THE ELECTROLYTIC VARIABLE
RESISTANCE TEST LOAD/SWITCH EP-RR 4 FOR THE CANBERRA HOMOPOLAR GENERATOR
R. A. MARSHALL
First Published: May, 1964
Re-issued: April, 1967
Department of Engineering Physics
Research School of Physical Sciences
THE AUSTRALIAN NATIONAL UNIVERSITY
HANCOCK ra’ A.C.T., Australia.
TJ163.A87 EP-RR4. f T J16 3* 1924126 . A8 7 EP-RR4 A.N.U. LIBRARY i This book was published by ANU Press between 1965–1991. This republication is part of the digitisation project being carried out by Scholarly Information Services/Library and ANU Press. This project aims to make past scholarly works published by The Australian National University available to a global audience under its open-access policy. THE ELECTROLYTIC VARIABLE RESISTANCE
TEST LOAD/SWITCH FOR THE
CANBERRA HOMOPOLAR GENERATOR
by
R. A. MARSHALL
First Published, May, 1964 Re-issued, April, 1967
Publication EP-RR 4
Department of Engineering Physics Research School of Physical Sciences THE AUSTRALIAN NATIONAL UNIVERSITY C anberra, A.C.T. A ustralia - 6 FEB 1968 CONTENTS
page
1. Introduction 1 2. Description of the Test Load 1 3. Use of Load as a Current Control Resistor and Switch 7 4. Characteristics of the Test Load 7 5. Operating Characteristic of Quarter Homopolar Generator 15 6. Theory for Load Conductance - Time Program 16 7. Test Load Conductance - Time Program 17 8. Cam Co-ordinates and Classical Pulse Calculations 18 9. The F ast Pulse 21 10. The Emergency Brake Pulse 23 11. The Bus Bar Test Cam 24 12. Cam Manufacture 24 13. References 27
ii 1. Introduction
When the Department's homopolar generator was nearing completion in its Mark I form the design and construction of a device for testing it was begun. The re quirements for this were as follows:
a. To be able to carry the 1. 6 million amps produced by the homopolar generator.
b. To be able to make and break the circuit with a voltage of 240 volts in it. This is the maximum voltage that is generated in one rotor disc.
c. To be able to have its resistance continuously variable in a pre programmed manner so that pulses can be shaped as desired.
d. To be able to absorb all the energy from one rotor at full speed. This is 290 megajoules at 900 revolutions per minute of the rotor.
Details of the nature of the tests can be found in references 1 and 2.
2. Description of the Test Load
The test load consists of a series of sheet steel electrodes 1/8 in. thick interleaved so that they have alternate polarity, Figure 1, which are hung from a gantry type frame as shown in Figures 2 and 3.
r -r 7v\\\\ , 'v\\ \\i t-J IO? 1 * ! I i _rr\ v, ■_ 'A y r r *r Figure 1. Detail of Interleaved Electrode Plates Description of the Test Load 2 The top ends of the tongues of these electrodes are dimpled and clamped to the alumi nium conductors which connect to the homopolar generator. Below these is an 800 gallon capacity tank which is supported in a tubular stem rising in its centre on a hydraulic ram, which enables the tank to be raised and lowered. This causes the caustic soda electrolyte to rise to a greater or lesser height between the electrode plates, giving the desired electrical resistance in the generator’s discharge circuit. This resistance can be made as low as 4 micro-ohms. Vj O O Ct t w ö y? ö o • o / c o n tro l val'/l Figure 2. Front Elevation of Test Load - Semi-schematic Description of the Test Load 3 Figure 3. End Elevation of Test Load - Semi-schematic Description of the Test Load 4 The ram is actuated by high pressure oil fed from an accumulator via a valve which is controlled by a control cam and the electrolyte tank's vertical position, (see Figures 4 and 5) The accumulator is used to enable the desired speed of move ment of the tank to be obtained without using an excessively large pump. This speed is such that the plates can be immersed in about a second. HYDRAUL 1C ACCUMULATOR &AZ BOTTLE AUb RAM tf / [f 9 Figure 4. Hydraulic Circuit for Test Load - Schematic When the tank is being lowered on the servo system after a pulse, it is possible for the valve to shut right off under some circumstances. This tends to occur at the end of the downstroke. To prevent this from causing large destructive pressures in the system, a pressure relief valve is included in the oil line to the ram. When this hydraulic servo system was first operated, it behaved in an unstable manner. This was cured by decreasing the gain of the servo by modifying the valve spool to make the characteristics of the valve