7 Nov
2000
7 Nov
'00
3:09 p.m.
At 11:06 PM 11/7/00 -0500, you wrote:
It would be 1979 before there was DSP fast enough to do
a FIR filter
digially for audio frequencies, Analog had been doing that for 30+ years.
One could duplicate it using modern analog opamps and still solve the
same problems. The best one I remember solved a bouncing ball, for
varying gravity and rebound rates and "drew it" on an O'scope face.
It was a Popular Electronics (maybe RE) design. Still very buildable.
It's often forgotten that there were different op-amp designs that
allowed
things like four quadrant multiplcation, LOG and ANTILOG amps, CLIPPING,
SUM, DIFFERNCE, ABSOLUTE VALUE when combined allow solutions
of great speed with good accuracy.
Ever since the late 60's, very large simulations of power
grids were being performed on analog computers for the purposes of
contingency analysis and dynamic stability assessment. These
special-purpose analog computers filled entired rooms and sometimes
had thousands of dynamic states (one integrator per dynamic state).
Back in 1992, I remember that I was running a simulation of a large
power system on a Sparcstation IPX with 32MB or RAM. The model had
several thousand states, and, with all of the storage overhead, the
machine was swapping frantically for a couple of days while the simulation
ran. An older faculty member saw that, and told me that he was solving
problems of that size in one morning in the 60's. Most of the work
involved setting the initial conditions and making sure that several dozen
plotters had enough paper to print all of the results :-) . The actual
simulation ran in just a few minutes.
Carlos.
Allison