Memotech MTX ROM Replacement

System Description : MTX 500 Serial No.20143, 4000-05 computer board with 3 x TMS4764 mask ROMs.

Problem Description : Replacing failed/obsolete ROMs

In 2017, Wolfgang Joerger generously donated an FDX system to me, in addition to the disk unit, the system came with an MTX500 computer which contained a 32kB RAM board and an FDX interface board that had been modified "after market" to add the RS232 serial port components.

The computer had a couple of problems :-

• A failed VRAM
• A faulty ROM

Replacement of the faulty VRAM was quite straightforward, given the number that I have replaced already. I won't go into the details on this page, but I replaced all of the VRAMs while I was at it. See the Repairs overview page for a few examples of how to fault find an replaced failed VRAMs.

The ROM failure was more interesting and warrants a page devoted to it - hence what you're reading now . . . . .

After replacing the failed VRAM, at power-on, the "MTX500" tried to start, but immediately dropped into PANEL mode and would not respond to any keystrokes. Given that the usual symptoms of faulty DRAMs are a black screen with a constant audio tone generated, I thought that a RAM fault was unlikely, and tried swapping out the socketed components in turn, including the CPU, CTC and VDP - all to no effect.

When I turned my attention to the ROMs, I found that the system ROMs had identifiers that I had not seen before. I have a UK spec MTX500 with a 4000-05 computer board in it and the system ROMs are TMS 4764 mask ROMs identified as "MTX1A", "MTX1B" and "MTX1C". The ROMs in this machine were identified as "MTX2A", "MTX2B" and "MTX1C", as it turns out, the "2" series ROMs are German language specific ROMs (this machine had a German character set keyboard fitted).

I was able to get the machine to boot successfully using the "MTX1A" and "MTX1B" ROMs from my UK spec machine, confirming that there was indeed a ROM problem so I needed to investigate the failed ROM(s) further, which proved to be somewhat problematic.

There are no compatible EPROMs for 4764 ROMs,  the 4764 footprint does not have a programming pin allocated - the only way to program them was to use a custom mask during fabrication and direct replacement with an EPROM/EEROM is not possible. Similarly, neither of the device programmers that I have was capable of reading the TMS 4764 directly to allow me to test the suspect ROMs. However, the pin-out of the TMS 4764 is very close to an AM2764 EPROM, with only 3 pins being different, so I built an adapter to allow me to read the 4764 ROMs in my TL866A as an AM2764 EPROM.

As I did not expect to have to read these ROMs again in the future, I had planned to knock up a "rough 'n' ready" adapter using a couple of cheap IC sockets, unfortunately though, the cheap sockets were pretty useless for doing that, so I ended up spending a bit more time that I intended to making a more robust adapter on a piece of prototype board with a ZIF socket fitted.

TMS4764 vs. AM2764 pin-out differences   4764 28C64 A11 18 23 A12 21 2 CE -- RD
Adapter board
Underside of the adapter board, showing the connections between the 14 pin socket and the 20 way headers (16 pins connected) that plug into the TL866A ZIF socket.

Using the adapter, I was able to read the "MTX1C" ROM and confirm that it was identical to the "1C" fitted to my UK spec MTX500. I was also able to read the "MTX2A" ROM and found that its contents were identical to the "A" EPROM fitted in the German spec MTX512 donated by Manfred Flume in 2016. When I tried to read the "2B" ROM, although the TL866A appeared to read the device successfully, the contents were blank - each byte returned "FF". So, either the ROM had never been programmed (unlikely) or it had failed in such a way that the contents could not be read by either the programmer or the MTX.

This left me with a failed ROM and no way of getting a direct replacement, the only solution was going to be to modify the computer board to accept an alternative device and program that with the "MTX2B" code from Manfred's German MTX.

The MTX 4000-04 series computer board has a number of links that allow different ROMs to be installed, although the 4000-05 series computer board can accept the same range of ROMs, the corresponding connections are hard wired which needed some patching to be done to the computer board.

I had intended to just patch the "B" ROM socket, but on examining the computer board it became clear that it would be simpler to modify all three sockets - even if it was a bit more work. The ROM sockets are quite close together, and whilst it might have been possible (just) to cut and patch the needed tracks with the sockets in situ, I decide that it would be neater to remove the existing sockets to expose more of the tracks that needed cutting, then fit new sockets before applying patch wires as necessary.

This is the ROM circuit diagram from the MTX Operator's Manuals. There are two versions of the Operator's Manual, the original Memotech version and the "new" manual produced by Phoenix publishing (the better version). Although the diagram is the same in both, the Memotech manual notes that, for the 4000-05 computer board, "ROM A, B C hard wired as position "a" of LK 1, 2, 4, 5".
On a 4000-05 series computer board, apart from the Chip Enable signals on Pin 22, the three ROMs are wired in parallel and the circuit diagram for all 3 ROMs is effectively as shown opposite.
Replacing the TMS 4764 ROM with, for example, a 28C64 EEPROM requires that the modifications shown in red opposite are made.
To isolate pins 2 and 27 from 0V and from each other, one track must be cut at six different locations as shown. The A11 and A12 address lines must be cut at the locations shown to allow them to be moved from pins 18 and 21 respectively on the TMS 4764 to pins 23 and 2 respectively on the replacement EEPROM.
On the solder side of the board, a single cut needs to be made to disconnect the Vcc connection from pin 24 on the TMS 4764 to make what will become pin 26 of the EEPROM and should be "not connected". In fact, the three pins that were Vcc for the TMS 4764 will still be connected to each other, but not connected to anything else.
This image has been annotated to show the patch wires that will be required once the sockets have been replaced. As well as repositioning the A11 and A12 address lines, pins 27 and 28of the EEROM will be shorted together to tie the programming pin high (5v). In addition, pin 20 (CE) of the EEPROM must be patched to RD.
New turned pin sockets installed. If you look closely (on the full size photo), you can just about make out most of the tracks cut in the locations identified above.
Solder side of the PCB, showing the patch wires
The replacement AT28C64s fitted If you open the full size image, you can see my DIY labels fitted to the EEPROMs as well as the shiny new VRAMs in their sockets.
Not being 100% confident that it would work first time, I tested the board before refitting it to the keyboard base. However, I needn't have worried, the machine booted at the first attempt and appears to work flawlessly!
Although this machine currently has a 32k RAM expansion board fitted, it is the only MTX500 that I have that has not either been upgraded to 64k on board or has a RAM expansion board soldered in place. Despite having a German keyboard layout, the machine will now have a new life as my MTX500 test system for proving the functionality of my expansion boards on a bare MTX500.