INMC 80 News


May–September 1981, Issue 4

Page 42 of 71

The Gemini EPROM Board

The Gemini EPROM board is a standard 8″x8″ Nasbus (compatible! –Ed) card, with silk-screen, solder resist, and with plated through holes. It holds the usual addressing logic/​buffering for the Nasbus and has 16 empty 24-pin sockets to accommodate the EPROMs and one additional socket which will accept the Nascom Basic Rom. There is also an on-board wait state generator that can be switched in or out as required. This has an advantage over the one on the Nascom 2 in that it only inserts wait states when the EPROM board is being addressed, and not in every cycle irrespective of whether RAM or EPROM is being accessed. As with the Nascom 2 the wait state occurs in both M1 and MREQ cycles. (If you check your Z80 data sheets you will notice that the critical cycle from the point of view of memory access is the M1 cycle, other Read/​Write cycles are 1/2 a clock period longer). As mentioned earlier the Board also supports the Nascom Page Mode of working.

The EPROM sockets are subdived into four banks of four sockets each. Each bank can be decoded to a 4K boundary, and each bank has a strap field associated with it that is used to configure the bank’s sockets for either 2708s or 2716s. (The 2708/2716 selection being on a bank basis means that the two types may be accommodated simultaneously on the board, but obviously only in separate banks). The 4K decoding is similar to that of the RAM cards, the wanted address being selected by a wired strap on a header plug. (None of the ‘Which bank is which?’ of the Nascom 2).

Before starting assembly of the card I plugged it into my system which continued to work happily. This is a procedure I would recommend that everyone follows – it can save time later when there are a lot more faults that could be about. This test will reveal any major faults on the board in those lines that are connected to the Nasbus. (Only once have I had a board that stopped everthing. In that case I found the fault was a tiny whisker of copper across pins 1-6 of the edge connector – the board just hadn’t been trimmed properly). The board took a little while to assemble – there seemed no end to those 24-pin sockets! Once again the board was plugged in, this time to catch any possible solder splashes. With everything Ok at this point it was time to add the ics.

All the components were there in the kit, the only error being that two ‘S’ ics were supplied instead of the ‘LS’ ones specified, but luckily they were common devices and I had some of the LS ones to hand. (Perhaps someone somewhere needs new glasses?) The only other problem I had with the assembly was that the supplied 24-pin header plug refused to go into its 24-pin socket. After a great deal of careful effort (and loud swearing) it partially conceded and now sits about 2/3rd of the way into its socket.

With the TTL ics in and the board plugged into the bus everything still worked, so it was time to add some EPROMs. I set the strap fields for 2716s,​the decode address for 8000 on Bank 0, inserted some 2716s and plugged the board back in. Lo and behold everything worked! However I subsequently discovered one small point that does not appear in the manual. If a 2716 bank is decoded as a 4K block and the 4K block is an even one (0,​2,​4,​6,​8,​A,​C,​E) then the 2716s must be placed in positions X0 and X1, if it is odd then they must go in X2 and X3. From this it follows that if a 2716 bank is decoded as two non-contiguous 4K blocks, then one must be odd and one must be even.

To sum up. The Gemini EPROM board is a well produced board that does the job it was designed to do without any apparent vices, and has been a welcome addition to my system.

Page 42 of 71