The other night I picked back up playing with insulated gas discharge tubes (IGDT), better known as plasma globes. The basic idea is to pump a sealed chamber filled with a noble gas (or air, in a pinch) down to low pressure (1s to 10s of Torr), and apply electricity at high voltage (a few kV) and frequency (a few kHz) to ionize the gas. Depending on the gas(es), pressure, voltage, and frequency, the results should range from a diffuse eerie glow to a bunch of lightning-like plasma streamers dancing around inside the chamber. The most well-known example are the Eye of the Storm novelty globes and their knockoffs.
Here are some pictures from tonight’s initial playing, using a scavenged fridge compressor as a vacuum pump and a cleaned-out glass pasta sauce jar as the globe. The effect inside the jar was generally unimpressive over a wide range of pressures, but the effect on the clear PVC tubing and the white plastic hose barb I used to attach it was certainly interesting! The hose barb connector glowed brightly and the plastic tubing lit up like a flexible neon sign, with a glow that crawled through the hose several feet toward the compressor as the pressure dropped. Nice poor-man’s neon effect, but I suspect air will outgas and/or leak through the walls of the PVC tubing rapidly (days?), ruining the vacuum and the effect. This calls for further study :-)
I didn’t play too much with additives except to revisit the combination of plasma globe and discrete mini Neon lamps. This looks much cooler with the neon lamps actually inside the evacuated chamber with the glowing plasma (and finally with a decent camera!). Everything so far has used plain old air.
Fridge Compressor as Vacuum Pump
The short story is yes, you can use the compressor from a refrigerator or dehumidifier as a quick & dirty vacuum pump (or light duty air compressor), provided the refrigerant has all been vented/leaked before you get to it (unless you have facilities to safely capture and recycle the refrigerant, which few do). The design of these things seems pretty universal, as well as ingenious: the “in” tube leads to a piston, which is driven by a big electric motor in the center of the compressor via a crankshaft. The output of this valved piston vents directly to the inside of the case, which is sealed and in facted welded shut, with the “out” tube being the only escape. The bottom 1/4 inch or so of this mechanism is submerged in a pool of light oil, which is sucked up by a cone on the bottom of the crankshaft mechanism, sucked through the shaft and flung out the top and against the outer walls of the compressor by centri[whatsit] force – providing simultaneous lubrication and transfer of heat from the pump mechanism to the outside case, which acts as a heatsink. This page has a very cool teardown of one of these compressors, complete with videos of the oil-slinging mechanism in operation.
As a side effect to this design, small amounts of oil are sometimes ejected from the outlet along with the air, necessitating some kind of primitive oil recovery/trap mechanism. For mine I attached a small length of vinyl tubing to the outlet, made a sort of drip loop in it, and attached the other end to a piece of plastic pipet taped vertically to the side of the compressor. The width of the pipet prevents oil from being blown out all the way to the top, while the drip loop in the tubing (intentionally seeded with a couple drops of oil) provides a simple but highly effective visual flowmeter, good for indicating the end of effective pumping (or a leaky system). A loosely fitting cap taped over the pipet helps reject dust.
It’s probably a good idea to eyedropper out occasional oil from this catcher and feed it back into the inlet to keep the piston lubed up. (I had to drain all the oil out of mine to transport it, then refilled it with an equal volume of some SAE 30 motor oil I had on hand. No ill effects thus far, but I’m pretty sure the original oil was a lot lighter.)
But how good is it? I have no actual meter that will deal in these pressures, so it’s hard to say. On this pump a jar of water would–just barely–start to boil at room temperature. According to this calculator, that approximates to roughly 20 Torr (.026 atmospheres). Not bad for free!
Possible sealing approach (haven’t gotten around to trying it yet)
Professional globes are all-glass affairs of a complex shape, with the glass fill tube(s) heated and melted shut while under vacuum to seal the globe. This is not really an option for hobbyists, since it must be connected to a live vacuum during sealing and a typical “globe” has a nonglass base (plastic or Plexiglas) where an equally nonglass fill tube emerges, since glassmelting is not compatible with the typical vinyl tubing connecting it to the live vacuum system in the process. Some just leave a small length of tubing and a closed valve permanently affixed. But good valves aren’t cheap, and cheap valves aren’t good. The approach I’d like to try is to epoxy a hose barb into the base with a very thin piece of hollow metal tubing (needle) epoxied inside of it, making the tube the only gas path into or out of the globe. This is connected via a short piece of vinyl tubing to one of two matching hose barbs on a small lidded jar (e.g. baby food – something with a “magic button” style rubberized lid), such that the needle snakes through the center of the tubing and just reaches the bottom of the jar. A small discardable “bucket” (e.g. soda cap) of 5-minute epoxy would be mixed up and placed inside the jar, the lid screwed on, and the needle sunk into the epoxy. The jar, and thus globe, are then evacuated via the 2nd barb in the babyfood jar lid. When the pump is switched off, the normal small leakage (maybe introduced via valve, if the pump setup is just that good) will allow the vacuum to relax slightly, forcing liquid epoxy up the metal needle, where it is allowed to cure and permanently seal the globe. The excess epoxy-filled metal needle and hose barb can then be cut off. Extending this setup to multiple flush/fill hoses on the globe would be easy enough, other than trying to hand-align these multiple needles at once.