Rishi wrote:Internally the separation between the A0-A8 and A0' and A8' isn't quite clear to me, why was it done that way to begin with ?
When the Z80 'executes the display file', pixel data that's sent serially to the screen, is grabbed from the data bus during Z80 refresh period. During that time, the Z80 has contents of I register on its high order address lines (A8-A15). So programmers can decide which memory area the pixel data comes from, by fiddling with the Z80
Interrupt register. Which is what you'll see in any display routine that puts something other than standard ZX81 character set on screen.
The lower order address lines (A0-A7) serve to select "which character?" from the ROM area that has the pixel data. But during refresh, Z80 has contents of R register on those lines. Which isn't useful for determining which character to display, since it's auto-incremented (+1) on each NOP+refresh cycle.
Those resistors allow the ULA to override the contents of the lower address lines (right before the pixel data is grabbed). What does the ULA put on those address lines? The character code that was grabbed earlier from the data bus during the Z80 instruction fetch that came before the refresh. And what 'instruction' was fetched? A character code from the display file!
(screen line within character (0-7) goes onto A0'-A2'). But immediately after, ULA forces the data lines low, so the Z80 sees and executes NOPs instead.
For clarity: the ULA doesn't care whether Z80's I register points to an area in ROM, or an area in RAM. That's the programmer's problem...
Likewise, the ULA doesn't care what pixel data it grabs, or whether it comes from ROM, RAM, or a floating data bus. But there
IS a difference there: if it came from the
internal ROM, the lower address lines were set by the ULA (A0'-A8' over-ridden since 'behind' the resistors). If it came from RAM (or some external memory), the lower address lines were set by the Z80 (!, not over-ridden by the ULA since RAM A0-A8 are 'before' the resistors).
So you'll see that what pixels actually end up on screen, is a
very, very intricate 'dance' between ULA, Z80, software, and the ROM/RAM memory layout. Again: as soon as you modify standard ZX81 display routine, what happens is the programmer's problem.
also i was wondering if i could recycle some of the 4116's (..)
Unless you like pain, "because I can", or meant to provide 'public value' (like: show a machine in a museum, that both works and uses actual 16K*1 chips): no point
whatsoever imho. Something along the lines of "
Total
Cost of
Ownership" (or in other words: simply not worth the trouble).
A static RAM, 64K*1 or even 16K*4 DRAMs (4416 type) are
ALL better options than those nasty 4116 chips. Save those for a museum if you want.
Especially given the low cost of replacement chips, and how -relatively- easy it is to hook them up.
But any help is always appreciated.
Standby alert 
) i am anyway considering re-using the box and edge-connector to make another 32K SRAM for a friend of mine. How hard is it to use the 4116's i am actually fairly sure all 16 of 'm are still ok, but there is something else wrong with the memopak causing a strange video pattern. I just don't like throwing stuff away, particularly semiconductors are not recycle-able unless as parts. (if you melt the stuff you get left with a non usable mixture of metal and silicate 
)
