flipflops; the Linear Address Generator (LAG) made out of a PAL and a couple of counters; the Bus Mgmt Unit (BMU) PAL, 4 multiplexors for the RAM addressing, 2 EPROMs, some bus drivers, a video output shift register, and the 6522 Peripheral Interface Adapter (PIA) which had 16 programmable I/O lines, which handled the keyboard and mouse interfaces and some other tasks. That was the core architecture, to which was added the dual serial ports, the internal and external floppy drive ports, the real-time clock chip, and the sound output. The PALs were the Programmable Array Logic parts made by MMI, which allowed one to write output logic equations to define each of the 8 outputs. They were fairly power-hungry (by today’s standards at least) but were cheap and flexible and Burrell put them to good use.
The underside of the wire-wrap protos was of course a maze of wires. One morning we were amazed to find that one of the protos was badly scorched... it turned out that late the night before, Burrell had been making some changes and managed to get power and ground shorted together somehow, and in his impatience to get the proto back up, he decided to use ‘brute force’ by removing all the chips and connecting 120VAC across the power and ground in an attempt to ‘burn out’ the short.... a misguided attempt, as one would expect! I don’t that proto was ever revived.
Later in 1981 we brought Colette Askeland over from the Apple-II/III division to lay out a Mac PCB – the first of eight iterations of the board between 1981 and the final version in late 1983. Looking at these boards now, they seem like quaint antiques indeed. Not one surface mount component! Here’s a list of the 8 board iterations:
#1 First PCB 1981 MC1001-00 no silkscreen... big holes around the perimeter for mounting the ‘fence’, the solid copper bus bar that was to run around the perimeter of the board and be tied to cold or chassis ground in an attempt to minimize the RF energy radiation from the digital logic. We eventually were able to cut the ‘fence’ back to just run along the back edge of the board.
#2 First Build, 1981 MC1001-01
--33 ICs (not counting CPU, 128k RAM). 3 Eproms, dual 6551 UARTs (serial ports), 6522 PIA
3 PALs: BMU (Bus Mgmt Unit), LAG (Linear Address Gen.), TSM (Timing State Machine)
#3 MC1001-02 Feb1982 build of 50
--32 ICs (not counting CPU, RAM). Z8530 SCC, word-wide RAM (instead of byte-wide), and only 2 EPROM sockets now. At the specific request of our esthetic guru Steve Jobs, the RAM array was spread out to match the regular grid spacing of the other chips, which didn’t work out so well: the increased trace lengths caused a troublesome increase in noise on the high-speed RAM signals, so the next board went back to the more densely packed array. This board also included the new STFCLK (‘Special Task Force’ clock chip) which was a custom chip, dedicated to keeping track of real time and able to store some system preference values. The mouse connector was moved to the back of the board. The Z8530 was a very new part from Zilog which incorporated dual serial ports with a new high-speed synchronous serial mode AND a hardware header-recognition compare register, so that incoming packets could be qualified by the hardware and the processor would only be interrupted to service packets addressed to that Mac -- this was to be key to implementing Appletalk (after a few more years of software development). The Z8530 was a cool new high-performance chip which, like the 68000, seemed over-powered for a small consumer-oriented machine – but Steve Jobs once again successfully negotiated a remarkably low price for the part based on our projections of buying millions of them. This board also marked a change from the old Apple-II 6-pin power connector to a 10 pin power/video connector, no doubt because we finally had George Crow on the team and his ‘analog board’ integrated the switching power supply and the video CRT