A detectors operating frequency determines several things.
First is its sensitivity to particular conductivity levels. Lower frequencies are better able to handle harsh iron ground mineralization, and in general terms, are more sensitive to higher conductivity metals such as copper and silver. But don't confuse "sensitivity" with "detectability". Sensitivity is most important when dealing with fringe "depth" and fringe "size" targets, those that are just on the threshold of being detectable and not being detectable. In that case, there is an optimum frequency depending on the conductivity and conductance of the target. Suppose you have a target that is silver, and it is just on the edge of detectability. In this case, a lower frequency will make the difference between being able to detect that target or not.
Higher frequencies on the other hand are more sensitive to low and mid-level conductivity targets like brass, lead, and gold alloys. Note here that there is no such thing as "pure gold" in nature (raw nuggets) and that most all gold jewelry consists of alloys, all of which fall lower on the relative conductivity scale than pure gold. In the case of natural gold nuggets, higher frequencies are able to detect smaller nuggets. But frequency is only one concern on gold-specific detector designs. RX gain is another that is very important, as is the detectors ability to be "accurately" ground balanced. One problem with higher frequency machines for prospecting work is that they are not only much more sensitive to the little gold nuggets, they are also much more sensitive to the ground mineralization, so a more precise GB system is usually incorporated to such machines. A good example is White's 40:1 GB control ratio, or even their later 4000:1 "ramping" system.
As far as general coinshooting is concerned though, you are not likely to see any major difference between machines on coin-sized and larger targets that fall well within the range of detectability of the particular machine.
Another consideration is depth of detection. A lower frequency machine will tend to detect deeper in iron mineralized ground, while higher frequencies are somewhat thwarted in the depth department in like ground due to their increased sensitivity to small targets, AND ground mineralization, whether conductive or magnetic. But, on the other hand,higher frequency machines can often be operated at higher gain levels that can level the playing field to some extent where depth is concerned.
Take that 3 kHz machine and the 30 kHz machine nugget hunting in some of the goldfields of the western states, and all other design factors being equal, you can easily prove my point. If the frequency didn't matter, then any frequency in the stated range would be fine for everything. This may be true of coin-sized targets at "detection depths", but not when we start considering fringe "size" and fringe "depths" of particular metals. George Payne wrote a discussion of optimum frequencies vs. specific coins awhile back that covers this topic to a "T". (exerpt posted below) But getting down to the nitty-gritty, there is always one "optimum" frequency for any specific metal, just like there is always one "optimum" tuning or adjustment point for any particular machine for a given set of conditions. But the greater the "detectability" of any specific target, the less imporant this "optimal" frequency or tuning adjustment becomes. I guess it's really a matter of what level of detectability is involved. For example, even the cheapest Radio Shack or foreign made detector can likely detect a one ounce gold nugget at 2 inches (extreme ground conditions notwithstanding), regardless of the operating frequency involved. But try the same thing with a "nugget" or speck weighing in at a fraction of a grain (1/480th of a troy ounce), and the benefits of specialized frequencies and other design considerations quickly become apparent. There is good reason for 50 or 71 kHz specialized gold detectors being able to detect smaller nuggets than a Radio Shack special. But like has already been mentioned, those design factors (as well as others) become much less critical for coin-sized and larger targets that fall well within normal detection range. There is also good reason why lower frequency machines get better depth and handle harsh ferrous mineralization better than high frequency units. But considering that there is such a wide range of differences in metals that we wish to detect, the best we can really do is find a "happy medium" or good average frequency base to work with, and then adjust other design parameters around that frequency for the best performance possible, either for the specific types of targets sought (specialized gold machines) or more "general-purpose" detecting. Understand too, that those multi-frequency machines claiming a wide range of operating frequencies are effectively modified pulse induction designs that perform poorly in detecting very small gold targets, regardless of their high-end frequency. This is something inherent in the PI-based design.
-----------------------------------
from a George Payne article.
"The target signal returned to the receive coil can be thought of as composed of two components, one we call x and one we call r. The polarity of the x signal (its direction) tells us if the target is ferrous or non-ferrous. The r signal has only one polarity. Also, the ratio of the x and r signal tells us the target
