March i8, 1920] 



The 



Gy rostatic 

 By S. G. Brown. 



^^V HAV E directed attention to several faults that have 



I^^K to be rectified if the compass is to be of use on a 



^^Ipp, and I shall now discuss the last, but by no 



" means the least, of the errors that may arise if the 



instrument is not properly designed. This error was 



not known when the gyro-compass was first brought 



out, and it proved a most difficult fault to correct, 



while its elimination has had more to do with the 



design of the later forms of gyro-compass than any 



t)ther factor. 



If a gyro-wheel is precessed towards, and kept 

 pointing to, the north by an ordinary pendulum 

 weight, it will work well on board ship provided that 

 the ship is steaming on a fairly smooth sea ; but if 

 the direction of the ship points anywhere in the 

 quadrants — that is, north-west or north-east, south- 

 west or south-east — and the ship rolls, the wheel will 

 rry to set itself so as to bring the rim of the spinning 

 wheel in line with the roll; and in a long-continued 

 and heavy roll the compass mav show an error of 

 20° or more — a most serious fault, and one that 

 would render the instrument quite useless in a heav\' 

 a. This error is called the " quadrantal error." 

 lie extent of the error depends upon the violence of 

 riie ship's rolling and the direction of the axis of the 

 wheel. If the ship points direct north, south, east, 

 "r west, the error is nothing, but it would be a maxi- 

 mum in anv of the directions before mentioned. 



I think Anschutz was the first to point out the 

 error and suggest a cure. This I gather from one 

 ot his publications in the year 191 1, in which, speak- 

 ing of the tendency of the compass to wander when 

 on board ship, he says: — "Theoretically, the influence 

 of rhythmic turning movements on a gyroscopic ap- 

 paratus must disappear completely if not onlv the 

 real, but a.lso the apparent, movements of inertia 

 of the movable system become equal for each plane." 

 If we go back again and study our simple gyro- 

 compass, we see that the movable system is not sym- 

 metrical. In the direction of the axis of the wheel 

 the effect of tilting movement is more or less resisted 

 by the spinning wheel— this may be termed the 

 stabilised direction; while at right angles to this — 

 that is, in the direction of the rim of the wheel — there 

 is no resistance to tilting encountered, and this direc- 

 tion we term the direction of free swing. A simple 

 form of gyro-compass pointing-, therefore, in a direc- 

 tion, say, north-west on board a rolling ship has a 

 force applied to it tending to turn it so as to bring 

 its direction of free swing into line with the roll. 



.•\nschutz gets rid of the error by multiplying the 

 number of his gyro-wheels and by making the 

 instrument as symmetrical as possible. In England 

 the quadrantal error was first discovered and studied, 

 I believe, by the Admiralty Compass Department. In 

 the year iqi4 the Sperry Co. claimed to have effected 

 a cure for the error by attaching the pendulous weight, 

 not directly to the gyro casing, but through a pin 

 arranged to move in a slot in the casing. In order 

 that the axis of suspension of the oendulum may 

 remain vertical when the compass oscillates with the 

 rolling of the ship, a small auxiliarv gvro was em- 

 ployed to stabilise the pin connection between the 

 pendulum and the g-vro casing. 



We therefore see in these applications of .\nschutz 

 and Sporry two General ideas. In the first case the 

 idea is to make evervthing symmetrical, like a ball, 

 so that there is no stabilised or free swing direction 



' r>isconrse delivered at the Royal Institution on Friday. Faniinrv -^o. 

 Continued from p. . 



NATURE 



n 



Compass. 

 F.R.S.' 



to the wheel, and; therefore, no tendency to turn; 

 while in the second a method is provided to pre- 

 vent the point of application of the pendulum weight 

 from moving and acting as a crank, and, by keeping 

 the pendulum weight always vertical in the north- 

 west direction, to destroy its power of turning the 

 compass. In the Brown compass the quadrantal error 

 is eliminated by making the weight operate completely 

 out of phase with the roll — that is, at 90° displace- 

 ment. 



If a gyro-compass is worked by a weight which 

 tends to precess the wheel in phase with the roll, 

 then there must be a quadrantal error, but there 

 will be no error if it is forced to operate completely 

 out of phase with it. It is also essential, as .Anschutz 

 has remarked, that the real moments of inertia shall 

 be the same in all directions of the movable system 

 of the compass ; that is to say, the moving system 

 should be in dynamic balance, as it is termed. 



If a child's hoop is suspended by a string and 's 

 swinging in one direction, the hoop tends to set itself 

 lengthwise to the direction of the swing. On the 

 other hand, if an exactly similar hoop be placed 

 over, but at right angles to, the first, and suspended 

 as before, then on swinging the hooos there will be 

 no tendency for them to turn, as they are now in 

 dvnamic balance. It is for this reason that the mass 

 distribution of the moving system of the gyro-compass 

 should be in dvnamic balance, and to carry this out 

 adjustable weights are fitted, usually in the direction 

 of the spindle of the wheel, to counteract the weight 

 of the supporting ring of the gyro casing, and thus 

 there is no tendency for the compass to turn, due to 

 this cause, when under the action of rolling. 



The Brown gyro-compass is shown diagrammatically 

 in Fig. 4. .\ is the gyro-wheel in its rasing B. 

 This case is carried on knife-edges M in the vertical 

 ring F, and is thus free to tilt under the action of 

 the rotation of the earth. The vertical rin^ turns in 

 azimuth on a frictionless mounting, consisting of an 

 oil-pump at the bottom and a ball-bearing, m, at 

 the top. XY is the three-phase motor that drives the 

 oil-pump. 



The gvro-wheel is the rotor of a three-phase motor, 

 and current is led into the moving system through 

 the three sets of iron contact rings R and S. These 

 rings do not touch, but the outer set are hollow, and 

 mercurv fills the space between them, so that there 

 is little friction. The vertical ring- is dynamically 

 balanced by two projecting weights D. Q is the 

 pendulous mounting, supported by gymbal rings and 

 bv the outer row of springs to take up shock. 



The gvro-wheel runs at 15,000 revolutions per 

 minute, and thus acts as a powerful blower, giving 

 an air-pressure equal to some 3 in. of water. Fixed 

 to the vertical rinr, but connected through the hollow 

 bearing M to the inside of the case, is the air-jet T.. 

 This iet blows into the two halves of the air-box K, 

 and thence through the- pipes T ; the air-pressure is 

 thus transmitted to the oil in the two sets of bottles 

 C and D. H is another air-jet similarly mounted, 

 and emploved to act upon a pair of contact-making 

 vanes I. 



The contacts T. through the agencv of the con- 

 troller. which is fixed on the switchboard, are to work 

 the repeaters and the step-bv-sten motor V; this 

 motor forces round the follow-up ring N to keep the 

 contact-makinf* vanes I alwavs opposite the air-let, 

 and in doing this all the repeaters on the ship follow 

 suit. U is the compass card fixed to the upper por- 



NO. 2629, VOL. 105] 



