The Numitron: An obvious idea that wasn't very bright
Ah. The seven-segment display. Just make an 8 out of some sticks, and then you can represent the Arabic numerals zero through nine just by taking sticks away! You can even make some letters happen if you’re clever about it. These things have been around forever: the earliest known representation of them is in this 1910 patent where Frank Wood claims to have devised this special monogram arrangement. Personally I’m a little skeptical that he was the first to think of this, especially since this earlier patent exists where George Mason devised a segmented alphanumeric display which is even more complicated, and creepy looking, but that’s not important to this video. Going back a century you might find these boxy numbers in price signage or perhaps scoreboards at a sports stadium - anywhere a permanent display which needed to show different numbers made sense.
Usually large so many people could see them, the segments might be lit with light bulbs (as Frank Wood envisioned) but they could even be mechanical in nature with differently colored flaps hiding or revealing the individual segments. But for many years, they remained pretty obscure. The seven segment display wouldn't really start to take off until the digital age. And once we got there and they started spreading, our old pals at RCA smelled opportunity so they got out the crayons and took a crack at it. And these are their handiwork.
They’re called Numitron tubes, and they are terrible. They’re also rare and frustratingly expensive, probably because they didn’t sell very well at all. Because they’re bad! This is what I can only describe as the crudest possible way to produce a technically functional seven-segment display for digital devices. Now you might be able to tell what’s going on in here just by looking at them, but before I explain their absurdity I want to go back in time because by doing so, you’ll understand why these briefly made sense.
Briefly. Once upon a time, there were computers. At first very large, very expensive computers, but we see the potential! As computing technology got cheaper and microprocessors started showing up in things like scientific equipment, cash registers, control consoles, and whatnot, we needed displays to allow humans to read and understand the data those processors were spitting out. Of course a general-purpose computer could justify using a CRT monitor and generating a video signal to display columns and rows of text, but something like a calculator didn’t need all that.
It just needed to show some numbers. So, cheap and simple numerical display devices which microprocessor-based systems could control became a very hot item. Over the years equipment manufacturers tried all sorts of ideas. One of the most famous is the Nixie tube. These neon-filled tubes produced a glow discharge around wire cathodes formed in the shape of numerals, resulting in beautifully legible displays which were readable in many conditions. But they had a rather huge caveat - they required high voltage DC to produce that discharge.
As in, around 180V. That meant that while the tubes were simple, the circuitry required to drive them was complex. Not to mention dangerous. Because of this, Nixies quite naturally attracted plenty of competition. But, many of its contemporaries shared another of its flaws: the need for a dedicated output for every character it can display. Since Nixies use wire-form numerals, a standard Nixie tube needs 10 inputs to function: one for each of its cathodes.
But a segmented display which forms numerals by combining different shapes together only needs one output per segment. That saves on wiring, but it also makes the circuitry which actually drives the display devices cheaper. In the early days, dedicated ICs known as BCD decoders would translate the four bits of binary-coded decimal input they received from a microprocessor into the correct output to drive a display device. The internal structure of these decoder chips was quite simple - just a few logic gates, really. But they could be a lot simpler with a seven-segment display compared with a Nixie tube since there were fewer outputs for it to control.
And, many of the same outputs are active on multiple inputs. A 2 becomes a 3 simply by swapping these two segments, and that reduces the number of transistors you need in the chip which makes the chip cheaper. That’s extremely in the weeds, though - when we get right down to it, the simple fact was the seven-segment display had too many advantages over Nixies, Nimos, edge-lit displays, projection displays and all the other wild ideas that were floating around. So although they were arguably less readable and more ugly, they were obviously going to become the standard display technology for numerical information. Except - it’s not really a technology. It’s just a specific graphical representation for numerals.
So using them may simplify control circuitry and make wiring devices a little easier but we still have to figure out how to actually make them! And that wasn’t trivial. In the late 1960’s, our technology to make small things light up brightly was quite limited. We did have teeny little light bulbs, of course, and we used those in myriad ways (including those projection displays) but doing that had a ton of drawbacks, especially visibility in different lighting conditions. To maximize contrast under bright ambient light, ideally the glowing element of the display should be directly viewable - which was one of the Nixie tube’s greatest strengths. So you might just make a seven-segment nixie tube, which was indeed a thing (more or less). But glowing neon cathodes still needed the same high-voltage DC as ordinary Nixies which was annoying and expensive to implement.
