[Edited since yesterday, to move the main temp sensor to the top of the heat exchanger, to make the pump variable speed, and to vastly improve the drainback system]
Consider the following system:
On the roof, a conventional solar thermal collector ... My preference would be for an evacuated tube
collector, to reduce the risk of it freezing in winter, but a flat plate collector would be fine in mild climates.
The collector has two pipes: an intake, and a return.
Inside the house, a heat exchanger between the solar heating fluid, and the domestic hot water.
At the top of the heat exchanger, where the gaseous heating fluid enters from the solar collector, a temperature sensor.
At the bottom of the heat exchanger, inside the heating fluid pipe, a sensor to detect the phase of the heating fluid (gas vs liquid).
After the sensors, a pump to send the fluid back up to the solar collector's intake, and a check valve, to prevent the water from running backwards through the pump, when the pump is turned off.
The pump is controlled by the liquid sensor and the temperature sensor: Whenever liquid is present, and its temperature is not above 190F, the pump will operate.
Furthermore, the pump's speed should be variable: when the temperature is 120F or below, it will operate at maximum speed. Between 120 and 190, the pump will operate at an intermediate speed, with hotter resulting in a slower pump speed, and cooler resulting in a faster pumping speed.
At the very highest point in the system, a check valve that can allow heating fluid out into the atmosphere, as a safeguard against the temperature sensor failing.
Fill the system with distilled water, then attach a vacuum pump to the check valve at the top of the system. Running this pump will first degas the system, then lower the pressure enough so that the water in the solar collector will boil. When a precisely calculated amount of water has been removed, the vacuum pump is removed from the system.
As a result, the solar fluid loop will contain only water and steam, and no other gasses. Because of how much water we removed from the system, it's pressure will be low, so it's boiling point will be low.
During the day, the heat of the sun will cause the water to boil. The resulting steam will flow up to the top of the collector, down the return pipe, into the top of the domestic water heat exchanger, and condense.
The condensate will activate the liquid sensor, activating the pump, which will move the liquid out of the heat exchanger and back to the solar collector.
At night, when the collector is colder than the heat exchanger, the water in the collector won't boil, of course. If there were any liquid distilled water in the heat exchanger, it *would* boil, but since that water is continuously removed to the collector by the pump, we can expect the heat exchanger to be "dry" when night falls.
Furthermore, because there's only enough liquid water in the system to fill the pipe from the pump's check valve, up to the top of the collector, the water in the collector will not overflow and run down to the heat exchanger. As a result, the liquid stays in the collector at night, and no heat is transferred to it from the domestic hot water heat exchanger.
With this design, the amount of pumping energy needed to one's drinking water is very small: Each pound of water pumped up to the solar collector is changed by the sun into a pound of steam. Each pound of steam condensed in the heat exchanger heats the domestic water by 1194 BTUs.
The only potential issue is freezing in winter, since the distilled water is left in the solar collector on the roof each night.
This can be solved by adding an automatic drainback system.
Firstly, at the bottom of the solar collector, there would be a temperature sensor. When the temp drops below 35F, the system would enter drainback mode, until manually reset.
Secondly, there'd be an insulated distilled water storage container. In normal operation (not in drainback mode), this tank would only have low pressure steam in it.
A pipe tee just above the main pump's check valve would lead to a solenoid valve that connects to the bottom of the storage tank. The top of the storage tank would be connected to another tee, in the pipe between the top of the heat exchanger and the top of the solar collector.
Whenever the solenoid valve is open, we have what's effectively a perfectly normal heat pipe, which moves heat from the inside of the storage tank, to the solar collector. Eventually, all of the liquid water ends up in the storage tank, with only steam in the solar collector and the heat exchanger.
As the system chills, the pressure of the steam and water will drop to water's triple point, and the temperature of the steam in the solar collector and the water in the drainback tank will drop to water's triple point. Because steam is a poor heat conductor, the water in the heat exchanger won't get chilled.
The steam remaining in the solar collector can potentially freeze (or rather undergo deposition), but because of how little steam there is (compared with how much liquid there would be without the drainback system) less damage can occur.
To leave drainback mode, we flip a switch.
This switch does two things: First, it closes the solenoid valve just above the pump's check valve.
Second, it opens a second solenoid valve, which is located below a tee between the first solenoid valve and the drainback tank. This second solenoid valve connects to a check valve, which leads to a pipe tee, which is located between the heat exchanger and the pump.
Assuming bottoms of the drainback tank and the heat exchanger are at approximately the same heights, or if the drainback tank is a bit above the heat exchanger, gravity will move the water from the tank to the bottom of the heat exchanger, where it will trigger the liquid sensor, which will turn on the pump, and move the water into the solar collector.