Terry:
This is an attempt to automate a series of soil tests, one of which is so-called "tri-axial test". In the simplest of forms, it is the way to determine the permeabillity of a soil sample to water: how much water wood seep through under certain conditions in a certain time. Or something like that (sorry, a civil engineer or a geologist would be able to provide a better description).
The sample is a cylindrical piece about 2" in diameter and 6" tall. Two pieces of lexane (called "platens") of the same diameter are placed on each end of the sample. The whole "sandwich" (top platen - sample-bottom platen is inserted into a thin rubber sleeve and placed on the bottom of the glass vessel (roughly 3" dia. X 10" tall).
Each platen has a bored cavity with a fitting leading to it. Before the vessel is sealed and filled with water, plastic tubes are connected to each of two fittings, wich, in turn, lead outside of the vessel.
Water surrounding the sample seals the rubber sleeve around it. After some soaking (which may take up to 24 hours!), water is being supplied into both the top and the bottom platen - under pressure. For example, the water supply into the top platen is set at 73 psi, the bottom platen - 70 psi and water in the vessel - 80 psi. This difference in pressure (73 - 70) creates flow - very slow one.
Each of the two water sources (one for the top and one for the bottom) is fitted with a burette - a glass tube about 20 mm in diameter with level marks on it (0.5 mm resolution). After each source has been filled with water and it sets at some level in each burette, there is air under pressure (73 psi in the first one and 70 psi in the second one) above each water column. As the test progresses, the levels change: decrease in the top-platen burette and increase in the bottom-platen one. Initiallly, the amount of level decrease in the first burette is greater than the amount of increase in the second one; eventually (after some more long hours) these amounts will become equal. That completes the test. At present, everything is done by hand: turning all the valves on and off, adjusting the pressures, reading the levels every 15 minutes or so - all done by an operator.
Although 70-80 psi are typical numbers for the test, the main air supply provides 200 psi. This means that any differential pressure transducer used to measure the level, must be able to withstand this much pressure without being destroyed - in case one of its ports gets disconnected or some regulator goes wild. One the other hand, it must be sensitive enough to detect 0.5 mm change in the level of water.
It would be a sweet small PLC project if not for the need to measure those levels with somewhat conflicting requirements (high resolutuion/accuracy vs. ability to withstand high ambient pressures). The ideas I've seen so far (Honeywell, Rosemont) look nice, although a bit bulky - they are packaged for power plant or industrial use and have a lot of extra stuff (comm. protocols etc.)
Pressure is somewhat like voltage - it makes sense only in relation to some other point, not by itself. Any ordinary pressure gauge is, in essence, "differential" - the second point being the Earth atmosphere.