Luckily, the vacuum fluorescent display had just popped on the scene. These devices required low-er voltage than neon discharge tech which was good, but the tubes themselves were more complex. These vintage IV-11 tubes might look simple enough but a closer look reveals there are several layers here. At the front of the tube is a cathode made of two heater wires which emit electrons through thermionic emission. There’s the seven segments forming an 8 at the back, of course, but look closely and you’ll see that right in front of the 8 is a control grid.
In operation, that grid is positively charged and will absorb the electrons coming from the cathode wires at the front of the tube. But the individual segments behind the grid can also be positively charged. When energized, they also attract electrons and are able to yank some of them through the control grid. When those electrons collide with the phosphors on top of the segments, they glow. Pretty neat. But that’s even more complex than nixies! And for every problem they solved they introduced another.
These tubes didn’t need high voltage, but they needed two voltages: a low voltage for the cathode wires and something around 30 volts for the anodes and control grid. Better than 180, of course, but still much higher than the logic circuitry of whatever you want to put these in, which was typically 5 volts. And, as you might be able to tell, these early tubes have pretty poor contrast under even modest ambient lighting since the phosphors are white when unlit. So, both neon tech and VFDs had their problems. But we were already building a lot of electronic devices which needed numerical displays, and there wasn’t time to wait.
So, the fine folks at RCA barfed out an idea. Here’s my impression of how that went: What if… why don’t we just make a weird little light bulb? [fart noise] And thus explains the Numitron! First hitting the market in 1970, these really are just weird little light bulbs… but with seven filaments. Each filament is stretched across little pegs on a support board so they become the segments of a seven-segment display. Then that assembly got shoved into a standard 9-pin miniature vacuum tube enclosure. Just apply power to those pins to light up the right combination of those filaments and you technically have a functional digital display device. Oh, and this model doesn’t have it, but they were available with a decimal point option which took the form of a tiny little X crossing these four pegs.
These days, of course, the trendy thing to do with weird old display tubes is to build a clock so that’s what I did. But in the spirit of the Numitron, I didn’t bother doing it well! I did the bare minimum. But it technically works. I think that was the slogan for the Numitron! Now, I’m about to be rather unkind to this piece of alleged technology but before I embark on my tirade where I count all the ways this is just a terrible terrible product, I need to admit that I also kind of like them. I’ve always been fascinated with the idea of building seven-segment displays out of weird things.
Like, imagine a clock where each segment is a four-foot fluorescent tube. You could read that thing from miles away! And, uh, when I was in college, I built a giant clock where the segments were sections of LED tape. RGB, of course. So Numitrons are very much up my alley. And while today the use of incandescent filaments in a seven-segment display might seem objectively absurd, that was in fact the point! These tubes are dead simple and require nothing special at all to use. They only need 5V to glow, so they didn’t need their own power supply.
Incandescent filaments are self-regulating, too, so there’s no need to add current-limiting resistors to the circuit. And since they were so tiny the segments only consumed about 25 mA each. This is actually the largest version of the tubes RCA made, but they don't even consume a watt of power with all segments lit. The ridiculous simplicity of these meant that you could drive them directly from a seven-segment BCD decoder chip, all powered by the same 5V supply as the rest of your newfangled digital widget.
That would save money in the hardware design, and since the Numitron itself is so freaking simple, the display tubes were cheaper than the competition. So it wasn’t a terrible idea. But, as you might be able to tell just from looking at them, the execution of the idea left quite a lot to be desired. If I could describe these in a word, it would be janky.
Or perhaps slapdash. Nothing about these feels like a finished product that should exist. They just look way too crude! The extremely thin segments lead to them being not very legible, and the arrangement of those segments is just… awful. There’s way too much of a gap between them and the vertical segments extend uncomfortably beyond the top and bottom segments which just looks weird.
