Assume you follow the procedure per rajkm: your CO2 tank is upright, the CO2 in the tube to the solenoid valve is gas. Now you lay the CO2 tank on its side. Nothing changes yet: the CO2 tubing is still full of CO2 gas at room air temperature. All of this system is under the aquarium, with the CO2 tube going up to the top of the aquarium to the diffuser in the tank. So the high point is the tubing at the top of the aquarium. Now you open the solenoid valve, this allows the pressurized CO2 in the tank to enter the CO2 tubing, but only the liquid CO2 because the gaseous CO2 is in a "bubble", a long thin one, along the length of the tank, not at the valve on the tank. As the liquid CO2 enters the tubing it almost instantly changes to dry ice, the solid form of CO2. This process cools the regulator to a very low temperature. The dry ice slowly changes to gaseous CO2, raising the pressure and temperature in the CO2 tubing and increasing the bubble rate as it does that. When the pressure finally stabilizes and the bubble rate is what you had set the needle valve to, the pressure in the CO2 tubing is below the regulator pressure setting, so it opens again and repeats the process. I suspect the regulator would have frozen up, then perhaps will warm up again and repeat the process again. I doubt that this would ever work satisfactorily, but I have never tried it. CO2 is a very tricky substance that likes to be either solid or gas, and it absorbs a lot of heat from whatever it touches to power its transition from liquid to solid to gas. It cannot exist as a liquid except under high pressure - around 500-700 psi. When liquid CO2 changes to dry ice, solid CO2, it becomes very cold, possibly causing the plastic CO2 tubing to lose its strength and to burst.