Micro­power

upper chips directly on top of the lower chips, but otherwise you would be well advised to solder DIL sockets for the upper layer onto the chips of the lower layer, as only one chip is put at hazard, and you will retain the option to remove the top chip easily in future. Before soldering on the top chip/​socket, wrap it in a couple of layers of foil as well, with the pins protruding through the foil. Again, the foil acts as an insulator and heat sink.

One unfortunate by product of this ‘recommended’ method is that when you have finished you have to dig bits of cooking foil from between the pins of your newly mated chips, and you may prefer to make yourself a Jig with a spare DIL socket which has all its pins commoned in some scrap veroboard. One last thought on piggybacking ICs; it is a good idea to make sure that pin 1 is clearly recognisable from below on the bottom IC, otherwise just as you finish a beautifully neat soldering Job, you realise that you can’t remember having checked the orientation of the bottom chip. If the end of the 4116 is not clearly notched, lightly scratch the underside of the IC next to pin 1, or mark it with a small dot of marker dye; and do this before you start any soldering.

Now for a quick look at the theory behind the modification. Fig. 12 of the instructions which accompany the RAM-B board show the memory ICs arranged in three banks of eight, with addresses commoned from the right of the diagram, and data commoned vertically. On the left there are essentially three lines to each bank; a column address strobe (CAS) which is common to all three banks, a unique row address strobe (RASx) to each bank, and a write strobe (WRx), again unique to each bank. The logic behind the latter two runs something like this. Each bank MUST be individually addressed and if you want to create another bank then you MUST create an additional RASx signal from somewhere. Since the existing banks are numbered 0 – 2, with their associated RAS0 – RAS2 signals, it makes sense to number the new bank 3, and call the new signal RAS3. The third signal is the WRx signal, which must be taken low when whenever you (the CPU really!) want to write to the respective bank. If the WRx signal is held high, bank x will become write protected, and if the relevant pins of the upper bank are soldered to those of the lower bank, the upper bank will take the same write protect status as the lower. At this stage, let me state that it is possible to arrange for the new bank of RAMs to have independent write protect , but to be frank, although I have run my board in this condition, I have never found the need for separate write protect on the new bank of RAMs, and in the end I Just removed the additional wiring as it was a bit of an eyesore. For this reason, I shall not cover this aspect of the modification, but if any reader is interested in the details, please contact me through this magazine.

From what I have said, it should be obvious that all 16 pins of the upper and lower 4116s can be commoned (is there really such a word in English?) except for pins 4, which must be separated between banks but comoned along the new bank.