No problem, feel free to do so. Please let me know once you're done, so I can add a link to your translation. If you want you can use HTML version from here, it's probably easier to copy and paste:

tht.net/~aoboroc/2004/05/panasonic-msx2-computer-256512kbytes.html

Soon I'll be modding second WSX, and this time I'll use good digital camera to make clear shots. Hope this will help future modders

The text and photo's and the ideas are by dhau. I made the pdf and published it at http://www.hansotten.com in the MSX do-it-yourself section.
Eventually it will end up at funet, the url given by dhau is a temporary location.

If you want the source of the pdf (Word format) to translate, just ask.

Hans

Idea is from Manuel Pazos, I just put it all together. I think he got information from a cool Japanese web site with information of different cool MSX upgrades. May be I even posted URL in this thread, don't remember now

my machine is also on http://www.geocities.com/philipspionier/dexx.html
check it OUT!! ;-)

Yep, of course. I got the information from Naga (and helped a bit by Tsujikawa). I'm not so good at hardware

Heya! I upgraded second machine! I have second MSX2+ with 512K

And I nearly killed it at the same step as the first time: desolder and lift pin 51, solder it to pin 52

Quote:

So now you get +5V in pin52 and not GND

Hi Guillian!

Thanks again for your help with 512K upgrade. Now I'm trying to understand the meaning of upgrade steps in order to be able to repeat it on other MSX2+ computers (Sony, Sanyo), which aren't as upgrade-ready as Panasonic computers.

I've read a bit of information on dram working, and now I understand what's D0-D3, A0-A8, /RAS, /CAS, /WE, /OE... So now I'm curious to understand completely what did we do using your help. I also understand that 464 and 4256 are similar dram chips, the only extra signal on 4256 is A8, which adds two bits to multiplexed address bus, and so chip allows to map 2 ^ (2*9) = 262144 words, 4 bits each. Two chips at the time provide a full byte.

The following is a speculation, and I need all the qualified input I can get (before I crack open one of my Sony MSX2+ computers or my Sanyo Wavy).

To get 256K on a T9769-based computer, I have to replace two 464 chips with two 4256 chips by using extension wires from 464 sockets and adding extra wire for A8 pin on 4256 to pin 47 "RA8" of T9769 (I speculate here, I'll trace where A8 goes, thanks to Yobi for T9769 pins). Once this is done, it's important to switch T9769 pin 52 "MPSL1" from GND to VCC. I assume that if this is not done, computer will still use only 64K, since A8 will be ignored.

To get 512K (maximum supported by T9769 without extra glue logic for multiplexing/demultiplexing), we can make use of pins 50 "CAS1" and 51 "MPSL0". We start with soldering two more 4256 chips as the "second floor" on top of previously added two with the exception of pin 17 "/CAS ". We should join pin 17 signals from both "second floor" dram chips to T9769 pin 50. But this won't give us 512K yet, since A8 is ignored for second dram bank. To correct this situation, we desolder T9769 pin 51 from the PCB (which is GND), and join it with pin 52, which is already VCC. This will enable full 18-bit addess decoding for second dram bank, and so we get full 512K.

Please let me know if you think I'm wrong ASAP, or else I will sacrifice one of my MSX2+ for the greater good of humanity

Well, first of all: I'm not too good at hardware. I'm sure other people here can give you better support than me. So I'm not sure whether it will work or not. Make it at your own risk (sorry!)

Quote:

To get 256K on a T9769-based computer, I have to replace two 464 chips with two 4256 chips by using extension wires from 464 sockets and adding extra wire for A8 pin on 4256 to pin 47 "RA8" of T9769

Bare in mind that x1464 pin layout is different from x4256. So you can't put a x4256 RAM chip into a x1464 socket. If there is not a suitable socket, you will need to wire the correct signals from the chip to the socket. Also add an extra wire (as you said) for A8 signal from the memory chips to the T9769.

You should trace the signals from the memory chips to the T9769 in the WSX/WX to be sure all is done right.

Also, check which is the actual computer memory. In this way you can know the T9769 configuration:
+ ------- + ------- + ------- + ------- + ------- + -------------------- +
| | 64KB | 128KB | 256KB | 512KB | Notas |
+ ------- + ------- + ------- + ------- + ------- + -------------------- +
| MPSEL1 | GND | GND | +5V | +5V | DRAM |
| | | | | | |
| MPSEL0 | GND | +5V | GND | +5V | DRAMs |
+ ------- + ------- + ------- + ------- + ------- + -------------------- +
| DRAM |81464*2 |81464*4 |64256*2 |64256*4 | <- DRAM to be used |
+ ------- + ------- + ------- + ------- + ------- + -------------------- +


The rest should be as you said, similar as you made it in the Panasonic WSX/WX

Thanks for a quick answer!

I am aware of pinout differences between 464 and 4256 ICs:

         4464                   44256 
   64-kword * 4-bit       256-kword * 4-bit
	
     +---+  +---+            +---+  +---+ 
 /OE |1  +--+ 18| GND     D0 |1  +--+ 20| GND
  D0 |2       17| D3      D1 |2       19| D3
  D1 |3       16| /CAS   /WE |3       18| D2
 /WE |4       15| D2    /RAS |4       17| /CAS
/RAS |5  4464 14| A0      NC |5 44256 16| /OE
  A6 |6       13| A1      A0 |6       15| A8
  A5 |7       12| A2      A1 |7       14| A7
  A4 |8       11| A3      A2 |8       13| A6
 VCC |9       10| A7      A3 |9       12| A5
     +----------+        VCC |10      11| A4
                             +----------+

Where:

A0-A9 - address lines (show which word to read/write to memory)
D0-D3 - data lines (four bits of word to read/write)
GND,VCC - power
NC - not connected
/OE - output enable (read from memory to data bus)
/WE - write enable (write to memory from data bus)
/RAS - Row Access Select (enable row selection latch)
/CAS - Column Access Select (enable column selection latch)

I am reposting your table as code to get it proportional:

+------+-----+-----+------+-------+--------+
|      |64KB |128KB|256KB |512KB  |Notes   |
+------+-----+-----+------+-------+--------+
|MPSEL1|GND  |GND  |+5V   |+5V    |DRAM    |
|      |     |     |      |       |        |
|MPSEL0|GND  |+5V  |GND   |+5V    |DRAM    |
+------+-----+-----+------+-------+--------+
|DRAM  |464*2|464*4|4256*2|64256*4|ICs used|
+------+-----+-----+------+-------+--------+

So according to your table, MPSEL0 signal enables second bank (i.e. use of CAS1 signal from T9769), and MPSEL1 enables A9 decoding (i.e. see more then 64K per bank). In the end we still need to connect both to VCC, but it's good to know what are the signal roles

I think I'll build my memory upgrade board on a prototyping PCB and use soft wire from 40-wire IDE cable to hook it up to computer mainboard with peices as short as possible.

Anyone outthere who wants to try and upgrade their Sony/Sanyo MSX2+ computers to 512K??? Speak up or keep your silence forever!

I got no feedback, but who cares? Here is the picture of the glorious T9769B chip (careful - 500K picture) from Sanyo Wavy 35J

My Wavy 35J was 512K memory now

Nice to hear that

I'm glad too Now I want to replace 44256 with a 1MB 3-chip SIMM (for 512K of it's memory, I'll connect extra address lines to ground)

Sony MSX2+ owners - you're out of luck