In any era, the story of electronics has very much been about figuring out how to make something happen with what’s available at the time. And as is often the case, the most interesting developments come from occasions when needs exceed what’s available. That’s when real innovation takes place, even if circumstances conspire to keep the innovation from ever taking hold in the marketplace.
This gem of a video from the Antique Wireless Association has a perfect example of this: the long-lost analog-to-digital converter vacuum tube. Like almost every mid-20th-century innovation in electronics, this one traces its roots back to the Bell Laboratories, which was keenly interested in improving bandwidth on its massive network of copper lines and microwave links. As early as 1947, one Dr. Frank Gray, a physicist at Bell Labs, had been working on a vacuum tube that could directly convert an analog signal into a digital representation. His solution was a cathode ray tube similar to the CRT in an oscilloscope. A beam of electrons would shine down the length of the tube onto a shadow mask containing holes arranged in a “reflected binary code,” which would later be known as a Gray code. The analog signal to be digitized was applied to a pair of vertical deflector plates, which moved the beam into a position along the plate corresponding to the voltage. A pair of horizontal deflector plates would then scan the beam across the shadow mask; where electrons fell on a hole, they would pass through to an output plate to be registered as a bit to be set.
Fast forward twenty years, and Dr. Gray’s basic idea was leveraged to build a 224 Mb/s analog-to-digital converter that just wasn’t possible with the transistors of the day. The innovation with this tube was to parallelize the output — instead of a single electron beam being rastered across the shadow mask at the appropriate position, a ribbon of electrons fell on an entire 9-bit row before striking an array of output detectors.
As usual for Bell Labs, the tube performed excellently, nearly matching the theoretical signal-to-noise ratio. But alas, another project in the lab to build an all-solid-state ADC had gained traction while the tube was being perfected, and much of what drove both projects fell apart as AT&T concentrated on microwave guideline and optical fiber for their digital networking needs.
As innovative as the vacuum tube ADC was, it never appears to have seen use in any production networks. But it’s still a fantastic illustration of what’s possible under constrained conditions. We’d love to see one of these tubes, if any still exist, resurrected and put through its paces.
Thanks to [Mark Erdle] for the tip.
“224 Mb/s analog-to-digital converter” – absolutely amazing for the 1960s!
I was lucky enough to contribute to two similar projects at Westinghouse R&D in the early 70s. Both started with Vidicon shells. In one, we photographed spoken phrases on an oscilloscope screen, vertically stacked them, and etched the image onto the oxide on a 1 inch silicon wafer. The tube addressed the phrases by moving the beam vertically and read out by scanning a “dithered” ribbon horizontally. Signal recovery was based on the difference in the secondary emissions between the etched and unetched areas. The concept was proven, but no application seemed to need such high speed access and inflexibility of content. But it WAS fun!
Info on a similar device : http://electricstuff.co.uk/glassadc.html
Gray code was used in the legendary Robot Research Inc. Model 400 Slow-Scan Converter.
The use of Gray code allowed for a less noisy representation of the image.
It’s described on page 17 in the manual.
https://www.manualslib.com/manual/1356478/Robot-400.html#product-400
1965… this was just merely before the advent of supercomputers clocked at 10MHz. Seems a solid state 12 Msps ADC was not really out of reach by then.
I saw one of these at the VintageTek Tektronix museum.
It was a side project they were working on in the 80’s for some digiscopes they had in development. I don’t think it ever made it to production but with their glass expertise I’m sure they could have done it.
The speeds sound real useful, so why come we no see these up into the 90s?? Did they work out super hyper expensive to mass produce. Did they expect semicon versions to get cheaper and faster sooner? Did they not expect tube tech to continue advancing as it did through cold cathode stuff and CRTs?
(Thought I posted something like this but it’s not here)
You appear to have watched the video before commenting
oops, I forgot to add the /s
Right, because theirs was the only possible use case for fast ADCs.
We lost a national asset when we lost Bell Labs.
Somehow we lost nearly all the great “Think Tanks”.
When I was a kid in the 60s, I got chewed out for being on the phone longer than 5 minutes, and my folks kept long-distance calls less than 10 minutes on the very rare occasions they called family two states away, with a cost pr minute of $0.50-$1.00 or more. Not long ago, 2 current and former colleagues called me from Denmark during a drinking session for $0.02 per minute. That’s the difference a monopoly makes, and how Bell Labs came to fund a crowd of very smart physicists. I suspect those monopoly dollars also funded a larger crowd of marketers who popped out “gee-whiz” press releases.
These days, the smart physicists are hired (in the US) by the Department of Energy Labs, and their way of sharing the fruits of their knowledge result in (for instance)beam time on the high-energy synchrotrons at several of the DOE labs, computation time on the DOE supercomputers, accessible to scientists all over by submitting a research proposal and the cost of travel time and scientists/grad students. I have been involved in projects where beam time let our scientists track diffusion speed of several types of ions in the wood cell wall at less than micron scale resolution, and generate a time-resolved 3D model of the ions and wood cell structures.
The brains still exist. They get paid by somebody else and they do different things, is all.