Delay lines are valuable tools. We can use them in various fields, from electronics, physics, telemetry, or even arts. In practice, a delay line … delays the delivery of a signal in time. You can think about it as a holding pattern that allows slowing down information delivery. This characteristic is critical in telemetry when the measured phenomenon is short-lived compared to the data recording time. For example, in an underground weapon’s test, the sensing information travels at a good fraction of the speed of light – that’s good, you want to get out fast – but then it needs to be delayed before the lossless data consumption. You would be surprised how omnipresent they are!
In 1953, John Presper Eckert, Jr. and John William Mauchly, the Electronic Numerical Integrator And Computer (ENIAC) fathers, patented a memory system using torsion delay lines (US Patent 2,629,827). Several technologies have been used since to implement effective delay line memories with increasing storage capacities. A few remarkable technologies are piezoelectric, acoustic waves propagating in mercury, and photons.
Torsion delay lines are twisting a metal wire for a short time to store information. The wire, fed through a transducer equipped with electromagnets, has two nickel wires welded onto it, so the wire is sandwiched between the nickel wires. Since nickel is magnetostrictive, the electromagnets change the length of the nickel wires temporarily. While one nickel wire pulls the wire in one direction, the other pulls it into the other, creating, in turn, a torsion in the primary wire. The torsion wave then propagates in the wire. The nickel wires are anchored into a massive – compared to the nickel wires’ gauge – dampening material (the extended white rubbery substance that looks like a ceramic cement power resistor).
For each bit to store in a torsion delay line memory, a torsion is created in the wire, which propagates along it. The encoded data is, therefore, a sequence of well-separated torsions traveling along the line one after the other. This explains why delay line memories have a sequential read/write. Quite similar to the fantastic bubble memories I talked about here. As you certainly guessed, the length of the wire tells how many bits you can store in it. This explains the use of coils. Indeed, such arrangements allow packing a few meters of wire in a relatively small footprint. Of course, the longer the wire, the longer the access times.
Regardless of the technology, delay line memories need to be refreshed. In a torsion delay line memory, this is done once a torsion wave reaches the end of the wire, where it is detected and converted back into an eclectic signal using the exact mechanism used to create it in the first place. This signal is then reinjected at the other end, hence the refresh. The dampening material at both ends limits the reverberation of the signal. Last, at the input and output, simple logic gates – along a clock signal – are used to read or write bits from/into the memory.
To illustrate today’s post, I photographed my 1979 USSR memory module built by ЭЛЕКТРОНПРИБОР. These memory modules were used in desktop calculators and could store from a few hundred bits to a few thousand, and had read times of the order of 500 us or faster. In the 60s’ these memories were cheaper than core memories (here) and were used by many manufacturers, such as Olivetti (Programma 101), Monroe (Epic 3000), or ЭЛЕКТРОНИКА, possibly the Iskra-122 (here). I find this technology simple, elegant, and quite impressive for what it is.
Of course, instead of a wire, you can use mercury, as mentioned earlier, or any advanced media. Actually, photons are used today to delay highspeed signals in space applications. Among the possible medium, one could pick water. Indeed, you can design a binary computer using water (that’s for another day 😊). Remember I mentioned Arts as an application domain for delay lines?
That’s what Melissa Dubbin and Aaron S. Davidson did in 2019 for the Okayama Art Summit (here). Their exposition is described as follows. “Delay Lines draws from their research and work with silica-based mediums such as glass, integrated circuits, and soft robotics. Dubbin & Davidson’s work was supported by Prof. Shuichi Wakimoto, Okayama University System Integration Laboratory. Custom borosilicate glass elements for the work were fabricated at UrbanGlass. The work has been acquired by the Ishikawa Foundation, Okayama, Japan.” True, this is not a delay line per se, but since I like it when science, technology, and art meet, I could not resist mentioning their work.
Happy Fourth of July!
Thanks for the article!
One thing, I learned recently, is that delay lines were used for basic timing. E.g., the PDP-1 doesn’t have a central clock, but various timing networks built from delay line modules. (I guess, with all those miles of wiring, localized timing makes sense.)