Why this FCC thing keeps coming up is beyond my thinking. There is indeed a Part 15 which covers incidental radiators. In other words, if an electronic device emits a radio signal (again that's RADIO), but isn't (wasn't) designed to be one, then Part 15 applies. A typical example of this is a home computer. Some aren't shielded very well, and they radiate hash which can be heard on any nearby radio. Some digital TVs, and even electronic clocks can do the same.
As for a metal detector. The CPU inside is indeed a computer, albeit small in scale. That part of the device is covered by Part 15. You can hear some of them using a radio receiver, but I'd bet they all meet Part 15 rules. However, trying to apply this to the coil itself, is ill placed.
The highest frequency modern metal detectors use, is just below 100 kHz. That's 100,000 cycles per second. However, any frequency below 100 kHz, DOES NOT fall under Part 15.
Most detectors nowadays, operate between about 5 kHz and 15 kHz. That falls in the audio spectrum. Some, not so well made detectors, can actually be heard if your hearing is very good.
A better example is the Garrett 2500. Its mean coil frequency of operation hovers around 7 kHz. In case you missed it, this is in the audio spectrum, not the RF spectrum! So, Part 15 doesn't apply! Period! But you still can't "hear" it, as the coil is very well supported so it can't physically oscillate.
As alluded to above, the amount of power applied to the transmit coil, doesn't have a linear correlation to the detection depth. Even increasing by a factor of ten, won't guarantee any improvement in depth. Remember, we have to also receive the reflected signal, and the more power that's applied to the transmit search coil, the more difficult receiver design becomes. And, we have to balance cost, weight, battery life, and a few dozen other things in the process.
I'm of the opinion that the maximum detection depth, for a hand-carried detector, is already in our hands. The next step is interferometry. Once low power consuming CPUs get powerful enough, we will indeed see the technology. Whether we can afford it, remains to be seen.
Alan Applegate, K