Here is the difference between the real ARK-5 device and the simulated one:
If any of you had ever seen a pre-WWII radio, you would have noticed that they were big, heavy and had buttons for different frequency ranges, usually SW and MW.
An old family friend had one. That specific radio had buttons for SW1, SW2, MW1, MW2, MW3 frequency ranges, with a single frequency tuning knob. That radio must have cost a pretty coin in its day.
Anyway, in the back I noticed that the tuning knob moved a series of disks. Each disk corresponding to a frequency range.
The knob moved all these disks at the same time, and the buttons selected which disk was doing the frequency tuning.
All the disks began at the same 0 point, but the disks did not have the same circumference range. They had notches where the frequency range ended.
Here is the problem with the simulated ARK-5. We have the same disks, but since we do not have a sample device, all these disks are the same length. So what happens when you move a specific frequency disk, like the 310-640? So if the disk was at the middle of the frequency range for 310-640, which should be 475Khz, that middle position is not exactly the middle position for the 640-1300 and the 150-310 frequency ranges. But, since we are assuming that all three disks are the same length, we calculate the middle frequency for the other two ranges: 150-310 and 640-1300. But these calculated frequencies are not exactly the real ones, and thus you need to fine tune.
My decision to implement the two memories was based on this: the need to fine tune the NEAR frequency due to the inaccuracies in the frequency calculation. In real life, there was no need to do that.