Witchcraft and Alchemy sputtered platinum onto glass with a garage vacuum rig to make a Panaplex-style display

A homemade Panaplex display in Kapton tape and wires, its segments glowing violet, beside a platinum periodic table tile — Image: youtube.com

“Of course, it’s completely overkill and it’s really hurting my wallet, but that’s not the point,” says the maker behind the YouTube channel Witchcraft and Alchemy, about using platinum for the digits. “I just trying to make something cool.” (Quotes here are from the video’s auto-generated captions, grammar and all.)

It started with a Nixie tube the maker broke. After gluing the pieces back together, the tube went into a vacuum chamber, wired through a feedthrough, to see if it still lit. There was no neon on hand. “But I do have argon for welding, which still should work,” the maker says, and instead of the usual orange glow it gave “A deep bluish violet glow.” The pump turned out to be too strong for the argon valve, so a second valve went in between the pump and the chamber to throttle it.

Play the video on YouTube

The platinum came from a side project. The channel had already posted a garage-built PVD DC magnetron sputtering gun, described as made “from some random parts and scraps i found,” and sputtering meant the digits could be masked into any flat shape on glass. For the anode the maker picked glass with “a layer of indium tin oxide on its surface,” which makes it conductive and keeps it transparent. The first test cell used a piece of 3D printer filament as the spacer. “The filament was a bad idea though.”

Then came the homework nobody had published. There isn’t “a lot of information” on what pressure and voltage a flat argon cell like this wants, per the video, so the maker hooked a photodiode to the cell and logged pressure, voltage and current. Breakdown hit “a crazy high 270 volt” at 0.6 and 2.5 millibars and dipped to around 210 volts between 1.2 and 1.5. One square centimeter came close to two watts. The maker settled on 1.5 millibars, while calling the pressure readings “complete garbage,” since the gauge “was salvaged from the trash” and isn’t calibrated for argon.

The shape came out of the parts bin. Sputtering only works in 2D, the maker didn’t want to route a seven-segment display’s middle bar, and “the Nixie driver chip only has 10 outputs,” which covers two characters of five segments each. “With most things what I invent it turns out that something like this already exists and it is called a panaplex display.”

Masking took three tries. Laser-cut spray paint came off with solvent and took the platinum with it, and laser-cut Kapton tape did the same thing when it was peeled. An etched aluminum stencil, pulled tight to the glass with magnets, finally worked, and the result looked perfect until “after 3 days of not even looking at it,” when the platinum let go. “Platinum actually really does not want to stick to glass,” the maker learned, and a titanium layer underneath fixed it: “the difference here is night and day.”

Steal this

Coating glass with platinum or another noble metal? Put down a titanium layer first. The maker’s bare platinum looked perfect for three days and then let go, and titanium underneath is what fixed it. Kind of like primer before you paint a door that slams a lot.
When you need a pattern on a fresh thin film, hold a stencil against the glass instead of laying down paint or tape you’ll have to remove later. Paint dissolved off with solvent and peeled tape both took the platinum with them, and an etched aluminum sheet held on with magnets didn’t. This is like painting around masking tape on a wall that came out of the paint can yesterday: the tape takes the fresh paint with it.
Running anything with high voltage inside a vacuum chamber? Keep the logic board outside and bring only wires through the feedthrough. The maker’s Arduino board went in “and hoped for the best,” and the video’s note afterward reads “Never put sensitive electronics and high voltage in the same vacuum chamber.” Kind of like keeping the phone out of the bathtub, even when the charger cord reaches.
If your only gauge is uncalibrated, log against it anyway and treat the numbers as relative. This is like a bathroom scale that reads five pounds heavy: wrong about your weight, still right about whether it went up.

The prototype lit inside the chamber, an Arduino and a Nixie driver multiplexing the segments, though only two characters at a time through the feedthrough’s wires. The standalone build went further: a CNC-machined back plate, a valve to close off after pumping down, Loctite EA9492 epoxy picked for low outgassing, and a 3 mm laminated glass window the maker made, “full of bubbles, but at least it’s really strong.” It held vacuum for 24 hours.

It didn’t work. “I would not consider this as a working display,” the maker says. Taking it apart showed resistance between segments that should have had none, the best guess being that “the kapton adhesive become conductive when it was exposed to the plasma,” and the conductive glass was damaged too. “That’s probably why Nixie tubes aren’t made out of epoxy and kapton tape.” A later version with a built-in valve “just kind of caught on fire. This has to be a first in a vacuum chamber.”

Why it matters

The factory version used the same idea. Panaplex was a Burroughs trademark, and the more common Panaplex II “has a conductive coating on the glass for the anode,” Brent Hilpert wrote on the ClassicCmp cctech list, the same move as the maker’s ITO glass. He also recalled Panaplex and similar gas-discharge digits in calculators “up to the mid-70s” and “extensively in arcade/pinball games.”

The factory had help the garage doesn’t. A collector’s spec sheet for a 16-digit Burroughs Panaplex II lists a neon fill “doped with Krypton-85,” a radioactive gas, running on a 200 V supply. The maker’s version is welding argon, a salvaged gauge, and epoxy whose datasheet “says nothing about its permability.”

The data is out there now. The breakdown curve and power numbers fill a gap the maker says exists for flat argon cells, with the warning about the gauge attached. The GitLab repo holds the KiCad layout of the segment artwork (the schematic file is empty), with a README that says “Free to use. Do whatever you want with it.”

The coating turned out to be the solvable part. Titanium fixed the adhesion and a stencil fixed the mask, but the sealed display died on the parts a garage can’t easily replace: glasswork, an adhesive that stays an insulator under plasma, and a seal that doesn’t leak gas over time. Hobby sputtering rigs built from scraps make the cathodes reachable for one person, and that points to more dead display technologies coming back as one-off builds, with the envelope still the hard part.

The maker says there’s “a really long list” of what to do differently and wants to turn the design into a working digital clock, but is “really tired of just cleaning glass and burning actual money” on platinum. The video ends asking anyone with experience for ideas, and the resistance between segments, which disappeared once the display was destroyed, still has no explanation.

More Retro