get down to the damp sand in a target free area. Now GB once again, and note the GB setting as it compares to the surface GB number, and then bobb the coil in AM to find a sensitivity level where it doesn't respond.
4. Also, are you noting when you dig targets that there are layers of black sand? So that the black sand is stacked like a cake as you go down through? Which will raise heck with the GB setting.
My overall guess is that you may be running the sensitivity too high. And the combination of salt & black sand may need a dual freq to prevent the target ID shift. Or there's enough black sand to chew up signal to prevent detection beyond 5 to 6 inches. Kind of like detecting inland at 9 to 10 inches, where you are at the limits of what the circuitry can reliably ID, so it throws up a 45, which I guess kinda says, there's something down there but I don't know what it is.
Please get back to us(the forum), because this is something interesting that doesn't get discussed much here. Too many dang relic hunters on here I reckon.
HH
BarnacleBill