Four Dark Sockets and Three Weeks to Wait
One hundred and seventy volts across a 22kΩ anode resistor. That’s 7.7 milliamps—well within the IN-14’s rated operating current. I’ve checked the math three times. The tube should glow.
It doesn’t.
The kit arrived this morning from a seller in Kharkiv. Six IN-14 tubes, a boost converter rated for 180V output, a bag of K155ID1 BCD decoder chips, and a PCB that looked professionally fabbed until I noticed the silkscreen was mirrored on one edge. The tubes themselves were packed individually in foam, each with a handwritten date on the base in Cyrillic: 1986, 1986, 1987, 1986, 1984, 1986.
I built the flight instrument clock back in February using IN-12 tubes. Those worked on the first try—orange glow, crisp digits, the satisfying depth of stacked wire cathodes behind glass. The IN-14 is a side-view variant: same Soviet manufacturing era, same mercury-doped gas fill, same upside-down “2” serving as the digit “5.” Different pinout, different socket, but the same operating voltage range.
Tube one lights at 168V. Dim at first, then steady.
Tube two: nothing. Raised the voltage to 175V, then 180V. Still nothing.
Tube three: flicker. The “7” cathode glows for half a second, then cuts out. Tap the glass—flicker again. Classic poor connection, either internal or at the socket.
Tube four: glow, but only on one side of the digit. The “3” shows as a backwards “C.” Cathode poisoning, probably. These tubes sat in a Ukrainian warehouse for forty years. If no one cycled them periodically, oxide deposits build up on the unused cathode surfaces. Sometimes you can burn it off by running higher current for a few hours. Sometimes you can’t.
Tube five: works. Orange, stable, all digits tested.
Tube six: the “1” glows blue-purple instead of orange. Mercury contamination from a compromised seal, or just a late-production tube with a heavier mercury dose. Not broken, exactly, but visually inconsistent with the others. It’ll stand out in a row.
So: six tubes, two working normally, one colour-mismatched, one with partial cathode failure, one intermittent, one completely dead. Sixty-six percent failure rate on “new old stock” components from forty years ago.
This shouldn’t surprise me. Yesterday I was winding the output transformer for the tube amplifier, and I broke the 42-gauge wire three times before getting the technique right. Vintage electronics punishes impatience. But with wire, you can start over. With NOS tubes, you’re limited to however many the Cold War left behind.
The troubleshooting process is tedious in a way that isn’t interesting to describe. Check the socket pins with a continuity tester. Measure the voltage at each cathode connection. Swap tubes between sockets to isolate whether the problem is tube or circuit. The boost converter is outputting correctly—176V, rock steady. The K155ID1 decoder is switching properly; I can see the output pins go low one at a time as I step through the BCD inputs. The fault is in the tubes themselves.
I’ve emailed the seller. He’s responsive—we’ve exchanged four messages since February—and I’m certain he’ll offer replacements. But shipping from Ukraine takes three weeks minimum, and I’ve already invested tonight in a build that can’t proceed. The PCB sits on the bench with two glowing tubes and four dark sockets, a thing that isn’t a clock and won’t be for weeks.
The original plan was to build something simpler than the flight instrument version. No METAR parsing, no aviation data. Just a clock. Six digits. Hours, minutes, seconds. The kind of project a first-time builder might attempt, because the code is trivial and the circuit is well-documented. I wanted to write about the basics: strike voltage versus sustaining voltage, why the digits are stacked 6-7-5-8-4-3-9-2-0-1 instead of sequentially, how the Soviet IN-14 uses an inverted “2” for the “5” to save tooling costs.
Instead I’m writing about dead components and the limits of vintage supply chains.
The two working tubes are glowing on the bench right now. The code is running: a simple counter that increments the seconds digit every thousand milliseconds. Watching a single Nixie count from 0 to 9 over and over isn’t satisfying in the way a completed clock would be. But there’s something to learn from it anyway. The transition from “9” to “0” involves all ten cathodes briefly—the previous one deionizing while the new one strikes. For a frame or two, both glow faintly, superimposed in the gas. It’s invisible at normal viewing distance but obvious when you’re staring at the tube from six inches away, waiting to see if it’s about to fail.
The seller says he’ll ship replacements Monday.