.....and what might work well in one type of ground may not work well elsewhere. In many cases, that "tweaking" amounts to nothing more than adding more in the way of user adjustability, such as manual ground balancing, adjustable threshold level, sensitivity or gain, discrimination vs. all-metal modes of operation, and so on. Simply stated, the more "operator control" you have over a machine, the more "tweaking" you can do without doing any kind of modifications to the internals. Some of the additions to certain machines amount to nothing more than adding such control where it was not provided on the machine "by design", at least not externally. Good examples are some of the manual ground balance functions added to machines with otherwise internal adjustment pots. Another is the simple exchange of certain components on some of the PI type machines that shorten the gate pulse delay to make them more sensitive to smaller items with less cross sectional mass (the Whites surfmaster PI was a good candidate). It's not always so much a matter of increasing the inherent capabilities of the machines, but increasing or simplifying the level of user control over the machine to do the "tweaking" externally according to the user's specific needs rather than having to depend on what the factory "thinks" is the "best average setting" for the greatest number of conditions.
But as far as actual component changes (other than adjustable controls), the operational stability of the machine is more often the primary concern. Any tradeoff in performance for stability can really be a two-edged sword. And when you're talking about an increase in target response especially, that usually includes ALL targets, including junk and ground signal.
Just a thought......