linear. Initially these had pronounced S shape with mest of the controlling being done over a small portion of the centre of its travel. Other points of interest concerning the design of test load are as follows: Caustic electrolyte was chosen rather than acid, even though its resistivity is higher, because this allowed steel, Description of the Test Load 5 which is both stronger and cheaper, to be used for the electrode plates. There are two main forces acting on the plates when current is being carried. Because the current-carrying tabs of the plates are separated rather than interleaved, this gives rise to a force of several tons at full current pulling the two groups of plates together. This is resisted by supports between the top clamps and by continuing some of the plate spacer tie rods across between the two stacks. The other force is one tending to pull the positive and negative plates into contact with each other if their spacing is not perfect; a case of unstable equilibrium. When the load was first built an attempt was made to support each polarity of plate by means of two angled sets of struts that were kept out of the electrolyte. However these failed during a test of the generator1 allowing the plates to touch and put a short circuit on the machine during which the current rose to 1. 9 million amps. Following this, the bracing was done with horizon tal insulated struts which get wet during a pulse. These can be seen in Figure 6 and proved satisfactory. CAM INPUT Feeo back chain CONTROL VALUE LEVEN Anchor point CONT/pOL VALUE Figure 5. Test Load Tank Height Feedback Control System - Diagrammatic Description of the Test Load 6 Figure 6. Photograph of Test Load, Showing Electrode Plates, Insulated Struts and Electrolyte Tank. The electrode plates are arranged in the two stacks to allow room for the tank ram to rise up between them. The devices shown around the plate tabs in Figure 2 are pick-up coils to enable rate of change of current, and hence current, to be mea sured during pulses (this is one of several ways employed to measure the current). Between the load and the generator, the bus bars which are interleaved to minimise inductance and forces, pass tnrough a commoning unit which, as the name implies, connects all bars of each polarity. The purpose of this is to prevent any unsymmetri- cal short circuit which may occur in the load from getting back to the generator in its unsymmetrical form. This could cause serious forces on the rotor bearings and con ductors. Nearly all of the energy delivered to the test load from the generator appears as heat in the electrolyte and to remove this the caustic soda solution is pumped out of the tank through a heat exchanger into a hold-up tank. It is then pumped back into the load tank, all this being done between pulses. So that successive pulses may be as similar as possible, it is necessary to ensure when the electrolyte is pumped back that the surface is at the correct level. This is done by pumping an excess quantity back, which subsequently flows out over a weir slit. The weir collector can be seen on the tank in Figure 6. Characteristics of the Test Load o o o os 11 0*4S l-o 14 0-96 l-s 36 (•44- 20 4S l' 91 2-S CO 2-39 ZO 1Z 237 325 IS III iso 84'2 IIS ZlS 9 o-G 1 3-59 400 2)7-5 3-52 475 104-7 4-06 4 s o 112'4 4-30 47S" 121-0 4-53 S'00 l}OS 4-76 SIS 1411 4*99 SSO 155-8 S'-2 2 S b 9 u c 6 188 S'CC 7 260 G-55 8 332 7*40 9 4o4 8-27 lO 476 9 /3 n S48 10-00 (Z 620 10-17 U 651 lb 74 14 764 |2-6( Fig. 7: Test load plate shape and relationship between depth of immersion (L), area IS (3-48 836 of immersion (A), tank movement (H). IG 9o$ I4-35 L, H in inches. A in square inches. n 930 IS22 IS 1 052 1609 *9 112 4 (6-36 20 1(56 n-*3 21 /26$ 18-70 22 1140 19-56 23 1412 20' 43 24 14*4- 21-30 L A H 7 3. Use of Load as a Current Control Resistor and Switch Although the load was originally built to test the homopolar generator, its main use now is as a current control resistor when the generator is being used to power such experiments as the million amp arc, and the high field magnets. In this latter case, it is also used as a safety switch connected into the magnet's safety inter lock system, as described in reference 3. 4. Characteristics of the Test Load 4.1 Plate Shape The plates of the load have the shape shown in Figure 7. They have constant total width w for the first 3-1/4 in. ( w = 24 in.). Over the next 2.44 in. w increases according to the formula 24 w = ------— / 1 - 0.365H which applies in the range H = 0 to H = 2. 