3 ms·

Is there a FPGA equivalent for these analog chips? I'm guessing no but that would be handy to have field programmable analog computers.

nickpsecurity 9y ago
Definitely: https://en.wikipedia.org/wiki/Field-programmable_analog_array https://en.wikipedia.org/wiki/Field-programmable_analog_arra... There was also a general-purpose, analog computer that people seem to have done similar things with. http://www.comdyna.com/gp6intro.htm http://www.comdyna.com/gp6intro.htm https://www.scss.tcd.ie/SCSSTreasuresCatalog/hardware/TCD-SCSS-T.20121208.027/Comdyna-GP6-Operators-Manual.pdf https://www.scss.tcd.ie/SCSSTreasuresCatalog/hardware/TCD-SC...
ChuckMcM 9y ago
I haven't seen any but what you would really need would be a challenge to put into a chip. Essentially you would want a wiring matrix of that you could use to connect groups of integrator / differentiator units to, and then switchable resistors and capacitors to set rates. Not as straight forward as one might like. An alternative might be a matrix of DSP 'cores', instead of switching an analog signal around you would move around an I/Q stream. That might be an interesting path to wander down for a bit.
- junkcollector 9y ago
  A voltage controlled varactor can be made from 2 non-complimentary mosfets with a common body bias. The differential amplifier can be done with as little as 3 additional fets, but probably more on the order or 7-10 for performance reasons. So you're looking at programmable on chip integrators for around 10 mosfets a piece. Of course, you'd use a technology size larger than what is used for digital logic because the analog toolchain would be more sensitive to problems like channel noise, gate tunneling, etc. You would not want to use a differentiator because they tend to amplify noise in an unstable and uncontrollable fashion and are not required for functional completeness (the analog equivalent of Turing completeness). Edit: IQ analog processing is extremely common in the high performance RF field but tends to be static and not reprogrammable. FPAA's exist on market but are extremely expensive for what you get. The general purpose analog computer was initially developed by Shannon (yes that Shannon) back in the 40s.

nickpsecurity 9y ago

Definitely: https://en.wikipedia.org/wiki/Field-programmable_analog_array https://en.wikipedia.org/wiki/Field-programmable_analog_arra... There was also a general-purpose, analog computer that people seem to have done similar things with. http://www.comdyna.com/gp6intro.htm http://www.comdyna.com/gp6intro.htm https://www.scss.tcd.ie/SCSSTreasuresCatalog/hardware/TCD-SCSS-T.20121208.027/Comdyna-GP6-Operators-Manual.pdf https://www.scss.tcd.ie/SCSSTreasuresCatalog/hardware/TCD-SC...

ChuckMcM 9y ago

I haven't seen any but what you would really need would be a challenge to put into a chip. Essentially you would want a wiring matrix of that you could use to connect groups of integrator / differentiator units to, and then switchable resistors and capacitors to set rates. Not as straight forward as one might like. An alternative might be a matrix of DSP 'cores', instead of switching an analog signal around you would move around an I/Q stream. That might be an interesting path to wander down for a bit.

junkcollector 9y ago
A voltage controlled varactor can be made from 2 non-complimentary mosfets with a common body bias. The differential amplifier can be done with as little as 3 additional fets, but probably more on the order or 7-10 for performance reasons. So you're looking at programmable on chip integrators for around 10 mosfets a piece. Of course, you'd use a technology size larger than what is used for digital logic because the analog toolchain would be more sensitive to problems like channel noise, gate tunneling, etc. You would not want to use a differentiator because they tend to amplify noise in an unstable and uncontrollable fashion and are not required for functional completeness (the analog equivalent of Turing completeness). Edit: IQ analog processing is extremely common in the high performance RF field but tends to be static and not reprogrammable. FPAA's exist on market but are extremely expensive for what you get. The general purpose analog computer was initially developed by Shannon (yes that Shannon) back in the 40s.