Computer Hardware Memory with Neon Lamps

Like other gas discharge lamps, the neon bulb has negative resistance; its voltage falls with increasing current after the bulb reaches its breakdown voltage. Therefore, the bulb has hysteresis; its turn-off (extinction) voltage is lower than its turn-on (breakdown) voltage. This allows it to be used as an active switching element. The circuit below demonstrates a basic neon lamp memory. The supply voltage V is between the breakdown and the maintaining voltages of the lamp so that in the quiescent condition the lamp is non-conducting.

If a positive "set" pulse is now applied to the input the lamp will fire and with the supply voltage greater than the maintaining voltage, the lamp will reming conducting. The amplitude of the set pulse must be large enough to fire the lamp and thus must be greater than the difference between the firing voltage of the lamp and the supply voltage. So

V_set > V_F - V

where

V_M < V < V_F

To read out of the memory circuit a positive "read" pulse is applied to the input. The aplitude of the "read" pulse must be less than that required to fire the lamp if the lamp is non-conducting. So:

V_read < V_F - V

If the lamp has been previously fired by a "set" pulse the "read" pulse will be passed through the lamp to the output, however, if the lamp has not been fired, the "read" pulse will be blocked by the non-conducting lamp and the lamp and the output will remain at zero.

Readout can also be accomplished with the use of a light detector such as a photocell or photodiode, avoiding the need to use a read pulse.

The reset the memory circuit a negative "reset" pulse is applied to the input. The magnitude of this pulse must be sufficient to drop the voltage across the lamp below the extinguishing voltage.

| V_reset | > V - V_E

Examples:

Evan Daniel made a "game of life" from Neon Lamps, demonstrating their use as memory elements:
https://www.youtube.com/watch?v=wlDrr4VRlQ4

There are no microprocessors or integrated circuits. The neon lamps act as both the indicator and the memory, so you are literally seeing the state of the cells. In0 is the output of the cell; In1-In8 are its neighbors. It needs two power supply voltages (+70 and +18). Clock is a square wave off the 18V, run it slow. 70V is low precision. 18V and is higher precision. Most of the resistors are 1%, hopefully which ones matter is clear. Transistor type matters; others would work instead, I suspect, but you might have to tune the resistor values. 2N5089 was chosen instead of 2N3904 for higher gain, which is needed. The edit wand is just the clock signal, AC coupled via 0.1uF cap, touched to the lower leg of the lamp. If things don't run just right, start by tweaking your 18V supply voltage.

Also: Lego Logic Blocks / Optical NAND

See also:

• https://en.wikipedia.org/wiki/Dekatron If a series of 10 gas discharge elements are configured in a ring inside a tube, pulses can cause a lit element to fire the next element and reset the inital element. This makes a ring counter that increments from 0 to 9 called a dekatron tube. These can serve as one digit of base decimal memory, as well as counters and readouts, since the lit element is visable.
• http://www.eetimes.com/document.asp?doc_id=1279395 Logic gates made from Neon Bulbs and photocells.
• https://books.google.com/books?id=JhgoAQAAMAAJ&q=%22memory%22 General Electric Glow Lamp Manual
  https://archive.org/details/GE_Neon_Lamps_1965
• https://www.muffwiggler.com/forum/viewtopic.php?t=70250&sid=82bd0603eb7e3d6def6507d2607e2f69 Discussion.

  .