CLASSICCMP digest 241

From: William Donzelli <william_at_ans.net>
Date: Fri Nov 21 20:23:53 1997

> I've done it, and it works, to an extent. I once built a binary
> counter out of neon bulbs (i.e. several flip-flops strung together)
> and with with some sorting out of neon bulbs and careful setting of
> B+ voltage got it fairly stable. Unfortunately, the ambient light
> in the room shifted operating points around considerably, as the
> ionizability of the neon bulbs shifts around depending on the
> ambient light level.

Yes, light is a factor. Some old telco equipment has lights (non-neon)
inside the case, just to stabilize things.

I was thinking about one bulb per bit. Imagine this: the memory is laid
out as an X Y grid, just like core plane. Each intersection has a bulb
(with diodes, resistors, etc.) to a common. Under idle conditions, each X
and Y line provides just enough current to keep the bulb voltage at the
keep alive level. To set a bit, the corresponding X and Y lines would get
a voltage boost - just enough that the correesponding bit fires, but not
enough to fire an entire row. Clearing a bit would work in reverse - drop
the X and Y lines just enough to kill the right bulb only. Reading a bit?
Well, I am still working on that. Perhaps it could involve a destructive
read (kill the bulb), and seeing what the common current does (no change
if the bulb was already off, a little change if the read killed the
ionization).

For stability, I suppose one could go to the bigger VR tubes: 0A2, 0A3.
0D3s are very common, dirt cheap, pretty when they glow, and the classic
"bottle" shape.

> Neon bulbs certainly are readily available on the surplus market; I've
> seen ads for large lots of NE-2's at a couple of cents a piece.

I have a large lot (several hundred pieces)! Actually, neon bulbs are
still used and still being made. And yes, they still are cheap, too!

William Donzelli
william_at_ans.net
Received on Fri Nov 21 1997 - 20:23:53 GMT