And they didn’t lean the 8 to accommodate the decimal point at the bottom so the display is off-center in the tube which is infuriating. But its problems don’t end there. Do you notice something wrong about the background? For reasons that I cannot fathom, RCA decided to make the support board for the filaments grey. That was a mistake. It should be black, or at least as dark as possible. Because, ya see, any ambient light that hits these things is going to brighten that background and make the glowing filaments harder to see.
And, if that weren’t enough, even in a pitch-black room the filaments themselves illuminate that support board enough to where contrast is reduced, making the already kind of hard to parse numbers just a little bit harder to identify. Now, I will give them the benefit of the doubt and say that there could be some materials reason this isn’t black - maybe a nice, dark background wouldn’t survive the heat of the glass-sealing process. But I mean, these VFD tubes have a much darker background so… And it might be hard to tell but these aren’t even straight - and I don’t mean my handwork when installing the sockets. The internal support structure holding the filaments is leaning noticeably in several of these tubes so with all the tubes fully seated, the numbers are slightly crooked and with multiple tubes you’ll never have a perfectly aligned display unless you commit to a lot of fiddling. Honestly, I don’t think anyone at RCA was particularly proud of these things.
I mean, if they were, I would think they would have tried a little harder. Just look at the datasheet. These are the other varieties RCA sold. A plus or minus tube. Wow.
And a plus or minus 1 tube. Really workin’ hard over there. In this promo piece about them, we don’t learn whose idea this was.
We just hear from Robert D. Reichert, the manager of the whole dang tube design department. And RCA didn’t bother putting a patent number or even patent pending on these boxes which are specifically branded Numitron. That made me wonder if they even bothered patenting it. Honestly the only nice thing about these is their name: Numitron.
That’s pretty cool. But it’s also lazy! This is literally just numi (short for numerical I guess) and tron. And back then tron was as infectious a buzzword as AI is today. I was starting to lose my mind over the origin of these things.
The idea seems simultaneously too obvious (this write-up in Popular Electronics even says as much) yet these are also just… not good. I could not rest until I found the patent and who invented these. Luckily I did, and here it is.
Apparently these were the brainchild of Richard Arthur Bonnette and Norman Lee Lindburg. Now, RCA was a huge research-driven organization back then so it’s plausible that these were just two names pulled out of a hat and I kind of hope it was. This does not scream quality. Editor’s note: while looking for a different patent (the one showing the projection displays which used little light bulbs) I ran across this patent issued in 1959 to Art Garfunkl - I mean A. R. Garfinkel. And… well, sure enough the idea was too obvious and RCA was not the first to have this idea. The implementation here is very different, though - it’s in a flat package (which is gonna come up again later) but most importantly, if I’m understanding the diagrams correctly, the filaments are producing light indirectly and it’s guided towards a mask with small slits at the front.
I’m guessing this concept never made it to production, and that drawing implies it’s got the widest bezels you’ve ever seen. But I wanted to throw it in here. You can see from the datasheet that RCA was grasping at straws trying to sell these things. Oh sure, low-voltage operation is very real and objectively a good differentiator.
But “void of clutter” is clearly a dig at Nixies, and the whole “oh, it’s incandescent with a wide spectral output so you can use color filters to obtain a display of any desired color” is true, but a stretch. I mean, sure, but good luck with blue. RCA designed them to last 100,000 hours and, well, that’s only gonna be possible if the filaments burn quite dimly. So they’re not producing much light towards the blue end of the spectrum and a blue filter will really cut their output. Now, to be fair the tubes can be a lot brighter than they appear in this clock.
They’re power-limited here thanks to the BCD decoders I had which are not capable of passing 25 mA. If I apply 5 volts directly to a tube, we can see that it can get this bright. Which is honestly impressive! That’s definitely brighter than a Nixie tube, and whiter than I was expecting. But this also really reveals how much the filaments are brightening the background and reducing contrast. That really shouldn’t be grey.
I mean you didn’t need a rocket surgeon to tell you that. While RCA didn’t seem to try very hard here, the concept was valuable enough to attract clones. There are plenty of variants available that were made in the USSR, but those are even worse! These IV-9 tubes have filaments that sag under their own weight. Later, incandescent seven-segment displays just like Numitrons but in flat packages appeared, and for some reason I have a memory that they were common in gas pumps for a brief time but literally I have no idea why that’s in my head.
