All of these efforts were preparatory for the greatest of all Chinese medieval clocks, the 'Cosmic Engine' of Su Sung, built in the year 1092. Su Sung's clock is known in considerable detail due to the miraculous preservation, over nine centuries, of his book Nezv Design for a Mechanized Armillary Sphere and Celestial Globe. This work describes the design and construction of the great clock in full detail. Some drawings in the book have recently been discovered to be those of Chang Ssu-Hsun in 976, to whose earlier clock Su Sung's must have borne a closer resemblance than had been thought until now. Modern working models of the clock have been constructed, and in Plate 80 may be seen one at the Science Museum in London. These models are, of course, based on the full descriptions and drawings in Su Sung's book.
  Su Sung's clock was actually an astronomical clock tower more than thirty feet high, like the previous one of Chang. But on top of Su Sung's tower was additionally a huge bronze power-driven astronomical instrument called an armillary sphere (see page 36), with which one could observe the positions of the stars. A celestial globe inside the tower turned in synchronization with this sphere above, so that the two could constantly be compared. We are told that the observations made on the demonstrational globe inside and by the observa- tional sphere above 'agreed like the two halves of a tally'.
  On the front of the tower was a pagoda structure of five storeys, each having a door through which mannikins and jacks appeared ringing bells and gongs and holding tablets to indicate the hours and other special times of the day and night. All of these time-indicators were operated by the same giant clock machinery which simultaneously turned the sphere and the globe.
  This machinery consisted, as usual, of a huge vertical water-wheel with scoops at the end of each blade, into which water dripped from a water clock. Every time the wheel turned one notch upon the filling of a scoop, there was a ratchet-pin which came down to prevent the wheel recoiling backwards. As for the forward motion of the wheel, it went forward one scoop every quarter of an hour. Needham describes the machine as follows:

The wheel was checked by an escapement consisting of a sort of weigh-bridge which prevented the fall of a scoop until full, and a trip-lever and parallel linkage system which arrested the forward motion of the wheel at a further point and allowed it to settle back and bring the next scoop into position on the weigh-bridge. One must imagine this giant structure going off at full- every quarter of an hour with a great sound of creaking and splashing, clanging and ringing; it must have been very impressive, and we know that it was actually built and made to work for many years before being carried away into exile.

More...