Most patents are worthless for learning anything, but Texas Instruments has some remarkably useful patents. That one, in particular, is a 61-page complete description of the calculator hardware with detailed schematics, as well as the software in flowchart and assembly form and the chip's instruction set. The patent has enough information to completely simulate the calculator (which I did: http://righto.com/ti). This patent was clearly written by the engineers and not the lawyers.
The funny thing is that what the patent actually claims is an obscure technique for connecting the ground lines in the calculator's ROM. In other words, almost all the detailed text in the patent is irrelevant to what is actually patented.
Thanks Ken, it's a great simulator. Is there any way others could write code for this simulator and play around with it as opposed to just running the original code?
And what are the best and worst features of this hardware architecture? Was there any little thing you think the designers should have done to improve it? Or was there something genius about it?
As far as the architecture of the calculator, it's kind of amazing that they were able to build a calculator chip at all, given the state of IC technology. The architecture is highly tuned to calculators, rather than being general-purpose. For instance, a "digit scan" register cycles through 10 scan lines to illuminate each digit. At the same time, it scans 10 rows of the keyboard.
Another interesting thing about the calculator chip is it has 11-bit opcodes. Powers of two are so ingrained in us that it seems bizarre to have 11-bit opcodes. But old computers often had random word lengths. The calculator's instruction set is very low-level, more like microcode than a "real" instruction set.
The funny thing is that what the patent actually claims is an obscure technique for connecting the ground lines in the calculator's ROM. In other words, almost all the detailed text in the patent is irrelevant to what is actually patented.