44 in., at which limits w = 24 in. and 72 in. respectively (H = 0 at L = 3.25). Over the remaining height w is a constant 72 in. The above formula for w is chosen so that if constant rate of area immersion is re quired, then this is obtained by constant deceleration of the electrolyte tank to 1/3 its initial velocity over this range of H. For constant velocity of immersion, it also pro vides for constant rate of change of area immersed. (The plate cutouts were provided to make it easier for the load to give an initial rate of rise of current from the genera tor of 1/10 of that for the main part of the pulse. This was originally thought to be desirable to prevent high current densities from occurring on the homopolar rotor surface during a pulse. Later calculation showed that this fear was unfounded.) 4. 2 Plate Area (one side of one only) 2 The area of the two tabs below the transition region = 78 in. ______2 The area in the two transition region = 131. 5 ( 1 - / 1 - 0. 365h) in. = 88 in2 for H = 2. 44 in. i. e ., the whole region 2 The area of the top parallel portion = 1318 in. Thus giving the area of one side of 1484 in* 2 one plate Characteristics of the Test Load 8 4.3 Plate Spacing The space between plates = 3/4 in. (Centre to centre spacing of plates 7/8, plates 1/8 in. thick) 4.4 Electrolyte Area The total current from the homopolar generator is carried througn 72 electrolyte sheets, each sheet being the area of one plate (the immersed portion thereof) and the thickness being the same as the plate spacing. 4.5 Electrolyte Resistivity Sodium carbonate has too high a resistivity, so the choice is between sodium or potassium hydroxides. The resistivities of these as a function of concentra tion is shown in Figure 8. These curves are given at 18°C, which is a standard tem perature at which electrolyte resistivities are expressed. KOH is 50% more expen sive than NaOH (which is about £90 per ton - in 1963) but is not much better, so NaOH is used. Point A on the curve represents the concentration beyond which it is not worth going for NaOH, i. e ., 10%, giving a resistivity of 3. 2 ohm cm. at 18°C. Figure 9 gives the conductivity at concentrations of up to 10% for NaOH. 4.6 Temperature Effect on Electrolyte Resistivity The general formula for this change is K^ = Klg (1 + b(T - 18) ), where b varies between 0.020 and 0.025 (see Chem. Eng. Handbook, Perry, p 1783). For NaOH, b = 0. 020 at 5% concentration. This means that the resistivity falls by a factor of 2.6 as the temperature rises from 18°C to 100 C. 4.7 Load Conductance K = 17,600 (L - 3.38) / p mho (4.7.1) (provided that the plate cutouts are fully immersed), where L is the plate’s depth of immersion (inches) and p is the electrolyte resistivity (Ohm cm). Another way to express this is K = 244 A/ p (4.7.2) Thus the resistance of the load with plates fully immersed and with 10% NaOH at 18°C is 8. 8 x 10“^ ohm. (With the electrolyte raised to 90°C, load resistance = 3. 6 x 10 ohm.) Characteristics of the Test Load 9 4. 8 Resistance and Back Voltages in Series with the Load —6 The effective resistance of the bus bars and load tee bars is 8 x 10 ohm. The^ effective resistance of the 1/8 in. thick steel lead-in tabs to the load plates is 1 x 10 ohm and the effective resistance of the plates themselves is about 2 x 10“6 ohm. Adding a further 4 x 10- 6 ohm for contact drops at clamp joints gives a total of 15 x 10 ohm. (This is probably too high, but gives a reasonable total as far as the calculations are concerned.) There is the contact volt drop of about 1/2 volt per set of homopolar generator brushes. One volt should therefore be subtracted for this when designing test load cams. 3-0 -- ConcbkttRA, t iw i t ' i (&RA/US o r MATVR/AL IN JOO CrKAMS OF SOLUTION ) Figure 8. R esistivity V ersus Concentration for NaOH, KOH, Na9CO , at 18°C. ^ O Characteristics of the Test Load 10 There is a further back voltage of about 2 volts due to the electrolysed films of oxygen and hydrogen on the plate. These behave like a simple cell acting in the backward direction. (Observations made during the bus bar testing program showed that the rate of build up of this polarization was rapid, full back EMF being built up by the passage of the order of 5, 000 coulombs.) 4. 