petrus said:
Something in between is a wide, wide area. A calculator with only four basic
functions can be build with almost any micro that has enough I/O-pins two
read the keyboard and drive the display of your choice. It may become pretty
slow if you go into larger numbers but it will nevertheless be faster then
the Eniac. (But heavier, slower and more energy hungry then the next simple,
cheap pocket calculator.) I was told that Microchip has a application note
treating floating point calculations but I did not check out myself.
petrus bitbyter
The 12 and 16 series have no multiply operation, so the floating point
libraries supplied by microchip are immense and slow. However, the
current crop of low-end pics are almost certainly faster, and hold more
code than the original HP calculators. A PIC18, with a multiply, would
make it even easier.
On the other hand, you can buy a casio fx-115MS that does everything
from complex numbers to integration for about $15, which is undoubtedly
far less than you would spend on parts, not counting programmers and
test tools to build it.
Another project, which is probably easier to pull off, and which will
give you lots of puzzles to solve and a thing you can use at the end
would be to build a VFD clock. You can get cheap vacuum fluorescent
display tubes at surplus places, They look cool, but require you to
design and implement various power supplies to handle the
grid/filament/anode. Take the timebase off of the powerline, and use
CMOS logic or something like that to generate the clock display. A
microcontroller (PIC or something) would make it easier, but you would
have to learn to program it.
That will be a fairly big project, not too hard, and you'll have an
interesting looking clock afterwards, not a slow, feature-bare calculator.