192 MEASUREMENT OF PRESSURES 



the cable by a piezoelectric or other high impedance gauge. This as- 

 sumption requires a perfect dielectric material for cable insulation, in 

 which the capacitance is independent of frequency and the dielectric 

 loss is negligible. All solid dielectrics exhibit some departures from 

 this ideal state, although the effects are sufficiently small as to be negli- 

 gible for polystyrene and polyethylene cables used in gauge circuits. 

 Cables with these dielectrics are desirable for small charge work where 

 they can be handled fairly gently. 



Unfortunately, the standard varieties of low-loss cable often do not 

 survive the rough treatment encountered in work from vessels with 

 large charges, and also exhibit spurious effects known as "cable signal," 

 as discussed in part C. The cause of failure is breakdown of the dielec- 

 tric by handling and salt water absorption. To avoid these difficulties 

 more durable rubber dielectric cables have been used almost exclusively 

 in the past for field work. Cables of this type all show appreciable 

 variations of capacitance with frequency, some being very poor indeed. 

 Although the low-loss rubber varieties are much better, it is still not 

 safe to assume that over any appreciable range of frequencies or time 

 intervals they can be represented as a simple capacitance and it is there- 

 fore necessary to determine the errors involved and devise means for 

 their correction. 



Over the frequency range fifty cps. to fifty kc./sec, the capacity of 

 a good low-loss rubber dielectric cable varies by ten per cent or more, 

 and the response to a gauge signal impressed on the cable will vary by 

 about the same amovmt over this range of frequency or the correspond- 

 ing range of times. If the cable response characteristics are determined 

 as a function of frequency, by bridge measurements or otherwise, it is 

 formally possible to correct any observed transient response for the 

 effect by use of the superposition theorem and operational methods. 

 As a practical matter, this procedure would be prohibitively difficult 

 because of the necessity for a series of response measurements, repeated 

 for every use of the cable, and tedious calculations. An alternative 

 solution is introduction of a compensating network, which has been 

 done, but with only limited success because no simple network can be 

 successfully used over the wide range of frequencies necessary. A 

 simpler procedure, which is adequate for reasonably good dielectrics, is 

 to calibrate the transient characteristics directly and use this calibra- 

 tion record for correction of the observed records, the procedure for 

 correction being quite simple if the effects are not too large. 



A suitable and very convenient procedure for calibration of the 

 cable characteristics is afforded by the Q-step procedure described in 

 detail in section 5.7, by which in effect a known quantity of charge is 

 impressed on the circuit from a step voltage source (see section 5.9(D)) 



