IMSAI 8080 kit, about $500, ca. 1976. I soldered every part myself, then, knowing very little about computers, couldn’t figure out why it didn’t do anything. A trip to the Byte Shop told me why.
The kit came with a CPU, a power supply and a case, but no RAM. :smack: (What’s RAM?)
Pushing my wallet to the limit, I invested in a 4K RAM board for about $300. I soldered every part myself.
With no keyboard, I had to toggle the front panel switches to enter a program, one step at at time.
With no display device, I read the front panel LEDs to get the data results.
I added a local Junior High math teacher to my team; he knew more than I did about computer stuff and had been teaching the kids to enter data thru punched paper tape.
Long-term storage? There was none. The more advanced hackers told us to buy a Tarbell cassette interface and a Penny’s audio cassette recorder. Now we could load Basic after writing our own drivers in object code.
A keyboard came next. Diving into surplus bins, I dug up two keyboards that had been ripped from teletypes or something like it. Had to figure out the wiring and pinouts, but we got one working eventually. We wrote our own drivers in object code.
But a Soroc terminal was being sold at a serious discount at the Byte Shop. It was a 9" monochrome screen with a keyboard rigidly attached and RS232 I/O up to 19,200 baud, block mode option. Text only, no graphics. List price: $1600, discounted to $900, so I bought one. Now I had both a display and a keyboard in one package. We wrote our own drivers in object code.
Pretty soon the cassette interface was shown to be inadequate for data storage, so I bit the bullet and purchased the very latest storage technolgy, a dual Micropolis floppy drive unit with S-100 interface card for $1895. Each single-sided, hard-sectored 5" floppy stored 360K. We wrote our own drivers in object code.
Hard drives were only a dream.
A printer was a luxury, but we were working on a graphics project and it was obvious we needed one. An experiment with a plotter was a failure (We wrote our own drivers in object code but couldn’t get the plotter to respond properly). We found a startup company, Malibu, that sold a dot matrix printer, 60DPI, 11" wide, using fanfold paper, for $2500. We convinced them to sell it to us for the wholesale price of $1500. It sounded like a cat screeching when it printed a line, and we had to write our own drivers in object code.
We built up the IMSAI with 32KB of RAM in a mix of 1K, 4K, 8K and 16K boards, mostly hand-soldered. About 32KB of ROM and other memory-addressed devices maxed out the addressing space and we could add no more. A Merlin graphics board got us to a 320 x 200 graphics display on a 19" monochrome studio monitor. To get resolution that high, the Merlin was a triple-width board and used interrupts that effectively reduced the CPU speed from 2Mhz to 1Mhz.
Experiments with Basic showed that it would be too slow, as it took 8 minutes just to clear the screen by writing to all memory locations. It would have to be object code all the way (we didn’t know what an assembler was at first).
With no operating system available, we wrote our own and burned it into a PROM with our homemade PROM burner. An ultraviolet light inside a small box made an eraser.
And with that, we produced music graphics at 60x75DPI that we put in front of orchestras and they played it without a problem, knowing that for the first time, there were no missing or extra beats in any measure, something that hand music copying could not guarantee.