9 Voltage limitation on Electrolyte Some experiments were conducted with NaOH solutions which showed that for any particular concentration (or conductivity), as voltage was raised across electrodes in it, a point was reached at which the current carried began to fall away sharply, and the mode of behaviour changed. We have called this the "critical voltage". At some voltage below this, slight arcing occurred in the electrolyte, but current re mained .steady. At voltages above this point, arcing became stronger and more noisy and the current carried began to fluctuate. This latter may have been due to steam 0-2 - To osrA tA j c o v w c m ir r a t Aa/K OTHF/f TCKPetATUKe T use rue ronowtvo- — ' J ! ' 5 '° Ai/lOO OtAMi j or sot unew) Figure 9. Conductivity Versus Concentration for NaOH at 18°C. Characteristics of the Test Load 11 layers forming on the electrodes, because sweeping the electrodes through the elec trolyte tended to raise the critical voltage. Electrode spacing or shape had little or no effect on it. Figure 10 shows this critical voltage plotted against conductivity for NaOH solution. Also shown in Figure 10 are measured values taken from the load during tests with the homopolar generator. These have been taken from reference 5. The two points shown as circles would be expected to straddle the critical voltage line. The pair of points shown as crosses straddle another critical line which divides "spark ing" from "spark free" operation. At the point shown as a double cross, no sparking was seen so the "sparking limit" curve passes somewhere above this. * • SMALL SCALE EXPERIMENTS O 1 INFORMATION FROM FULL SCALE ' I OPERATION OF TEST LOAD J ( from REF. 5) - 3 0 0 xA \x 8 2 *\ utu d y-SPARKING LIMIT 2 \ - 3 0 0 £ X cui A o "critical voltage" -100 T _ ! I I 001 0-05 0-10 CONDUCTIVITY ( ohm' c m ') Figure 10. "Critical Voltage" and Sparking Limit Versus Conductivity for NaOH Solution Characteristics of the Test Load 12 4.10 Electrolyte Temperature Rise in the Load Because the resistivity of an electrolyte falls rapidly with increasing temperature, if a (constant) voltage is applied across it, then the current density car ried will rise. This increases heating and accelerates the process. To calculate the effect, consider the electrolvte between two plates. Then the rate of temperature rise is given by: dT J2 o -1 ~ 77 = - - ■■ ■ _----- C sec - 2. where J is the current density (amp cm ) S is the specific heat of th^ electrolyte (cal. gram - 1 . w is its density (gram cm ) In the case of the test load, this reduces to: dT o -1 C sec dt 4.2 C2 S w where V is the voltage between plates (volt) k is the conductivity of the electrolyte (ohm cm ) C is the distance between plates (cm) S and w as above, w may be taken as unity C is 0. 75 x 2. 54 cm. k is a function of temperature, so the equation becomes: V2 dT k18 (1+:^ > dt 15.2 S Noting that for the load, V is a function of time, this reduces to: -S g l- = _Ü18_ v2 dt 32 + T 760 S Integrating: 32 + T, k Vf , p f 18 f 2 11 ^ 32 + T ^ ~~ 760 S V dt (4.10.1) o V o where subscripts o, f refer to initial and final values. Characteristics of the Test Load 13 Specific heat has been left in this equation because it can be considered as having a value greater than one because of the thermal capacity of the plates. The pulse tests of 1962 indicated that S may be about equal to 2, giving a reduction of Tf by a factor of 5 compared with taking S = 1. 0. Integral V2 dt will be greatest for the surface of the electrolyte because it is in contact with plates for the whole time of the pulse where as lower levels of electrolyte will have volts across it for only part of the time. Thus heating will be greatest at the surface. There are three mitigating effects as far as electrolyte at and near the surface is concerned however. The first is that for the initial part of the pulse while the tank is rising, cold steel is continuously entering the liquid. This will have a considerable cooling effect. The second is that this action will also cause mixing of the electrolyte, which will also help. Thirdly, small scale experiments indicate that at constant voltage and at current densities of about 30 to 40 amp in , a fall in resistivity due to rising temperature does not produce a ri se in c u r rent. In fact the reverse seems to occur, probably due to the insulating effect of the gas films and/or concentration depletion at the electrodes. There is therefore a good chance that the calculated accelerating condition does not occur in practice. This is a good thing, because if surface and sub-surface boiling were to