VRAM board for Yamaha CX5MII

Guardian (3597)

28-06-2015, 19:10

Here's the first design of the VRAM board for Yamaha CX5MII computers. The board includes 2 4464 RAM chips and should contain 64kb of VRAM.

Schematics:

Board layout in Gerber view:

What puzzles me is that in the computer's schematics the connections may be shown wrongly. The AD7 connector pin is connected to the chip's A0 pin and AD0 connector pin is connected to the chip's A7 pin. This doesn't sound logical to me. Any ideas whether this is right?

If someone has this board, would be nice to check the above. Also would be nice to check whether the location of AD1, AD2 and AD0 on the connector is correct.

Anyway, I can share the schematics and gerber files, so whoever is interested could make his own boards. You can do this by ordering at www.pcbway.com site and uploading the Gerber files that I will provide on request.

IMPORTANT! It would be really nice if some hardware gurus here could check the schematics for errors. I am not professional engineer, so there may be mistakes. Though I made everything based on the original board's connections (except for AD7 and AD0 - on my board they are connected to A7 and A0 pins respectively).

Aangemeld of registreer om reacties te plaatsen

Van Grauw

Ascended (10768)

29-06-2015, 00:18

Original board pics: 1 2 3 4 5

I didn’t take pictures of the bottom of the boards unfortunately… But as for the VRAM board, on the darkish picture where it’s connected, by using a contrast filter maybe you can see it more clearly. If needed I can open it up to take more later.

On the last pic, is that airwire connecting CAS0 to the header connector? The schematic says CAS1 but that seems a bit strange for 16K VRAM… I can’t tell from the top if CAS1 is connected.

AD0 and AD7, maybe it is just for convenient wiring. Would it matter?

Van sd_snatcher

Prophet (3659)

29-06-2015, 03:56

I wonder: Why 64KB of VRAM instead of 128KB?

Van o.geerdink

Hero (588)

29-06-2015, 07:21

The machine has 16kb vram so with 64kb vram added it still won't be enough for normal msx2 functionality?

Van Wierzbowsky

Guardian (3597)

29-06-2015, 09:33

o.geerdink wrote:

The machine has 16kb vram so with 64kb vram added it still won't be enough for normal msx2 functionality?

64kb on-board + 64kb with this external board makes it 128kb. I am not sure it's possible to make an external board with 128kb without logic chips. I am not even sure whether this board will work as is, without the additional logic. The original board was designed for 4164 chips, so just replacing them with 4464 may not work. I need an expert's piece of advice...

Van Jipe

Paragon (1614)

29-06-2015, 14:39

on DRAM you can mix adress signal or mix data signal
is not conventional but work
on Elektor 64K Sram signal are mix for make an easy board

Van maxis

Champion (512)

29-06-2015, 16:35

Alexey wrote:

I am not sure it's possible to make an external board with 128kb without logic chips. I am not even sure whether this board will work as is, without the additional logic.

Alexey,

The pins on the VDP are connected this way to make sure that all the banks/pages are correctly refreshed by the V9938. Since upper screen modes are not implemented with 4416 and the memory refresh might require only 2ms per ROW, the addresses A0 & A7 are swapped.

Actually 16Kx1 and 16Kx4 devices have the addresses A0-A7 connected to the ROW address buffers and A1-A6 to the COLUMN address buffer. I.e. the pins A0 & A7 are only used in the ROW selection and can be swapped.

The register R#8 enables/disables the advanced timing (page mode) and address multiplexing scheme while accessing the external DRAM chips.

Regarding the 128K on the daughter board, only CAS0 is routed to the daughter board connector on YIS503/II. Also only 64Kbytes of VRAM can be addressed per CASx bank. Therefore maximum 64 KBytes of VRAM can be addressed by the plug-in board which makes it compatible with the MSX2 (you can always bring CAS1 signal separately for 128K).

When working with 128K of VRAM, putting one big DRAM chip with an additional logic wouldn't help, since V9938 uses the bank interleaved access in screen 7 & 8.
I.e. only the static memory has enough bandwidth to replace the two bank interleaved access.

Van Wierzbowsky

Guardian (3597)

29-06-2015, 17:31

Thanks, Maxis. So based on your explanation I can put 2 more 4464 chips on the external board, connect them in a similar way as the 2 first chips, but their CAS inputs will have to be wired directly to the VDP's CAS1 output, right? This should make it a 128kb expansion board.

Btw, I don't get why on the schematics there's CAS1-V on the connector while you say that there's CAS0 there. I don't have the computer, so I can't check which CAS is routed to the external board.

Van maxis

Champion (512)

29-06-2015, 18:09

Alexey wrote:

Right, you can put all the 4 chips (4464) on the same daughter board. In terms of parasitic capacitance applied to the address and control lines, this solution (4x4464) is better than 2x4464+8x4164.

Alexey wrote:

Btw, I don't get why on the schematics there's CAS1-V on the connector while you say that there's CAS0 there. I don't have the computer, so I can't check which CAS is routed to the external board.

I don't remember how the YIS503/II schematics is, however, V9938 has CAS0 always connected to some VRAM while CAS1 sometimes hanging in the thin air.

P.S. Вроде бы в оригинале на плате у YIS503/II установлена только "дочка" с двумя 81416, а на материнке микросхемы не впаяны. Поэтому, предположил, что дочка "сидит" на CAS0,а основной банк 4164 на CAS1.
Можно использовать 120нс память от Фуджитсу, Тексаса, НЕК или Самсунга. 150нс лучше не ставить совсем.
Удачи!

Van Wierzbowsky

Guardian (3597)

29-06-2015, 18:22

OK, I will make the board with 4 chips and a wire to CAS1 then. Would be a nice solution for those who don't want to install 4164 chips on board.

Van Grauw

Ascended (10768)

29-06-2015, 19:58

Alexey: sounds great! Do you still want pictures of the bottom side?