Differential methods of vacuum leak detection

In the August 1951 issue, Mr. Blears and Leck described their technique for leak detection in which a manometer reading is obtained which depends on the composition of the gas in the continually pumped vessel under test. The sensitivity of the method is limited by the background fluctuation of the manometer readings. If a gas having a different sensitivity in the gauge is substituted for air, in the neighborhood of a leak, this substitution will affect the pressure indication. This allows leak detection by addition or subtraction methods. A somewhat different subtraction method has been used in these laboratories. Our manometer consists of a single ionization gauge protected by a liquid air trap. The only requisite for the gas used in testing for leaks is that it should have a negligible vapor pressure at liquid air temperature; if this condition is fulfilled, the presence of the gas is detected by a lowering of the pressure reading, independent of its nature. We chose carbon dioxide on account of its ready availability, non-inflammability and relative non-toxicity. The four main causes of fluctuations in the manometer reading are as follows: (a) Variations in pumping speed, which can be damped out by placing between the evacuated system and the pump a short length of tubing, of low intrinsic pumping speed compared to that of the pump. This method is also used by Blears and Leck. (b) The rate of gas evolution from the walls of the system is not necessarily constant, but as the indication of the trapped ionization gauge is independent of the evolution of condensable vapors, the fluctuations of the manometer readings correspond only to that of the residual permanent gas pressure which will generally be low enough not to be disturbing. (c) Pressure variation may be caused by the unsteadiness in the rate of flow of larger leaks. (d) Fluctuations in the emission from the gauge cathode will cause small variations in the readings, even when the grid current is stabilized by an electronic device. The circuit shown in the accompanying figure provides an increase in stabilization by use of a null method. Part of the cathode current serves to balance the plate current of the gauge. So long as the pressure remains constant the fraction of the cathode current necessary to keep the circuit balanced is largely independent of total cathode emission. Thus the sensitivity of the meter reading is practically independent of the pressure. We used a galvanometer giving a 1 mm deflection for a current of 10^-9 A. The main differences between the technique of Blears and Leck and ours can be summarized as follows: Their method provides compensation for fluctuations described above (a), (c) and (d), whereas ours provides compensation for items (a), (b) and (c). While, in our opinion, the extent to which item (d) in their method or item (b) in ours is successfully dealt with is debatable, we have consistently obtained the following results: The smallest pressure variation that we can observe is 2 x 10^-8 mm of mercury when the residual pressure is 5 x 10^-5 mm of mercury. The maximum ratio (total leak flow)/(flow of studied leak) at which a leak can still be detected is 2500. In a system of moderate volume (about 1 litre), pumped out at the speed of about 100 c.c./sec, a leak of 0.01 mm³/h is readily detected. Reprint of a paper published in the 'British journal of applied physics', vol.3, Jan 1952, p. 29.