occur, considerable amounts of caustic soda solution could be thrown out of the load. (This did in fact happen on one occasion as noted in reference 5.) This voltage limitation on the electrolyte and the surface heating effect have two important consequences in deciding how fast the plates should be im mersed and what electrolyte concentration should be used. In the interest of keeping the critical voltage as high as possible, electrolyte conductivity should be as low as possible. This also causes least surface temperature rise. For any desired rate of current increase then the rate of increase of load conductance K is fixed. But K is proportional to plate area immersed times electrolyte conductivity. Thus a given dK/dt may be obtained with higher rate of immersion and lower conductivity, the limit to this being set by either the maximum rate of immersion of the plates or the minimum value of K required. 4.11 Gas Evolution -2 Experiments also showed that at 20 amp in. , the gas films formed on the electrolytes do not make resistive films, and also that the rise in level due to gasing does not cause a rise in current due to the resulting increase in electrode area. This level rise which amounts to about 5 in. in the load during a full pulse, is there fore ignored in designing cams. 4.12 Electrolyte Tank - Mechanical Behaviour The tank plus electrolyte weighs 8, 200 lbs. and is supported on a 4 in. diameter ram. The rest pressure in the ram with tank poised is 650 p. s. i. The maximum rate that the electrolyte tank can be raised at is 30 in. per sec. The actual rate of immersion of the plates is greater than this because of the effect of the Characteristics of the Test Load 14 displacement of the plates. This gives approximately a 15% effective increase in immersion rate. The maximum acceleration that it can be given upwards is 1/2 g. There is a blow off valve in the hydraulic circuit that does not permit higher values. The reason for this valve as mentioned before is to prevent the tank from banging when it comes back to the bottom position after a pulse cycle. At this stage, if the control valve shuts for any reason, very high pressures can be built up. The maximum acceleration that it is wise to give it in a downward direction is again 1/2 g. An acceleration greater than one g would pull the tank away from the electrolyte. Accelerations approaching g in this direction may cause the free surface to take on peculiar shapes and cause the resistance of the load to behave unpredictably and unevenly from place to place in the load. This is the reason for the above limit of 1/2 g. The tank falls back down under gravity. It is not "powered" down. 4.13 Control Valve Characteristics The servo control valve which admits oil to, or releases oil from the tank ram, will pass oil at a rate depending on the amount it is open and the differ ential pressure across it. This oil flow rate can be converted into an equivalent tank velocity and this is shown in Figure 11. The curves in this figure are based on the experimentally obtained formula: v = 5.42 x / A P where v is ram velocity (in. sec *) x is valve opening (in.) A P is differential pressure across valve (p. s. i.) This equation is not valid for x greater than 0. 26 in. 4.14 Servo Ratios A schematic arrangement of the feed back system is shown in Figure 5. It can be seen from this that for the valve to open an amount x, the cam input required is 2x extra above that required for the tank height and zero velocity. In the design of a cam, once tank height, velocity and acceleration have been fixed for all times during the cam operation, it is possible to decide what the valve opening must be at all times, using Figures 11 and 5. This will be dis cussed in detail later. Characteristics of the Test Load 15 * 010 OZO Figure 11. Test Load Tank Ram Velocity Versus Servo Control Valve Opening and Differential Pressure Across Valve. 5. Operating Characteristic of Quarter Homopolar Generator The full homopolar generator with the four discs in series will have the following characteristic values: C = 1250 farad Full speed = 900 r.p.m. ( oo = 30 tt sec ) Voltage at full speed = 960 volt Energy at full speed = J u2 x 10 = 1/2 C v =5.76x10 joules P where J is the polar moment of inertia of one rotor r Operating Characteristic of Quarter Homopolar Generator 16 This corresponds to an excitation field in the homopolar generator of about 16,500 g a u ss. Using these numbers, the characteristics of the quarter homopolar generator,where the energy is then stored in one rotor rather than in two and where current is taken from one disc rather than from four in series, can be found. They are: C = 10, 000 farad Full speed = 900 r. p.m. Vmax = Voltage at full speed = 240 volt Emax = energy at full speed = 2. 