>Martin McCormick <martin@DC.CIS.OKSTATE.EDU> wrote:
>>         I remember reading somewhere that a few of the early tube-type
>> computers of the 1950's used CRT's as ram devices.  The method for storing
>> and reading the display involved bouncing electrons off the phosphor as
>> described for storage tube monitors.  I believe that the devices had to be
>> refreshed just like dynamic RAM every so often to keep them from fading.
>
>It was "Williams Tube" memory.  It did need to be refreshed, and did not
>acchieve tremendous density.  It also was not terribly reliable.  It is
>rumored that the first video games were played on these; kinda neat having
>the frame buffer and display in the same device.


Without a doubt, a cool application.  But how about the use of CRTs in
analog computers as arb function generators?

"The input voltage X1 is applied between the horizontal deflection plates
of a cathode ray tube through a suitable d-c amplifier.  The voltage X0
between the vertical deflection plates is taken to be the output voltage,
this voltage is made to vary as a function  X0=f(X1) of the input voltage
X1 by a feedback arrangement which forces the electron beam to follow the
boundary of an opaque mask placed over the lower portion of the cathode-ray
screen.

The bias voltage tends to force the electron beam away from the mask.  As
soon as the spot on the cathode ray tube screen emerges from behind the
mask, however, a photocell mounted in front of the screen applies an error
voltage across the input terminals of the vertical deflection d-c
amplifier.  This error voltage is phased so that the beam is forced
downward toward the edge of the mask.  The feedback loop comprising the
cathode-ray tube, the photocell, and the vertical deflection amp will,
then, tend to keep the spot on the screen traveling just below the edge of
the mask as the ouput voltage X1 changes."

(from "Electronic Analog Computers", Korn and Korn)


The basic arrangement was good for 1% to 100kHz or so, and a refined
arrangement was expected to show better accuracy.  Less 'clunky' than
wiring up servos and pots, and somewhat more elegant than the diode based
breakpoint schemes.


newell