Re: about the plus4

On Fri, Apr 20, 2012 at 12:11 PM, Gerrit Heitsch
<gerrit@laosinh.s.bawue.de> wrote:
> Yes... Due to the way the 6502 works, using 450ns 2114 was very close even
> for a 1 MHz system (it's too slow for a C64 though)

Too slow for memory that was getting hit on both phases of Phi/2, yes,
but by the early 80s, one could get a 250ns 2114 for not too much more
than the 450ns part.

> Today that's a joke, but look up the price for a
> 6116-200 (2Kx8 CMOS SRAM, 200ns) in 1980.

Yes.  Not exactly cheap.

>> By the time the 6567 was designed in 1981, DRAM prices were coming down
>> low enough to put 64 KB in the machine and the badline design allowed it to
>> be slower RAM.

Yep.

> Using the 4164 DRAM also meant that you didn't need the extra voltages. OK,
> the VIC and SID needed +12V like the 4116, but at least you didn't have to
> deal with -5V for the 4116. Ask ZX spectrum owners how much headaches the
> 4116 and their support circuits can cause.

Dynamic PETs used 4116s, as did a number of 64K expansion boards (CBM
and 3rd party).  I don't recall any specific problems with the -5V
supply circuit
in the PET, but they did have this honking-big transformer and a tap to run
the negative supply (and just a simple LM7905 + caps regulator circuit).

What would have been an issue is a compact design with space taken up
by a pad of 32 4116 chips compared to the space taken up by 8 4164s.  In
the early days, 64Kbit DRAMs were not cheaper than four 16Kbit DRAMs,
but the line on the graph was clearly heading in the right direction, and the
amount of real estate was clearly a win from day one.  The PET design
(video + 3 VLSI I/O chips + large ROM bank + IEEE-488 + dual tape
control) is already so busy that 16 RAM chips kinda just hang out on
one edge.  Bumping that design up to 32 RAM chips wouldn't be
insurmountable.  By the time the C-64 was being designed, it was a
much more compact video circuit with higher-density ROMs and one less
VLSI I/O chip, so the RAM field impact rises.

-ethan

       Message was sent through the cbm-hackers mailing list