I haven't reviewed all the posts, but I can't think why anyone would suggest that you use two ground layers. A solid power layer will have similar signal integrity characteristics. If you have more than one power rail, you can divide the power layer into "zones" for each power rail. In some designs, you also need separate ground areas for analog and digital grounds (joined at one point, at an edge of the ground area). So, one ground layer and one power layer.
With adequate bypassing, the power planes will act the same as the ground plane, establishing the impedance of your signal traces.
Ok well maybe you should read the whole post because what you saying about splitting ground planes causes more harm than good if not done exactly correctly and its a really outdated way of doing it. Watch that video that was recommended its really good.
"if not done exactly correctly"
You seem to be making a statement, but then you completely undercut yourself by saying it will work, but only if the designer is not crap.
The idea of isolating analog and digital ground planes is sound. Digital electronics can produce huge ground currents, such as the impulses from a switching power converter. No matter how well you design that circuit, it will have very large ground currents that will create ground noise, disrupting sensitive analog circuits. So, the two need separate ground areas, connected at a single point. Any digital signals that cross between the two areas, should do so as close to the point of connection as possible.
I've used this approach in any number of designs, and it works.
As previously mentioned in the thread you admitted you didn't read, and covered extensively in the very good video Doctorandus_P shared.... High frequency return currents (above audio frequencies) are concentrated in the path of least inductance. With a track over a solid ground plane, the return currents take the same path as the track except through the ground plane, because that's the path that encloses the smallest loop area. At best, splitting the ground plane is not effective for controlling high frequency return currents, because with a solid plane they already take the optimal (least impedance) path available. At worst, splitting the ground plane can cause serious problem with EMI, SI, and basic function by forcing HF return currents to flow around--or even couple across--the gap in the plane, increasing inductance, and spreading fields out further across the board and beyond where they can impact more signals. As you note, signals should be routed across the junction point to avoid this, but that also forces the signal and its return current closer to the sensitive ADC/DAC--or you have to make that junction area larger, stretching the definition of "single point", to preserve signal separation without tracks spanning the split. So, bottom line, split planes can work, but they're not going to be better than a single solid plane for high frequencies, and will cause more harm than good if not done exactly correctly.
Splits/Slots can be effective at controlling lower frequency return currents (in the audio range down to DC), but that's only relevant if you have enough current in those lower frequency ranges to cause meaningful voltages across the DCR of the return path. "Meaningful" of course depends on the sensitivity of the circuitry involved as well as the magnitude of the currents concerned.
I don't know why you have ignored what I wrote. Ground planes must be connected. The signals can be routed over the point of connection between the two areas, to allow the return currents to follow the same path as the traces. This is easy to do properly. Just route the "high speed signals" along the path the return currents will follow. It is seldom required to minimize the length of traces crossing from one area to another. The signals just have to cross between the planes at the point of connection.
What the split plane is better at, is limiting the impact of high current, RF spikes, that are created by one circuit, from crossing the split, and impacting another circuit. A perfect example are the large transients generated by a switching power supply. By giving this section of the design a separate area of ground plane, the current spikes are limited to that area, preventing the currents from flowing over the other parts of the ground plane, where the IR drops can cause excessive noise.
If there is a section of analog circuitry that can be disrupted by digital noise, the semi-isolated area prevents noise from entering. Think of it as an iso-potential ground area.
If you don't believe this, consult any good reference on designing such switching power circuits, such as an LT (now Analog Devices) data sheet. They offer very useful design guidance and theory of operation information.
You are only talking about the propagation of high speed signals
across a split in the power plane. This is not what I am proposing.
If you don't understand what I am describing, say so, and we can stop discussing it now.