Re: Dynamic RAM / memory expansion

From: Nick Coplin (cbm_mail_at_lycos.com)
Date: 2001-01-19 12:05:28

Hello Ruud,  (this is going to be a log reply....)

The other answers on this topic explain generally how RAM is refreshed.  There are three typical schemes, the C64 uses the simplest which is basically that the VIC-II reads the RAM using its internal counter.  Every raster line 5 low-order (0-7)address lines are read (and hence refreshed) before the fetch operations are made for the character /sprite data etc. The VIC.TXT document floating on the internet explains this.

6164 chips need 7bit refreshing, 41256 need 8-bit refreshing, the 1meg chips need 9-bit refreshing etc.... The VIC-II fortunately provides 8-bit refreshing. Hence 41256's will work. (the "7bit" refers to the size refresh cycle, 128 cycles required, or pins A0-A6 have to be cycled)

However, 4464 chips (or 41464) used in the C64c require 8bit refreshing, and so on +1 over the above requirements.  The extra "width" of the chip requires an extra bit for cycling because the number of cells is the same in a 41464 as it is in a 41256). Hence, the C64c works with 4464 chips because of the VIC-II's A0-A7 refresh counter.

Now...  30pin 256kB SIMMS unfortunately generally use two 44256 chips (and sometimes one 41256 for parity). The 1MB SIMMS use either 8 or 9 1MegaBIT chips.  The problem in both cases is refresh will not work on all cells.

Having given some though to this before I found several tricks the Z80 guys use to get around this problem (the Z80 has an internal 8bit refresh counter).  The solution I think for the C64 is to use the BA signal (which goes low before the refresh /fetch cycles) to control a counter.  The higher bits of this counter can provide higher order bits for the SIMMS, but only for the first 5cycles of BA=low). After the 5th byte, the normal "VIC page" address needs to be shown in order to fetch the correct video data.

So... with a counter, multiplexer and some glue, RAM with greater than 256kilo-bits could be refreshed (btw: the time between refreshes doubles as memory size quadruples, which is very handy). I haven't done more other than develop this theory.... so I haven't got a circuit in mind.

Hope this helps.

Regards, 
Nick

PS: to expand the VDC RAM to 256kB, one would use a board eight 41256 chips for this reason.

--

On Thu, 18 Jan 101 19:03:35   
 g.baltissen wrote:
>Hallo,
>
>I've expanded some C64's and C128's by replacing the original 4164 by 
>41256's. But I have some 30 pin 256 KB and 1 MB and even some loose 72 pin 
>4 and 8MB modules laying around. Who has experience in connecting these to 
>a C=? The problem is that I have no idea if the refresh used for the 
>4164/41256 is good enough for the bigger RAMs. I can imagine that they need 
>more addresslines to used during a refreshcycle.
>
>--
>Groetjes, Ruud
>
>http://home.hccnet.nl/g.baltissen/index.htm
>
>
>
>-
>This message was sent through the cbm-hackers mailing list.
>To unsubscribe: echo unsubscribe | mail cbm-hackers-request@dot.tml.hut.fi.
>


Get your small business started at Lycos Small Business at http://www.lycos.com/business/mail.html
-
This message was sent through the cbm-hackers mailing list.
To unsubscribe: echo unsubscribe | mail cbm-hackers-request@dot.tml.hut.fi.

Archive generated by hypermail 2.1.1.