by Tom Jennings, tomj @ wps . com; last updated 31 Jan 1999
The capabilities of informational displays have
historically lagged other information and computation technologies,
just as they do today. Unlike current bit-mapped displays,
relatively un-bound to the symbols they present, historically
display devices have generally had a one-to-one symbol-to-mechanism
mapping -- character(s) silk-screened onto translucent substrate,
back-lit with an incandescent lamp, a motor-driven wheel of thin
metal flaps with painted numerals, or a labeled panel lamp.
For a 20 year period starting in the early 1950's, Nixie indicators, aka "nixie tubes", were the dominant high-density digital (and rarely alpha) display device. It seems so trivial now... but it was a vast improvement in information density, and a numeric value displayed on a row of Nixies at least approximated how actual people wrote it down on paper. Of course there were a lot of copy-cats and genuine improvements, some still in use today.
Tangentially related are various forms of decimal counting tubes, such as gas filled three-phase or "glow transfer" counting tubes, and Burroughs' crossed-field decimal counting tubes, developed in the peculiar period of proliferating technology families that coexisted and intermingled before lithographed semiconductors took over the world.
Nixie indicator tubes are character displays based upon vacuum tube technology. Essentially complicated neon glow lamps, Nixies contain ten thin metal electrodes, shaped into digits or symbols (. or + or -), and arranged in a small stack visible through the glass envelope. Like a neon bulb, a digit-shaped electrode glows orange when a voltage is applied between it and a common anode. Rather than new technology, this was creative use of existing tried and true.
Nixies are quite readable; they produce a nice non-flickering orange or red/orange light, reasonably visible in daylight, and they last a reasonably long time (5000 hours typical).
Under the guise of World Power Systems, I make a Nixie-based clock, the Model 11 series.
The characters in early models were all 0.3"-- 0.6" digit height and visible from the top/end of a bulbous glass envelope; they eventually became available in a range of sizes up to 2", and later more commonly seen with the stack of digits viewed from the side of the tube, in a conventional miniature radio tube T5 outline. One undesirable aesthetic concern is that since the digits are stacked front-to-rear, unlit digits partially obscure the lit digit, and the digits seem to shift to and fro as they change, since the stack of digits is about a quarter-inch deep.
The original end-view version shown at right is from Burroughs Corp. 1958 ad copy; the digit "8" was not an arbitrary choice, as it's generally the 2nd-from-top digit in the stack ("3" on top), and looks pretty nice; some, like "0" and "1", are behind a forest of dark digit electrode metal. Note the well-formed digits, in my opinion nicer to look at than modern seven-segment displays.
Designed by the Haydu brothers, George and Zoltan(1), probably 1952--1953, Burroughs Corp. bought the design and developed it into a product by project manager Saul Kuchinsky 1953--1955 (see Burroughs Corp WWW reference). Lore has it (EE Times web page, could not re-locate reference) that the name "Nixie" came from a working acronym for the new product of "Numerical Indicator eXperimental", e.g.. NIX-1, and the name stuck.
Most Nixie tubes are driven by 1-of-10 demultiplexers, however, a bi-quinary version exists. (Bi-quinary is a compact way of representing decimal numbers under certain design criteria; some early decimal computers used bi-quinary logic internally. It consists of four bits of data representing 1,3,5,7,9 and 0,2,4,6,8 and a 5th bit to select even/odd, more or less.)
There is a standard TTL integrated circuit to drive Nixies -- 74xx series TTL family 7441 and 74141. These convert TTL-level binary-coded-decimal to 1-of-10 high-voltage open collector drivers. Now utterly obsolete, they haven't appeared in databooks since the early 70's, though I put a copy of a 7441/74141 datasheet from an 1972 National Seminconductor databook here on WPS in the Archive section; the chips are in fact still made by NTE Electronics and sold through Mouser Electronics. January 2002 warning: it turns out that the 1998-date-coded NTE chips sold by Mouser are in fact surplus chips, cleaned up, their tops painted black and reinked with NTE's logo and false date! Two of us lifted the old paint off with a razor blade and found that the NTE chips were in fact TI chips from the 70's! Every chip sampled was old. It wouldn't be so bad if they weren't charging $4 for them and calling them new. Mouser claims to be unaware of the situation. Thanks to David Forbes for discovering this little deception.
Synergistically, Beam-Switching Counting Tubes (see below) directly drive Nixies without need for [once] exotic high-voltage transistors. An example of a Burroughs DC-105 decade counter subassembly is shown here. It contains one beam-switch tube (largest cylindrical assembly), a Nixie poking out the front and two vacuum tubes for glue on the back (one visible here).
Other display variations: Pixie indicator tubes contain their ten small digit-cathodes arranged in a circle like a clock face instead of a stack; otherwise they are electrically similar to Nixies. All 10 digits are visible in the glass envelope, but only one glows at a time. (This is by convention; you could light more than one digit in either Nixie or Pixie but it would not be useful.) These fell from favor rather quickly. Trixie indicators were complete Nixie display assemblies with transistorized drivers. [Burroughs Corp. datasheets; catalog 918A (6-61-5M); fact sheet, undated; May 1961 price list; Trixie transistorized readout datasheet, 11/60]. Burroughs Corp. also made various products based upon beam-switching and Nixie tube technology; including counting and scaling lab instruments, OEM-type counting subassemblies, the Beamplexer, a beam-switching-tube-based multiplexer for creating multiple virtual channels on a conventional oscilloscope, and an Information Display Board designed for airlines, bus stations, etc. using Nixie tubes, shown here.
