October 28, 2026, “Curiosity” section
Sometimes the history of technology is driven forward not by geniuses, but by the humblest components. Let’s start with the K52 capacitor—a sealed can the size of a thimble: inside, a porous tantalum core soaked in liquid electrolyte; the silver casing serves as the outer cathode plate. Its price on the market back then—pennies. Its reputation—the opposite: this very part went down in history as the likely culprit behind the loss of Fobos-2 (Phobos-2), the USSR’s last interplanetary probe.
I’ll tell it in three layers: what this component was, how it broke the mission, and why the engineer who lived through that accident predicted its repetition in 2004, long before it happened.
Tantalum capacitors came out of the golden age of Bell Labs. In 1950, R.L. Taylor and H.E. Haring described the technology: a tantalum anode with an ultrathin oxide film as the dielectric. In 1954, Preston Robinson of Sprague Electric patented a practical design, and on June 26, 1958, Sprague secured the sealed liquid-electrolyte casing—the very “wet slug” that survives to this day virtually unchanged.
The idea remains elegant: maximum capacitance in minimum volume, stable characteristics, the ability to operate... where nothing else can—in military and aerospace hardware. In the USSR, the family was assigned the K52 index: K52-1, “volumetric-porous hermetically sealed”; K52-2—also sealed, in a metal cylinder. They were made in Novosibirsk, at the Oksid plant, and in Sarapul; the production line is still alive today—Elecond JSC continues churning out K52-1…K52-30 to this day.
What did they pay for that compactness? Liquid electrolyte, which behaves unpredictably in a sealed casing under extreme conditions. And a silver casing—a superb conductor that, under certain conditions, starts branching into crystals inside the can. This is not some exotic phenomenon, but a known effect: silver migration in the presence of moisture and voltage. A sealed casing with liquid electrolyte provides perfect conditions for it.
The recollections of V.A. Molodtsov, technical head of the flight-control group, are the primary source for this story. He writes that even before launch, during endurance testing, silver-crystal growth was detected on the plates of the K52s. For an interplanetary probe, this meant the risk of a short circuit in the secondary power supplies (VIPs) of the onboard digital computer, the BUK—the spacecraft’s computational heart.
The option of “replacing all the capacitors” meant the VIPs would have to be redesigned almost from scratch. The option of “moving the launch to the 1990 window” meant rebuilding the spacecraft: different ballistics, different mass requirements. They chose a third path: “they relied on Russian ‘avos’ and launched on schedule.”
What followed was the chronicle of a predictable catastrophe, stretched across eight months:
Take note: triple-channel redundancy, designed precisely to survive single failures, failed exactly as the textbook says it will. When the same component sits in every channel, redundancy turns into three copies of the same failure point. In SRE language, that is a common mode failure—the most infuriating kind, because it makes redundancy a useless waste of mass.
There was not enough telemetry to reliably reconstruct the entire sequence of the accident. But the coincidence speaks too loudly: on January 21, the VIPs had already gone down as a trio; on March 27—again as a trio, and that was exactly when contact was lost. Molodtsov and most project veterans are certain: it was that very capacitor that finished off the probe.
The final chord of this story comes not in 1989, but in 2004. Molodtsov’s recollections end with a prophecy: the Fobos-Grunt (Phobos-Grunt) then being prepared “repeats the same mistakes—a fundamentally new spacecraft on its first flight must solve an extraordinarily difficult task.” And, for added bite: “only a pale-face can step on the same rake twice.”
It came true with frightening precision: on November 9, 2011, Fobos-Grunt failed to leave Earth orbit, got stuck there, and on January 15, 2012, fell into the Pacific Ocean. The official version from the interagency commission was “heavy charged particles” that damaged memory chips. IEEE Spectrum then asked snidely whether that was not a little too convenient—a pretty two-word pairing, “chips” and “space,” that required no search for those responsible on Earth.
The moral of this story is not “Russian engineers are bad”—not at all. The mechanics of failure are universal and reproducible in any country and any company:
Modern engineers read this story as a textbook example: common mode failure cannot be cured by the number of copies; it can only be cured by diversity—different components, different suppliers, different principles. Three identical capacitors in three channels are not three chances. They are the same landmine three times over.
And meanwhile, tantalum K52s live on: Elecond still makes them, for military and space-certified applications. A technology from 1958 that outlived both the USSR and avos.
Sources: