Thinking of keyboard interfaces (HP9915)

With this discussion of interfacing PC keybaords to vintage machines, I've been thinking about a project I've had in mind for some time now.. Firstly soem backgroud. The HP9915 is essentially an HP85 in a half-rack-width metal case. It could be run on a bench, or there was an HP mounting kit top put it in an instruemtn rack. Electornically, it's similar to the HP85. There is a buit-in tape drive. Ther is no printer. There is no built-in monitor, but there's BNC socket o nthe back that outputs TV-rate composite video (so getting a display is no problem). The keyboard controled IC is present, but obviously there's no built-in keyboard. There are some user-definable buttons on the front which are wired as the programamble function keys on an HP85. The keybaord row and commn lines are also buffered to TTL levels and broufht out on a DB25 socket on the rear of the machine. There was an HP keyboard that conencted here, this is basically just a matrix of switches wired to a DB25 plug. It's the same matrix as the HP85 keyboard, and that one is docuemtned in the service manual (although with at least one typo!). However, the keyboards are much harder to find than the machines. It's a 10*8 matrix with separate lines for the shift, control and capslock keys (these lines are simply grounded by the the appropraite key) My first question in designing this is : Is there ever a time when you press 2 keys simultaneously on an HP85 (other than shift or control along with another key). In onter words, is there ever a time when 2 locations of the keyboard matrix are closed at the same time? The reason I ask is that I can't think of one, and if such a case doesn't have to be considered there's a very simnple interface (a mutlipezer to sleect a scan line drivign the enable input of a demultiplexer to select an input line) that could be used. It'd be 2 or 3 cheap TTL ICs. I'd not want to connect a microcontroller directly to an external connector in any case (risk of damaging it, I notice that HP had much the same idea about the keyboard controller IC in the 9915, all the signals brought out on the connector are buffered). Thing is, it's a lot easier to replace a non-programmed TTL IC than having to program a repalcement microcontroller. So I'd either want to add buffers even if I was goign to try to use the microcontroller direcrly (which for reasons we've already discussed I think it 'pushign it'). The mux/demux ICs wouldn't need any more buffering IMHO, they could be simply wired to the connector. Of course if I do need to be able to have multiple 'keys pressed' at the same time, then I would haev to use the crosspoint switch or dual port RAM solutions we've already discussed. -tony