As a footnote to all this, yes, Burroughs Corporation was in fact William Seward Burrough's family. Allegedly the father or grandfather sold out of the business sometime in the 20's, as he saw no long-term future in the calculating business. While possibly a strategical mistake, they did well cash-wise, and it provided young Bill with a monthly check to fund his writing (and ancilliary activities) throughout his life.
In the crazy days of early computing, approximately the late 30's through the mid-50's, every imaginable (and now unimaginable) mixture of technologies were tried as paths to reliable computer/calculator design. In essence, the major impediment was density -- it was known that practical computing devices would need hundreds or (gasp) thousands of logical and arithmetic building blocks, each tiny block requiring a minimum of one, and sometimes a dozen, vacuum tubes and discrete components. (For comparison, installations with a hundred vacuum tubes was considered "highly complex" in the 1930's and early 40's.) Therefore, many things were tried that performed inherent computational functions -- operational amplifiers for multiplication and addition in analog computers, multi-position mechanical and electronic devices for counting and adding, and specific physical phenomena to emulate data or computation. There was also a fantastic mixing of technology types during this period that is hard to grasp today -- vacuum tubes, precision mechanics, physical side effects (eg. electrolytic tanks), switches, lamps, paper tape, semiconductors and human intervention -- all combined into a single device.
Decimal counting tubes were developed in the early 50's from known but previously-useless phenomena to perform functions that otherwise required a chassis full of tubes and discrete components. The design life of such things was not long -- broadly speaking, the devices below were in use from the very late 40's through the very early 60's, when even mediocre circuitry based upon lousy transistors were superior to just about everything else before it. Ericsson Telephone Ltd. was a major manufacturer, which sold them under the name "Dekatron".
NOTE: Many of the technical descriptions and references are from "Digital Computer Components and Circuits", R.K. Richards, Van Nostrand Co, 1957, and referred to here as [DCCC].
Sometimes called "glow-transfer" counting tubes, or Dekatrons (Ericsson's trademark), these noble-gas-filled counting tubes are based upon plasma/arc strike/break hysteresis physics. In addition to counting, the internal electrodes glow visibly, and hence are arranged to be visible through the end (top) of the glass envelope, to show the counter's current value. It is common for multi-decade counters and scalers to have the counting tubes poke their heads out the front panels, with each digit's visible "bits" labeled 0 - 9. (Type 6802 shown here, its octal base sitting on a scrap of brass for the photo.)
Various electronic circuits shape the pulse(s) to be counted into overlapping pulses that "bump" the plasma arc from one electrode to the next (hence the "glow transfer" from electrode to electrode.). Counting is done by picking off a signal from the Nth (generally 10th) electrode; an output pulse is thereby generated every N (10) pulses. (In fact most had two or three electrodes per digit-bit to perform the counting, but the general principle is the same.)
Each counting cycle takes anywhere from hundreds of nanoseconds to tens of microseconds to complete, since gas ionization is a slow bulk process; 100,000 counts per second is considered fast, a million pulses per second is about maximum. The IBM Electronic Statistical Machine, Type 101, for tallying punch-card results, allegedly contains three-phase tubes.
To the left is a reasonably typical device, sold by the Erie Instrumentation division of Erie Resistor. This four-decade presettable counter was designed for process-control (counting turns of wire on a bobbin, for example) rather than general lab use. Shown also is a closeup of the left-most pair of decades, showing (only slightly) more clearly the electrodes and panel labelling.
|Simplified schematic showing phasing of the three glow-transfer electrodes per counting step.|
This rather obscure device merits mention mainly because of the synergy of design with the Nixie tube, and in contrast to the directly-viewable gas counting tubes of similar function.Crossed-field counting tubes rely on peculiar electron/magnetic flux interactions known "since the earliest development work on magnetrons" [DCCC p420], which is anywhere between 1920 (magnetron diode invented by Albert W. Hull at General Electric Res. Lab., Schenectady NY) and British magnetron development in 1938--1940 ["The Invention that Changed the World", Robert Buderi, Simon & Schuster, 1996, pp 84--]. "Crossed field" refers to the fact that the magnetic and electric fields are perpendicular. In a magnetron, a beam of electrons rotates rapidly past cavities under the compression of a magnetic field; in crossed-field counting tubes, are replaced with beam-forming electrodes that precisely "bump" the electron beam from one electrode to the next.I believe Burroughs Corp. was the most popular manufacturer of crossed-field counting tubes (they called them "beam switch" tubes).
I have an N.O.S. Burroughs 6700 in the box with pinout sheet and connector; it has a large, discrete magnet encircling the glass envelope. I also have a number of more highly evolved models, type Burroughs BX-2003 and BX-2004 "BEAM-X Switch" tubes; these have only a mu-metal shield, the magnet being internal, and the overall device much smaller. I have no data nor references on these latter devices. L.M. Ericson Co (Sweden) , produced a similar tube, type AD3, according to [DCCC].
As mentioned above, the synergy with the Nixie indicator is that the counting tube electrodes directly drive the digit-cathodes of the Nixie tubes, "effectively replacing 18 transistors (10 high voltage ones) and forty diodes", according to Burroughs Catalog 918A, presumably a good thing in 1958. Catalog 918A also refers to miniature beam switching tube types BD-203 and BD-316, so likely there are a whole raft of these things languishing in dustbins around the world. Bulletin 826 and Supplement 1 describe the Burroughs line of beam-switching tubes -- I do not have.