88 x 10 joule 6. Theory for Load Conductance - Time Program Since the homopolar generator may be considered equivalent to a capacitor, its behaviour when connected to a variable resistance R (or conductance K) may be calculated in the usual way, i. e ., V.K From this t _1 V - V / Idt ( 6 . 1) o C o also dV K dt V C ‘ giving t _1 / Kdt ( 6 . 2) C o where K is measured in mho. Using this relationship and knowing the initial voltage V and load conduc tance K as a function of time, then V may be found. Also since I = VK (6.3) then I may be found. Theory for Load Conductance 17 (In practice, it is convenient to revert to the use of R at one stage in the tabulated calculations so that the resistance of bus bars etc., may be allowed for. This is con sidered in detail later.) The current pulse from the homopolar generator as originally intended con sisted of a damped sine wave with maximum current of 1. 6 x 10® amp. occurring at 3/4 second and falling again to zero at 1. 8 seconds. This is referred to as a Classical Pulse (see reference 4). It cannot be obtained from the full machine when tested with a resistive load (as distinct from inductive) because rotor reversal is not possible. The maximum energy pulse that can be taken resistively and have the same form will take less time than 1. 8 seconds. Assuming that the machine voltage at the end of this pulse is 1/9 of Vmax, i. e ., rotor speed is 100 r. p. m ., then the time for this pulse is 1. 1 seconds with the maximum current occurring at 0. 5 second. In the case of the quarter machine however, current is carried in one of the two discs of the rotor only. This means that twice the current must be carried in this one disc to get the same rotor retardation as with the full machine. Thus the maximum energy pulse that can be taken is a pulse length of 2. 2 seconds, the 1. 6 x 10^ amps occurring at 1 second. 7. Test Load Conductance - Time Program A preliminary idea may now be obtained of the conductance time program required to take a classical pulse. This is most conveniently done using a tabular method. See Table 1. Lines 1 and 2 give the current for this pulse at 0. 2 sec. TABLE 1 Conductance vs. Time for Classical Pulse 0 t ( i e c ) o 2 *4 * 6 8 t-0 11 /‘4 1-6 1-8 I (W > 0 •a*/oG ( z 'S 1-6 1-5 /-I -1 *4 o l< it -o84Kio& - i t - i s •31 •ll *27 */9 *11 *04 U t/c 8 21 2.8 3/ 3( 27 19 /< 4 $ !<<*/< o 8 2S S 7 88 U 9 (46 16S /76 180 : 0 V (80 112 151 12 3 9>Z Ct 34 (8 4 o 1) K ( mho) o 4(CSO 8,600 (2,200 17,400 14'COO IS,ZOO R ( micZoohm) C fO 216 ll 6 82 56 *1 18 Test Load Conductance 18 intervals of time. Line 3 gives Idt for these intervals, the arithmetic mean of I for the interval being taken. The value of C used in line 4 is 10, 000. Line 5 gives 1 _ Vq - V from equation 6. (V - V — J Idt) and line 6 gives the generator vol- o C o tage at times t assuming that the initial voltage of the generator is 180 (i. e ., 680 r.p.m. initial speed). This shows the voltage reaching the impossible value of zero at the end of the pulse. It is not serious because the tail end of the pulse is not as important as the beginning and middle. Line 7 gives K by equation 6.3 (I = VK). It is desired to get the plates fully immersed in 1 sec., so that control of K is maintained past peak current. (Control is lost once the plates are fully immersed because then maximum K is obtained.) To get K ^ 30, 000, with plates fully immersed, electrolyte conductivity required is obtained from equation 4. 7.1. 1 K______0. 08 mho P 17,600 x 21.62 Critical volts for this conductivity NaOH are 150 volts. This is close enough to the 180 volts above to provide a good test for brushes and the validity of critical voltage and conductivity curve of Figure 10. 8. Cam Co-ordinates and Classical Pulse Calculations It will be most convenient to describe a specific pulse design and for this purpose, the Classical Pulse has been chosen. The calculation is taken in two parts, the first gives the cam co-ordinates based on the required tank movement; the second gives the current based on the tank movement, initial generator voltage and electrolyte conductivity. This is given in Table 2. Lines 2, 3, 4 are worked out together to give tank height H, velocity V, acceleration a with time t (line 1) at intervals of 0.1 second. In this case, maximum velocity up and down has been used (30 imsec--*-), full immersion of the plates used (H = 21. 