If nothing else, as a reminder of how amazing it is that we can build up such robust abstractions on such wobbly underpinnings. And that *someone *still has to think about this analog stuff, even if you don’t.
Software Defined Radio is a lot of fun. I wrote my own SDR program, mostly from scratch, and implemented a ton of traditionally analog stuff in software. Things like phase-locked loops, IIR filters, and frequency demodulators. The experience of implementing this stuff taught me a great deal.
Maybe, but I was really curious about how a mechanical computing system built with parts that could aren’t micro- or nanoscopic would look, if was comparable to a modern computer in function
It is interesting, and can be fun to talk about, but yes in the end a bit of whimsy
For about two years in the mid 50s, I was responsible for the operations of a rather large analog computer. It was specialized to solve systems of up to 7 simultaneous differential equations of the form dx_i/dt = f_i(x_1,…,x_7) where the functions were linear combinations of linear and quadratic terms, that is k_jx_j and l_jkx_jx_k where the k’s and l’s are constants. (Any chemist will recognize these equations as chemical reaction equations.) The machine, made up of operational amplifier packages made up of subminiature vacuum tubes, occupied 7 consoles each about 2 feet wide and 7 feet high. Programming was by running phono cables from outputs to inputs and was trivial. My main job was repairing the op amps.
Addition, subtraction, and integration were trivial. Multiplication of two variables was a kludge, based on an approximate squarer and the quarter square identity: xy = (1/4)((x+y)^2 - (x-y)^2)). It operated on a 5 Khz square wave.
You could obviously build an analog Turing machine, but to match the computational ability of your phone, it would probably be the size of a continent and take the age of the universe to do any significant computation.
I think “analog” in the context of computing means not that there are continuously variable voltages and/or currents at the most immediate level in the behavior of the circuits, but that the value of a single voltage or current scaled over its possible engineering range is an analog or proxy for some other value of interest. In “analog” circuits there is an “analogy” between the voltage here and some different value in the real world, such as temperature or pressure or altitude or position of a machine part.
The fact that even in digital computers the lowest level components operate over continuous ranges of voltage or current isn’t the same sense of “analog”. And none of those voltages, by themselves, are meant to represent continuous variables out in the real world; they only do so in groups according to a somewhat arbitrary logical code such as IEEE 754 formatted 32 bit floats.
That’s know stuff, called MEMS. 10 years or so back we had lots of sessions on how to test them. As I understand it, the chip inside your airbag system uses them for one. So production technology at the moment.
Indeed–the accelerometers and gyroscopes in every phone also use MEMS. Some main oscillators also use MEMS instead of quartz. There was an amusing incident a short while ago where a helium leak from an MRI machine killed every iPhone in the hospital. The helium infiltrated the MEMS oscillator and changed the resonant frequency enough to kill it (temporarily, until the oscillators could be “aired out”).
I’m not aware of anyone using MEMS for actual computation, but it could be done.
