The four-stroke Atkinson differential engine uses a clever arrangement of linkages to move its two pistons in a single cylinder, out of phase with one another. It can use reed valves, like a two-stroke engine; both the intake valve and the exhaust valve are at the same end of the cylinder.
The sequence is as follows. Suppose the intake and exhaust valve are at the right end of the cylinder, and the pistons are close together, with the intake and exhaust valve between them. First the pumping piston, on the left moves to the left, opening the intake valve and sucking air and fuel into the cylinder. Then the working piston, on the right, moves to the left, covering the valves and compressing the fuel-air mix against the now-stationary pumping piston. Then the spark fires, and the working piston moves to the right, providing the power stroke. Finally the working piston passes the exhaust valve, the hot gas escapes through it, and the pumping piston follows it to the right, preparing to pump in the new fuel-air mixture.
On ##electronics cloudevil was talking about piston-powered blowers, displacing multiple liters of air per stroke, suggesting that they could be much quieter than traditional types of blowers, though of course they’ll still produce turbulent airflow. But such a blower will fail to be quiet if it’s using poppet valves or reed valves, since those produce an impulse every time they open and close.
You could imagine using a round, triangular, or teardrop-shaped port or hole in the side of a cylinder as a valve; when the piston passes over it, it opens or closes, but not impulsively. But this has two problems.
First, in the Atkinson engine design, the intake and exhaust valves are at the same end of the cylinder, so if there’s no reed valve or anything, they’ll be open at the same time. This is no way to make an air pump.
Second, if there’s a pressure difference across the valve as it opens, the airflow through it will still start suddenly and with a lot of turbulence, so it will be noisy, though maybe less so than a reed valve.
Both of these problems can be solved by moving the valve openings to opposite ends of the cylinder and redesigning the cycle for pumping without such events.
First, the left cylinder is at the left end of the cylinder, just to the left of the intake port, and the right cylinder is to the right of the intake port. Second, the right cylinder moves to the right, almost to the exhaust port, at the right end of the cylinder; in this way air is sucked into the cylinder from the intake port. Third, both cylinders move in unison to the right, first closing the intake port and then opening the exhaust port. Fourth, the right cylinder stops, while the left cylinder continues moving to the right, expelling the air through the exhaust port. Fifth, both cylinders move in unison, close together, back to the left.
These movements can easily be scripted by cams to minimize the bandwidth of the pistons’ movements, thus eliminating the direct production of sound above, say, four times their movement frequency, which might be 2 Hz. Then only turbulence and surface roughness are left as noise sources.
Why four times? If the pistons moved back and forth in unison a single time, or with a simple difference in phase, they could move in a perfect sinusoid, thus generating no sound from their sheer movement at frequencies higher than their movement frequency. But the movement I described above is not simply sinusoidal, so it would involve some harmonics. 8 Hz is inaudible, but 20 Hz or more might be audible and highly annoying. Of course air turbulence and surface roughness will generate higher-frequency noise no matter what the cylinders’ movement.
You might be able to find a lower-displacement purely-sinusoidal movement pattern with the right characteristics — most crucially that the cylinders be closer together when moving leftwards than when moving rightwards, and the same distance apart when the intake port closes and when the exhaust port opens. If a single sinusoid can’t do the job, you might be able to find something that only uses two or three harmonics rather than four, and thus enable higher operating frequencies while remaining purely infrasonic, maybe up to 5 Hz or 10 Hz. But I’m confident that with four harmonics you can do it.
You can take the same approach and apply it to the problem of making an engine, too, in the sense of a device that converts heat energy into mechanical energy.
The most direct approach is to feed steam or pressurized air into the intake; then the suction stroke becomes the power stroke, but you still have a sharp noise when the pressurized air gets over to the outlet port, and that noise of course represents wasted energy. Similarly, when the small space between pistons moves over the intake port, the pressure in it is the exhaust pressure, which is low. If you revise the cycle somewhat so that the space between pistons is zero when they open the input port, then continues expanding after the input port is closed, you can get all the adiabatic energy in the input working fluid. Alternatively, rather than making the space zero, it could simply be smaller than its size when the exhaust port closed by an amount sufficient to bring its pressure up to the intake pressure.
However, if you want it to be a standard four-stroke internal-combustion engine, you need the following cycle:
This requires some fancy camwork; the followers pressing on the cams during the expansion stroke is what drives the engine forward. Alternatively it may be possible to design a linkage that does all this, which may be beneficial in terms of being easier to adjust.
Adjustment of the cycle might be useful for a variety of reasons:
A second cylinder configured as described above can be used to harvest further energy from the exhaust, in the manner of double-expansion or triple-expansion steam-engines; this will also reduce noise further, as the second cylinder serves as a sort of muffler for the first. (You could also use such an engine as a muffler for a conventional internal-combustion piston engine, surely not a novel idea, since triple-expansion steam-engines go back generations.)
An engine with two or three cylinders cascaded in this way can be fitted with valves to redirect the flows of gases to answer demands that change from moment to moment: now a triple-expansion engine with a single combustion chamber, one cylinder feeding into the next, for greater efficiency, now a three-cylinder engine with all three cylinders burning fresh fuel for greater power.
By virtue of moving the expanding gas chamber down the cylinder as it expands, the loss of heat to the cylinder walls is reduced — while this benefits an internal-combustion engine less than a steam-engine, it may pose difficulties in keeping the hotter parts of those walls lubricated.
As the space between the pistons can be reduced indeed to zero, requiring no accommodation for the opening of valves, the spark-plug is redundant in these classes of engines; it can be designed to ignite purely by adiabatic heating as in Diesel’s engine.