Jeff, you mentioned 4-Tone so I presume you're using a Makro Kruzer series model, correct.
Question: Which search coil are you using?
Question: I'm curious, why are you using '0' Discrimination?
Question: You gave the GB reference, but what kind of ground are you searching in?
Question: How long have you been using the Kruzer, and how many hours have you put in so far?
Question: Why did you fresh-bury a penny at 2"-3"? I presume you are referring to a US 1¢ coin, so what type is it? Older 'bronze' type or a modern 'zinc' variety?
Question: Presuming you scanned the area to know it is free of a masking metal target, did you place the coin very flat in the bottom or the hole so it wasn't canted at an odd angle? Did you leave the hole open and simply scan across an open hole with the penny 2"-3" deep, or did you fill the hole with the removed soil? If you filled the hole, did you pack the soil so as to be similar to the adjacent ground looseness or compactness?
Answer / Comment: It is not uncommon to get a different audio response, or even a broken signal or an erratic response from either a coin in a fresh dug open hole or in a filled-in fresh coin planting. The reason, or multiple reasons, can be the change in the ground mineral make-up due to the open but visible coin in a hole or the filled hole with disturbed ground. Digging and burying a 'fresh planted' target will alter the ground mineral signal the metal detectors have to deal with. That ground signal processing combined with a disturbed environment plus the metal target can be a challenge for many/most detectors.
Question: You stated" "
I'm coming from a Deus, so I am accustomed to variance on VDI numbers, but this is incomprehensible." Have you used other detectors with VDI numeric read-out? Were they, as well as the XP Deus, radically inconsistent on a fresh-planted penny, and in similar mineralized ground? How about VDI consistency on finds of long-lost coins?
Answer / Comment: In typical searches on a wide-range of ground environments that were somewhat low on nearby masking trash, I have used some makes and models that were not/are not very pleasing when it came to having a stabile Target ID read-out. However, I am pleased to say, that I currently own and use an assortment of devices that provide me very consistent visual read-outs on the more consistent targets we search for, such as coins. I have enjoyed very clean audio and visual responses with Makro's Racer 2 and Multi-Kruzer, and that was using an assortment of search coils.
However, let me clarify one thing with regard to search coils. Very, very seldom do I use a 7X11 DD coil. By 'seldom' I would guess that with that size coil on any Nokta, Makro or other brand model, I
might, .. possibly, ... maybe, use that size coil 2 to 4 times a year and, at the most, total 6 hours a year. The bulk of my detecting is spent using mid-sized coils like the 7" Concentric on a Makro, 5X9½ Double-D on a Nokta CoRe or Relic, or smaller-size coils to better deal with dense brush, building rubble, common trash and ample ferrous debris. Using the smaller to mid-sized search coils is where I enjoy excellent audio responses and visual display.
Thoughts on the matter .. 
• Use a low-level of Discrimination, such as '03' which is my default setting.
• Forget fresh-planted 'tests' and, instead, just rely on using the detector and search coils you have to search for all naturally-lost targets. Natural ground environments for any area without fresh disturbance just to plan a target, and monitor the responses you get; analyze differences in target response using different sweep speeds; carefully recover the targets and note their depth, and position or orientation to the coil and their physical shape. In time you will better understand any detector's capabilities, both strengths and weaknesses.
• When you question your detector's behavior, compare the response, on a freshly planted coin or a naturally lost object,
before disturbing the ground, using a different make or model detector with similar settings and a similar search coil. Then you'll have a better idea if what your Makro model is doing is 'normal' or way out-of-whack. My experience is it will be working just fine.
• I usually hunt higher mineralized ground than the GB reference you gave, but there is more involved than just the ground phase value. You might be experiencing some ground environment challenges or, as Tom Slick mentioned, there's something wrong with the device or coil.
I enjoy my Regular-Use Detector Team and have very reliable performance with my Nokta and Makro detectors. Very solid, dependable detectors.
Monte
