Build · 2023-04-05
Water-tube valve: the self-regulating air charge
Can a long, narrow tube with its end under water act as a one-way valve that holds a thermoacoustic engine at its ideal internal pressure without manual bleeding?
Yes, and it ran harder than any other vent: the water column behaves as a resonant second piston, not just a valve. ~18 V on the prototype, high enough amplitude to tear diaphragms. Retired because it drinks the water on cool-down.
- Kind
- Build
- Status
- Built
- Date
- 2023-04-05
- Tube
- 6 ft 6 in × ⅛ in
- Tube volume
- 0.817 in³
- Output (prototype)
- ~18 V
- Cooling
- None (prototype)
Video
What it is
Explainer and test of the water-tube valve. Before it, the engine needed a manual valve released mid-heat. The tube adds a tiny buffer volume and the water acts as a one-way seal, so the engine regulates itself.
Results
- →Bubbles leave the tube as the engine heats; once enough air has left, the diaphragm self-starts and amplitude grows as the vacuum deepens.
- →A plain one-way valve doesn't work: it vents all the air and the diaphragm never pulls back. The water column provides the needed give.
- →With no cooling the prototype stops once the whole body heats up; the cooled single-piece design followed.
- →On cool-down the vacuum is strong enough to drink the whole cup of water, which is what drove the switch to the long vent pipe with a small duckbill at its end.
Lab notes
Why a vacuum
The diaphragm only moves usefully if the engine sits slightly below atmospheric pressure: heated air pushes it out, cooled air pulls it back. Venting the expansion air on warm-up gets there; the question is how to vent without letting air back in.
How the tube works
The tube extends the resonator by 6 ft 6 in while adding under one cubic inch of volume. Hot air travels into it, cools, and is pulled back each cycle; at the far end, water lets bubbles out but not back in. The water level visibly oscillates about an inch with the engine.
What the water column really is
The 0.82 in³ tube volume is close to the ~1 in³ the diaphragm displaces per half-stroke, so the water column takes up most of the stroke. It is not a valve so much as a second piston: a slug of water bouncing on the engine's gas spring. A liquid column on a gas spring is a Helmholtz oscillator with its own resonance, and when that resonance sits near the engine's the two drive each other and the amplitude climbs. That is why this configuration ran harder than any flap valve, and hard enough to tear several diaphragms.
Mean pressure
Any vent that passes air out but not in settles the engine about one pressure amplitude below atmospheric: air leaves whenever a peak exceeds atmospheric and never returns. The water column reaches the same mean pressure as a duckbill but with a soft, resonant end condition instead of a dead flap.
Why it drowns the engine
The problem is cooling, not running. The engine holds roughly 35–40 in³ of air; cooling from running temperature to ambient shrinks that by about 10 in³. The tube holds 0.8 in³, so the water has nowhere to go but into the engine, and no tube length fixes a 10× volume mismatch.
Open question
- Trap bottle: a sealed 15–20 in³ bottle between the engine and the water tube, both lines entering at the top, like a lab vacuum trap. Water climbing on cool-down collects in the bottle and bubbles back out on the next heat-up; the engine never sees liquid.
- If the trap works, the water column's resonance can be tuned on purpose: tube length and bore set its frequency, and matching it to the engine is the lever that made this run so hard.
- Past any vent: a sealed engine charged to 2–3 atm. Acoustic power scales with mean pressure, and every vented design is capped below one atmosphere.
Notes transcribed from the build video.
Related
Questions about this one? Ask in the Discord. It's the fastest way to reach me.