25 May
2005
25 May
'05
5:14 p.m.
Subject: Re: neon tube memories
From: ard at p850ug1.demon.co.uk (Tony Duell)
Date: Thu, 26 May 2005 00:38:52 +0100 (BST)
To: cctalk at classiccmp.org
Only takes 4bits to count to 10. Though it could be a switch tail
ring counter.
Unless someone has som ehard data on them being
used "digitally"
I'd assume the HP counter use was simply as generic, non-switching
circuit elements.
The HP5245 counter certainly used neons as memory elements -- I have the
instrument and the service manual which has a section on the operation of
the counter module.
Each counter module contains 8 transistors wired as 4 bistables. Diode
steering is added to make it count either in 1242 BCD code or 1248 (the
latter being one of the options available for the instrument).
The collectors of the transistors go to the neons via resistor (and
diode?) networks. By varying the voltage on one of the lines you can
transfer the state of the counter to the 8 neons and latch it there.
The neons are mounted in a plastic block with a thick-film circuit on
top. The thich film elements are CdS photoresistors arranged in a decoder
tree which directly drives a nizie tube.
Quite a hack to get a counter, latch and display driver in just 8
transistors! A write-only
neon memory would still make a nice addressable
display!
Somewhere I have a neon display which seems to be a dot-matix unit (7 dots
high by perhaps 100 long) with transfer electrodes like a Dekatron tube.
You apply 7 'bits' to the right hand column, then toggle the transfer
electrodes appropriately and all the dots move left one column. Repeat to
built up the pattern you want in the display.
-tony 
6:21 p.m.
Tony wrote:
Quite a hack to get a counter, latch and display
driver in just 8
transistors!
Allison wrote:
Only takes 4bits to count to 10. Though it could be a
switch tail
ring counter.
Brent wrote:
Flip-flops: 2 transistors per bit. 4 bits * 2
trans/bit = 8 trans.
But 8 transistors used in any obvious way does not also get you a
latch and display driver in addition to the counter. That's what
makes the use of the neon lamps in the HP design exceptionally clever.
Eric
26 May
26 May
12:53 a.m.
Tony wrote:
Precisely.
Quite a hack to get a counter, latch and display
driver in just 8
transistors!
Allison wrote:
Only takes 4bits to count to 10. Though it could
be a switch tail
ring counter.
Brent wrote:
Flip-flops: 2 transistors per bit. 4 bits * 2
trans/bit = 8 trans.
Eric Smith wrote:
> But 8 transistors used in any obvious way does not also get you a
> latch and display driver in addition to the counter. That's what
> makes the use of the neon lamps in the HP design exceptionally clever. 
6:48 a.m.
>>>> "Allison" == Allison
<ajp166 at bellatlantic.net> writes:
Allison> Boroughs Panaplex, I have a 32char version that I power up
Allison> and it works. I've designed using the 40 char version
Allison> pretty cool. Even got to visit the facility in NJ back in
Allison> '73 where they were made along with the other gas tubes
Allison> (nixe and the like.).
I don't believe a Panaplex uses the memory properties of gas
discharges. But the PLATO terminal panels did.
I have a short (4-8 pages or so) technical description of the
mechanisms and the waveforms involved. Some of the magic has to do
with stored charges on the walls of the gas cell. In the PLATO
panels, the electrodes apparently have a thin insulating layer
deposited on them, so they are not directly exposed to the neon. The
"sustain voltage" is a fairly complex AC waveform, and the addressing
is done by superimposing additional pulses on top of that, in a
"coincidence" addressing scheme.
All this worked very nicely in a quarter megapixel display (several
thousands were made) and in bigger ones too (I've seen military
displays of a megapixel -- 16 inch squares.)
paul
5:12 p.m.
Quite a hack to
get a counter, latch and display driver in just 8
transistors!
Only takes 4bits to count to 10. Though it could be a switch tail
ring counter. A
write-only neon memory would still make a nice addressable
display!
Somewhere I have a neon display which seems to be a dot-matix unit (7 dots
high by perhaps 100 long) with transfer electrodes like a Dekatron tube.
You apply 7 'bits' to the right hand column, then toggle the transfer
electrodes appropriately and all the dots move left one column. Repeat to
built up the pattern you want in the display.
-tony
7148
days inactive
7149
days old
6 comments
6 participants
participants (6)
-
ajp166＠bellatlantic.net -
ard＠p850ug1.demon.co.uk -
eric＠brouhaha.com -
hilpert＠cs.ubc.ca -
pkoning＠equallogic.com -
tomj＠wps.com