25 May
2005
25 May
'05
4:01 p.m.
If I recall, the 'famed' 'review of computer memory technologies' article
(1953) by Eckert[*] included a mention of a 2D matrix of neon lamps, but it was more of a
proposed technique than an implemented one.
I think the problem with neon lamps (or cold-cathode gas-discharge devices) for memory in
a 2D matrix of any size was the widely varying characteristics of individual devices. The
selection of voltages to keep every device happy or the selection of enough devices with
sufficiently matched parameters was problematic. (There are at least two parameters - the
strike voltage and the dropout voltage - of consequence.)
Even gaseous voltage regulator tubes, specifically targeted for a particular voltage (such
as VR150/0D3) and with only one parameter to worry about, have quite a variation between
tubes.
[*] Reprinted in IEEE Annals of the History of Computing, V20/N4, OctNov 1998.
The article also mentions other bizarre suggestions for memory such as
temperature-sensitive pigments, coherers,
corona-discharge, etc.
Eric Smith wrote:
Tom Jennings wrote:
For neon, from flawed memory, strike-over is
around 100V, and
sustain around 80V. [...] You could probably make a coincident-voltage
memory from them... but they'd be slow I think, but pretty.
Sounds like a fun project!
Lessee... If the numbers are 80V and 100V...
You have an X-Y matrix. Steady state, you apply +7.5V to each X line,
and +97.5V to each Y line, so each tube has 90V across it.
You turn on a particular tube by dropping the X line to 0V and raising
the Y line to +105V. The selected tube sees 105V, while the other tubes
in the same row and column see 97.5V.
I'm guessing that if you measure the current in those lines, you can
tell when it strikes, or maybe you just wait long enough before dropping
the lines back to the steady state.
To turn off a particular tube, you raise the X line to 15V and drop
the Y line to 90V. Now there's only 75V across the selected tube,
while other tubes in the same row and column will still have 75V.
Again, there should be a current drop when the selected tube turns
off.
And you read a tube either by trying to turn it on, or trying to turn
it off. If the current doesn't change enough, it hasn't flipped.
The voltages will need to be controlled fairly well, since the
margins are narrow.
Will neon bulbs have reduced lifetime if you run them on DC?
Eric