I just read it somewhere somewhen. Those clones were all short-lived, though, just like the Numitron, because of a little thing called the light-emitting diode. They had already been invented by the time RCA got the Numitron out the door, but they were extremely expensive in 1970. And, the first seven-segment displays which used LEDs were tiny things where each segment was actually multiple LEDs on a tiny little chip, and they relied on magnification lenses to make the tiny glowing digital dots appear larger.
That’s why the Numitron made sense to use. But it didn’t take long at all for LEDs to get cheap, and once they did? The Numitron was doomed. LED packages like this quickly became the defacto standard for seven-segment displays.
These are much like those really early and large displays where ordinary light bulbs were stuck down holes with lenses on the front, but miniaturized by replacing the light bulbs with tiny LEDs. Those LEDs are firing up at small lenses embedded in epoxy resin and that forms an incredibly rugged and simple structure. The lens shapes can also be tweaked for maximum legibility, and resin’s a lot easier to deal with than glass, too. So once production lines for these things spun up the Numitron just looked ridiculous. And let’s not forget the vacuum fluorescent display. These things are still in use today - in fact there’s one in my stove.
The single-digit tube form factor found here became just as ridiculous as the Numitron before long, but its the problems with contrast in ambient light were easily fixed with filtering and soon we learned how to make VFDs flat and pack them with hundreds of individual segments. You might have 8 seven-segement displays all in a single package for a calculator, or even dot-matrix alphanumeric displays. Through multiplexing, a technique where the entire display is broken up into several sections addressed one at a time in rapid succession, we could make very complicated and very customizable displays featuring custom graphics and multiple colors, all with excellent readability, longevity, and simple control circuits.
Oh, right and then there was the liquid-crystal display. Which, funnily enough, was invented by George Heilmeier at RCA! And before the Numitron! It took quite a while to become viable, though, and RCA’s first version used a concept called dynamic scattering mode, and wasn’t super great. Other innovators improved upon it over the years, and I’m really only including it as a footnote since it wasn’t self-illuminating and it was not quite a contemporary of the Numitron, at least not ones that actually worked well. But once we figured them out the Numitron looked even sillier.
Still, I can’t help but admire the Numitron. It’s perhaps the perfect example of the phrase “so crazy it just might work” distilled into a product. Trouble was, while it did technically work, its advantages became irrelevant approximately seven minutes after RCA brought it to market. Of course that’s not entirely true, but I am all but certain that device manufacturers, when shopping for display technologies, looked at how poorly these render numbers and how jankily they were constructed and quickly concluded they’d rather pay more for something a little more polished.
Slapping these in your digital device screamed cheap. If there’s a lesson we can learn from the Numitron, it’s probably that there are times where the simplest possible option really is too simple. If it had a little more time in the oven and RCA tried executing it a little better, maybe it would have seen more success. But it was always going to be stuck with the shapes you can make with a filament. In other words, thin, straight lines. With every other display technology offering much more legible characters and even custom glyphs, the Numitron would never be able to branch out beyond seven-segment and maybe alphanumeric displays.
But I’ll bet those would look even worse than this. But hey, at least they managed to get it to market when it actually made some sense. Unlike that Videodisc player. Oh look, LEDs! ♫ slapdashedly smooth jazz ♫ I have to record that line again.
Ignore all the continuity errors.... It’s a running clock. For your benefit… [waits until the hour rollover] OK …and, since the Numitron itself is so dang simple, the tubes werech cheap chchpbbt. Fell apart.
The earliest known representation of them is this eh debekadababa Ah, the seven-segment display. Make an aight… what? The use of incandescent filaments in a seven-segment display might seem objectibely… objectibely? Because of this, Nixies quite naturally attracted plenty of commetition. But, I’ve… mmm. RCA really did some numbers here, didn't they? Or should I say, they really did a number on themselves? Perhaps this is the product that signaled the beginning of the end for RCA. Imagine if they had stuck Numitron tubes in the Selectivision players though.
Maybe that would have sold it!
2024-07-02 20:22
