But the ZX81 was extremely limited, and other factors (like the keyboard
that didn't have real keys, and the method of putting the characters on
the screen, even the fact that the Z80 had built in hardware to refresh
RAM might have been a factor, I can't remember if it used static or
dynamic RAM) cut the cost of the unit so it could be sold
cheaply.
I paid something like $500 for my OSI Superboard II in the fall of 1981.
8K of RAM, a full keyboard, a video interface, a serial port and cassette
interface, BASIC in ROM. It used nothing but TTL for holding the main
components together. It was fairly big, which does impact on cost.
There were a bunch of iterations going on in those early days. Initially
a lot of things had to be done with TTL logic, lots of space, though not
necessarily expensive. Then larger scale ICs came along once the
companies could see a trend in what was needed. An early floppy disc
controller was pretty complicated, then made simpler when Western Digital
started making ICs (but still needed too many external components), then
not much more than the IC and buffers when Western Digital came out with a
revised controller.
In the same time span, RAM became more dense. The Altair 8800 only came
with 256 bytes of RAM, but not much later 1K x 1 RAM were relatively
cheap, then density kept going higher. Again, if you had to have a lot of
ICs for a specific amount of RAM, that used up space which ran up cost,
and increased current consumption, which meant a larger power supply. In
parallel with static RAM becoming more dense, dynamic RAM became more
dense, but also either became easier to interface, or designers became
better at interfacing them. Since dynamic RAM wsa always denser, it really
improved things, so by 1977, the Apple II could come with enough sockets
for a full 48K of RAM (the rest of the address space was kind of used up
by I/O).
Designers got better. The Apple II used the same circuitry to refresh the
dynamic RAM and to control the RAM for the video output. The computer wsa
pretty complicated in terms of circuitry, all that TTL, but it was simpler
than some of what came before.
Or take the VIC-20 and the C64, that came out about the same timein the
early eighties. The former used static RAM, which meant it didn't have
much, but by then it was a single 24pin IC, which beat the 16 16pin RAM
ICs in my OSI Superboard. The C64 went to dynamic RAM, allowing for 64K
of RAM with 8 ICs, but density of ICs had increased so the circuitry
wasn't particularly complicated.
The Apple II went through a few iterations, using custom ICs to lessen the
component count, but to a large extent it wasn't new design, just changing
the ttl ICs to one or two large pin custom ICs.
Some of what happened was there was a demand (or perhaps market created)
for cheap computers. So in order to address that market segment, costs
were cut, some of it was in less function (that 8K VIC-20) but some of it
was better manufacturing technique, including reducing IC count. Once you
can foresee selling a lot of an item, it becomes a different world when
you can only sell a relative few
Demand can drive things pretty fast, which may mean that nothing
particularly mattered in bringing cheap computers to the masses except the
idea, at which point everything was done to reach that market segment.
And once people were buying, that helped drive costs, and prices, further
down.
Michael