Indeed, that is the only reason that I can think of, the requirement for the RAM to be fast enough for the video/refresh access. In order to keep costs low, a slower chip was permissible for the other data bits. That's why some instructions say it's okay to fit the faster chips in both positions.MaFli wrote: ↑Fri Jun 02, 2017 6:15 pmEach 1K of RAM consists of 2 of these chips. One for the high nibble, the other for the low nibble. But why is one half done with an 2114LC and the other with a 2114LC-1. The -1 version has a 300ns access time, whereas the unmarked is 450ns. Same was done in the original ZX80.
The only thing I can think of, is because of dataline D6' that is used in the NOP code generation during video display and has to be fast enough. That's probably also the reason why it is the only dataline with an additional pull-up resistor to speed it up a bit more. The slower device is less critical and was probably cheaper.
If you have both types of 2114 SRAM, you could try swapping them around in both the ZX80 and the RAM expansion...