Memotech MTX - Keyboard Fault

System Description : MTX512 Serial No.(15503), 4000-05 computer board with 64k DRAM chips.

Problem Description : System dead, black screen with constant tone at power on

Resolution Summary : Reseated ROMs, replaced failed 74LS244 

Darren sent me two faulty MTX computers, his original "MTX500" (which has a MTX512 keyboard) and an MTX512 that he had obtained later, and made the attractive offer that, if I fixed his original machine, then I could keep the other one for the collection.

Darren reported that the "512 boots to a ready prompt, but the keyboard doesn't work, the 500 simply shows a black screen."   When they got to me, neither of the machines booted and both displayed a black screen on power on. A description of the repairs to the MTX500 can be found on this page.

When powered on, the MTX512 had the Memotech "black screen of death", i.e., a black screen with a constant tone from the sound chip. These fault symptoms are typically seen with faulty DRAMs, but can be caused by pretty much anything. The symptoms mean that the CPU is not executing much, if any, of the ROM code. The sound chip is usually reset/silenced early in the boot process, if things aren't getting that far, there is a pretty basic fault with the system, including faulty DRAMs, CPU or ROMs.

I was going to replace all of the socketed chips in turn, so I started by replacing the CPU (to no effect) and moved onto the ROMs. When I looked at the ROMs, although they appeared to be fully seated in the sockets, when they were removed, some quite extensive tarnishing was seen on the socket contact. Since a spare set of 4000-05 ROMs wasn't immediately to hand, I removed and replaced the existing ROMs, making sure that they were fully seated in the sockets, and tried to power on the system again.

This time, with no keyboard connected, the system booted to "Ready" prompt and appeared to be working. - Problem 1 apparently fixed with very little effort ! (Longer term, I will probably replace at least one of the ROM sockets.)

With the keyboard plugged in, I began to see the keyboard issue reported by Darren. The keyboard appeared to be working to some extent, some keys worked, but others didn't and it appeared that some others key(s) were stuck, giving random "gibberish" on the screen.

My keyboard fault finding page describes the MTX keyboard scanning process and keyboard drive/sense line interface. If there had been any obviously not working or stuck keys, then analysis of the drive/sense circuits can help identify the faulty IC(s). In this case, there were no obviously suspect keys, so a different approach was required.
Most of the keys are read through Input Port 5 using the 74LS244 Octal Buffer/Line Driver (board location 2A).
The remainder of the keys are read through Input Port 6 using 2 bits of the 74LS244 Octal Buffer/Line Driver in board location 1D). A further 2 bits are used for the keyboard language selection switches and the remaining 4 are used for the printer port (Input Port 4, not shown).
Putting a DVM on the keyboard sense lines showed that, as expected, most of the lines were reading ~5V, i.e., being pulled up to +5v by R1 to 8 (Port 5, location 2A, KB0 to KB7) and R22 (Port 6, location 1D, KB9). For a UK system, the language selection bits are usually set low with SWA, but opening the switches proved that the bits were successfully pulled up to ~5v. However, with no keyboard connected, the KB8 line was reading around 2v, suggesting that it was being dragged down by the 74LS244 so this seemed like a good candidate for replacement. I removed the existing chip, added a new socket and fitted a replacement 74LS244.
This fixed the problem and the keyboard is now working as expected.
That was pretty much all that was needed on this machine. The keyboard isn't brilliant, but all but one of the keys were pretty much acceptable. The <ENT> / <CLS> key on the numeric keypad was very rough, so I did change that one out.