The Selectron was an early form of digital computer memory developed by Jan A. Rajchman and his group at the Radio Corporation of America (RCA) under the direction of Vladimir K. Zworykin. It was a vacuum tube that stored digital data as electrostatic charges using technology similar to the Williams tube storage device. The team was never able to produce a commercially viable design before magnetic-core memory became almost universal.

Development

Development of Selectron started in 1946 at the behest of John von Neumann of the Institute for Advanced Study, who was in the midst of designing the IAS machine and was looking for a new form of high-speed memory.

RCA's original design concept had a capacity of 4096 bits, with a planned production of 200 by the end of 1946. They found the device to be much more difficult to build than expected, and they were still not available by the middle of 1948. As development dragged on, the IAS machine was forced to switch to Williams tubes for storage, and the primary customer for Selectron disappeared. RCA lost interest in the design and assigned its engineers to improve televisions

Around this time IBM expressed an interest in the Selectron as well, but this did not lead to additional production. As a result, RCA assigned their engineers to color television development, and put the Selectron in the hands of "the mothers-in-law of two deserving employees (the Chairman of the Board and the President)." was a by vacuum tube configured as 1024 by 4 bits. It had an indirectly heated cathode running up the middle, surrounded by two separate sets of wires — one radial, one axial — forming a cylindrical grid array, and finally a dielectric storage material coating on the inside of four segments of an enclosing metal cylinder, called the signal plates. The bits were stored as discrete regions of charge on the smooth surfaces of the signal plates.

The two sets of orthogonal grid wires were normally "biased" slightly positive, so that the electrons from the cathode were accelerated through the grid to reach the dielectric. The continuous flow of electrons allowed the stored charge to be continuously regenerated by the secondary emission of electrons. To select a bit to be read from or written to, all but two adjacent wires on each of the two grids were biased negative, allowing current to flow to the dielectric at one location only.

alt=glass vacuum tube|thumb|An operating SB-256 tube with all bits set to “1”.  The tube is in the quiescent, "maintenance" state, with read plate active.  Also shown, a 12AU7 tube for size comparison.

In this respect, the Selectron works in the opposite sense of the Williams tube. In the Williams tube, the beam is continually scanning in a read/write cycle which is also used to regenerate data. In contrast, the Selectron is almost always regenerating the entire tube, only breaking this periodically to do actual reads and writes. This not only made operation faster due to the lack of required pauses but also meant the data was much more reliable as it was constantly refreshed.

thumb|left|300px|Selectron cross section

Writing was accomplished by selecting a bit, as above, and then sending a pulse of potential, either positive or negative, to the signal plate. With a bit selected, electrons would be pulled onto (with a positive potential) or pushed from (negative potential) the dielectric. When the bias on the grid was dropped, the electrons were trapped on the dielectric as a spot of static electricity.

To read from the device, a bit location was selected and a pulse sent from the cathode. If the dielectric for that bit contained a charge, the electrons would be pushed off the dielectric and read as a brief pulse of current in the signal plate. No such pulse meant that the dielectric must not have held a charge.

The smaller capacity 256-bit (128 by 2 bits) "production" device was in a similar vacuum-tube envelope. It was built with two storage arrays of discrete "eyelets" on a rectangular plate, separated by a row of eight cathodes. The pin count was reduced from 44 for the 4096-bit device down to 31 pins and two coaxial signal output connectors. This version included visible green phosphors in each eyelet so that the bit status could also be read by eye.

Patents

  • Cylindrical 4096-bit Selectron
  • Planar 256-bit Selectron

References

Citations

Bibliography

  • Republished in IEEE Annals of the History of Computing, Volume 20 Number 4 (October 1988), pp. 11–28
  • The Selectron
  • Early Devices display: Memories — has a picture of a 256-bit Selectron about halfway down the page
  • More pictures
  • History of the RCA Selectron