The CPC6128 second bank of 64K RAM is controlled by a PAL 16L8 chip. It has the Amstrad part number 40031.
On the CPC 6128 schematic, it is top centre: CPC6128 Schematic however the X inputs aren't distinguished.
Fixed version (Gerald)
Original version from Porchy suffer from a bad handling of the RAMDIS signal. This cause screen artefact when accessing an external extension RAM like XMEM.
A14OUT = !( !A14
# !A15 & !Q0 & Q2 );
A15OUT = !( !A15 & !A14
# !A15 & !Q1
# !A15 & !Q0 & !Q2 );
Q0 = ( D7ANDD6 & nRESET & D0 & !A15 & !nIOWR
# !D7ANDD6 & nRESET & Q0
# nRESET & A15 & Q0
# nRESET & nIOWR & Q0 );
Q1 = ( D7ANDD6 & nRESET & D1 & !A15 & !nIOWR
# !D7ANDD6 & nRESET & Q1
# nRESET & A15 & Q1
# nRESET & nIOWR & Q1 );
Q2 = ( D7ANDD6 & nRESET & D2 & !A15 & !nIOWR
# !D7ANDD6 & nRESET & Q2
# nRESET & A15 & Q2
# nRESET & nIOWR & Q2 );
nCAS0 = ( nCAS
# RAMDIS & !nCPU & nCAS0
# !A15 & A14 & !nCPU & Q2 & nCAS0
# A15 & A14 & !nCPU & Q0 & !Q2 & nCAS0
# !nCPU & !Q0 & Q1 & !Q2 & nCAS0
# !nCAS1 );
nCAS1 = !( !RAMDIS & !nCAS & !A15 & A14 & !nCPU & Q2 & nCAS0
# !RAMDIS & !nCAS & A15 & A14 & !nCPU & Q0 & !Q2 & nCAS0
# !RAMDIS & !nCAS & !nCPU & !Q0 & Q1 & !Q2 & nCAS0
# !nCAS & !A15 & A14 & Q2 & nCAS0 & !nCAS1
# !nCAS & A15 & A14 & Q0 & !Q2 & nCAS0 & !nCAS1
# !nCAS & !Q0 & Q1 & !Q2 & nCAS0 & !nCAS1 );
File:CPC6128.JED : Fixed version of Amstrad 40031 GAL replacement
File:CPC6128.hex : Fixed version of Amstrad 40031 GAL replacement, Hex Intel version.
Initial replacement equation (Porchy)
The following equations were worked out by Porchy (member on CPCWiki Forum). These can be used to program replacements:
A15OUT = (!X2 & !X1 & A14
# !X3 & !X2 & A14
# A15);
!X1 = (!A15 & D7ANDD6 & RESET & !IOWR & D0
# !X1 & RESET & IOWR
# !X1 & !D7ANDD6 & RESET
# !X1 & A15 & RESET);
!X2 = (!A15 & D7ANDD6 & RESET & !IOWR & D1
# !X2 & RESET & IOWR
# !X2 & !D7ANDD6 & RESET
# !X2 & A15 & RESET);
!X3 = (!A15 & D7ANDD6 & RESET & !IOWR & D2
# !X3 & RESET & IOWR
# !X3 & !D7ANDD6 & RESET
# !X3 & A15 & RESET);
!CAS1 = (X3 & !X1 & A15 & A14 & !NCAS & !RAMDIS & !CPU & CAS0
# !X3 & !A15 & A14 & !NCAS & !RAMDIS & !CPU & CAS0
# X3 & !X2 & X1 & !NCAS & !RAMDIS & !CPU & CAS0
# !NCAS & CAS0 & !CAS1);
!CAS0 = (X3 & X2 & X1 & !NCAS & !RAMDIS & CAS1
# X3 & !X1 & !A15 & !NCAS & !RAMDIS & CAS1
# !X3 & A15 & !NCAS & !RAMDIS & CAS1
# !X1 & !A14 & !NCAS & !RAMDIS & CAS1
# !X3 & !A14 & !NCAS & !RAMDIS & CAS1
# !NCAS & !RAMDIS & CPU & CAS1
# !NCAS & !CAS0 & CAS1);
A14OUT = (A15 & A14
# !X1 & A14
# X3 & A14);
File:Amstrad6128.jed Original JED File posted on CPCWiki Forum
PAL MMR register
This register controls how the extended RAM is banked into the CPU address space.
| MMR | 64K page | S | MM | CPU Memory Mapping | |||||||
| 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 | &0000-&3fff | &4000-&7fff | &8000-&bfff | &c000-&ffff |
| 1 | 1 | x | 0 | 0 | 0 | Base 64k / Bank 0 | Base 64k / Bank 1 | Base 64k / Bank 2 | Base 64k / Bank 3 | ||
| 1 | 1 | p | 0 | 0 | 1 | Base 64k / Bank 0 | Base 64k / Bank 1 | Base 64k / Bank 2 | Page p / Bank 3 | ||
| 1 | 1 | p | 0 | 1 | 0 | Page p / Bank 0 | Page p / Bank 1 | Page p / Bank 2 | Page p / Bank 3 | ||
| 1 | 1 | p | 0 | 1 | 1 | Base 64k / Bank 0 | Base 64k / Bank 3 | Base 64k / Bank 2 | Page p / Bank 3 | ||
| 1 | 1 | p | 1 | b | Base 64k / Bank 0 | Page p / Bank b | Base 64k / Bank 2 | Base 64k / Bank 3 | |||
PAL I/O port
Note that no settings are stored in the Gate Array, but the PAL and Gate Array share an I/O port address.
Bit 14 of the PAL selection address must be at 1 on CPCs equipped with CRTCs 0, 1, 2. It can be at 0 or 1 on CRTCs 3 and 4.
For compatibility reasons, it is strongly advised to always set bit 14 to 1 to select PAL.
PAL Type Detection
10 OUT &7F00,&C0:POKE &4000,&C0 20 OUT &7F00,&C7:POKE &4000,&C7:OUT &7F00,&C0 30 IF PEEK(&4000)=&C7 THEN PRINT"PAL chip absent":END 40 OUT &BC00,&F:OUT &3DFF,&C7 50 IF PEEK(&4000)=&C0 THEN PRINT"PAL I/O layout on CRTC 0/1/2":END 60 IF PEEK(&4000)=&C7 THEN PRINT"PAL I/O layout on CRTC 3/4":END 70 PRINT"Error!"
See also
- CPC 464/664 cannot deal with A14/A15 for bank 0 like the 6128 does. So external RAM expansions differ in their behaviour regarding &C3 mode. See Discussion on the forum and Another discussion