HV plasma Christmas decorations

[Rich Feldman]

During the lost weeks I posted a preliminary report about luminous octahedrons made from T5 fluorescent lamps. The tubes were overhead at work until replaced with LED's in 2018. I am still learning things about off-label uses.

Among the Platonic solids, octahedron is unique in having an even number of edges at each vertex. That's electrically and mechanically convenient.
Here's the first "small" octahedron.

.
Hubs are made from black poly tubing that, by "serendipity of fit", pushes nicely onto the T5 lamp ends. It held up for a few days outdoors in rain and cold. A retired string of miniature lights provided wire designed for connecting lamps in series with ultra low cost connectors.

.
Hanging from a tree in my front yard, next to a fake neon (white LED) star, the octahedron was too bright with 30 mA from a NST.
Dialed down to about 10 mA, with Variac upstream of the NST, it looked good the next couple of nights.

.
I expected that low-current operation in frosty cold weather would be hard on the cathodes. In normal service they are kept hot by 170 mA discharge current.
Didn't expect a failure after less than 24 operating hours, but at the beginning of fourth evening session the ornament didn't light.

Cranking up the variac, HV arcs appeared in unintended places. Back indoors, individual lamp tests found one (#4) that wouldn't strike. It also has open filaments on both ends. Many other lamps lit with normal-ish voltage (typically 0.15 or 0.16 kV indicated, range 0.12 to 0.22) but had a pink color at one end, and spot checking found some open filaments. More details to follow.

For Round 2, I made more hubs and put together a "large" octahedron with twelve nominal 28 watt lamps, each 1163 mm long. Here displayed in a cubicle farm. This variac set to 90% of 0 to 140 V, so around 30 mA at 60 Hz. Indicated voltage was 2.84 kV, and didn't change in six hours of continuous operation on 12/20. To be continued on January 3.

[Rich Feldman]

Getting close to 24 hours of total run time on the long-tube unit, indoors with about 30 mA of current at 60 Hz.
All tubes look uniformly bright, except for darkened ends which aren't obviously getting worse by the day.
I guess the voltage will go up as lamps approach end-of-life; today it's held steady at 2.82 kV after warmup.

.
I think the small octahedron, which failed after about 24 hours at 10 mA in frosty weather, suffered from cold cathode syndrome.
With plenty of voltage to keep the discharge going, the emission mix and then tungsten coils were eaten away by sputtering.
Maybe the dark deposits eventually trapped most of the available mercury, which revealed the pink glow of argon.

Note that in a normal lamp discharge, there are about 1000 atoms of argon (at ~ 3 torr)
for every atom of mercury ( ~ 0.003 torr vapor pressure, if the tube's coldest spot is at 30°C ).
Tube volume can hold 12 micrograms of Hg in vapor form at 40°C, so the initial fill (1.5 to a few milligrams) is to cover Hg losses until something else fails.
.
The octahedron casts an interesting shadow of itself on the ceiling.
Contrast would be much higher if the main lights (now LED's in tubes) were off.

.
There's an X shape, which is shadows of upper meridians cast by meridian lamps in the same planes.
And there's a tic-tac-toe shape (#) which is shadows of upper triangles, cast by the equatorial lamps in same planes.

Guess that if the whole thing were spinning, fast enough to appear as a solid body because of persistence of vision, we would see fixed dark lines where far-side tubes are eclipsed by near-side tubes.

[Rich Feldman]

The public display overhead has ended a day early, after somebody bumped a corner by accident.
When octahedron rotated through a large angle, an external arc started at the top vertex.
I bet it involved thin wire woven into the edges of red ribbon that suspended the apparatus from a grounded S-hook.
One side of the polyhedron went dark, and the ribbon caught fire.

A couple of us were right there and Immediately pulled the plug. Supported the octahedron by hand & climbed up to put out the fire and/or unhook the ribbon.
Got it down at about the same time black plastic hub strips melted and two lamps separated at the top.
Clearly some design details need to be improved for safe semi-attended operation.

Damage is limited to one hub, its jumper wires, and the ribbon. I'm planning to continue the lamp-life experiment this season, in a safer place with regard to risk of fires and electric shocks. Shape might fit through doorways now, without having to take apart another hub. Final voltage reading before the accident was 2.98 kV, of which about 25% would have appeared between each top jumper and ground.

p.s. If the NST had had Secondary Ground Fault Protection, the octahedron would have simply gone dark until reset, with no consequential damage.
www.allanson.com/wp-content/uploads/Pro ... ection.pdf