3 gives immersion L = 24 in.), and the accelerations at the peak fitted in. O The deceleration (66 in. sec ) is gentle. This has the effect of drawing out the peak current over a longer time than if maximum deceleration is used. The acceleration back down has been taken at near the maximum 1/2 g (1/2 g is 193 in. sec"2). The 0. 04 second dwell at full immersion has been included just to make the calculation of H, V, a tabulate more easily. P (line 5) is the pressure required under the ram to give the required ac celerations, 650 p. s. i. being that required when a = 0. Cam Co-ordinates and Classical Pulse Calculations 19 Oo <2 u o o o Cs o o — ui 0 O 6 g Table Table 2: Calculation of Co-ordinates Cam for a Classical Pulse. __ — o 9 — JD -a — R S P H Y S.S v ^BRAR'1 Cam Co-ordinates and Classical Pulse Calculations 20 P (line 6) is the pressure that is left to be throttled across the control valve. It is the initial accumulator pressure minus P and minus 10 p. s. i. per in. rise of the tank (which is the rate of fall of accumulator pressure as the tank rises). x (line 7) is the valve opening required to give the required throttling of oil into the ram from the accumulator. This is obtained from Figure 11 using the values of A P and v (lines 6 and 3). A r (line 9) is the basic cam radius to produce the required H (r = 0. 5H, see again Figure 3. 14). Lines 8 and 9 added give r + A r which is the cam co-ordi nate required at each t. Converting these co-ordinates into a cam is described in section 9. A (line 11) is obtained from the Table in Figure 7 using H (line 2). The next step is to decide on a value of electrolyte resistivity. Referring to Table 1 it can be seen that a rate of increase of K of about 24, 000 mho sec- * ( = ) is desired. Line 11 shows an increase in immersed plate area of 2, 500 in.2 sec— -1 ( = dA/dt). dA/dt Using equation 4.7.2 244 — 244 This gives P required as 25 K dK/dt ohm cm. Tabular calculation (not shown) using this value showed that too high a voltage would be required to obtain a 1.6 million amp pulse. Trial and error gives a value of p =18 as all right and this is used in Table 2. Using equation 4.7.2 enables the load resistance R to be tabulated (line 12). To this is added bus bar resistance R giving total resistance R in line 14. B Line 15 K/C is the reciprocal of R divided by the capacity of the 1/4 homo- polar generator ( = 10, 000) and multiplied by 10® to make the units mho farad-1 Line 16 is 0.1 times the value of K/C in the middle of each time interval to give K St/C for each interval. Line 17 is the summation of these. From equation 6.2, it is seen that line 17 is the natural log of the voltage ratio. The antilog is given in line 18. This enables the voltage to be obtained - line 19. VQ has been taken as 180, this giving the desired 1. 6 million amp pulse. Current can now be found line 20, by dividing (line 19 minus 3), by line 14. The minus 3 is for the back voltages which act to oppose the applied voltage. Lines 10 and 20, i. e ., cam co-ordinates and current versus time, are the main results of this tabulation. However a further result of interest is obtained in line 21, namely J V dt for the pulse. This enables an estimate of the electrolyte surface temperature to be obtained by using equation 4.10.1. In this case we get: 32 + T ______f _ _ 21,600 _ 1,58 °ge 32 + T 18 x 760 S S o Cam Co-ordinates and Classical Pulse Calculations 21 Thus for S = 1, 32 + Tf/32 + Tq = 4. 85, giving Tf = 196°C for Tq = 15°C and for S = 2, 32 + Tf/32 + Tq = 2. 2, giving Tf = 71°C for Tq = 15°C. This in practice turned out to be all right. Table 3 is the calculation of a pulse using this cam, but starting from full speed and 240 volts and using an electrolyte resistivity of 32 ohm cm. (The right re sistivity can be hit on at second try by extrapolating from the first try and the current that would have been obtained from Table 2 if VQ had been 240 volts, i. e ., current peak = 2.14 meg amp, because for a given resistivity, current peak is proportional to initial volts.) . 2 In this case f V dt = 48, 290. Thus for S = 1, 32 + T/32 + T =7.3, giving Tr = 310°C for T = 15°C t o t o and for S = 2, 32 + T/32 + T =2.7, giving T = 95°C for T = 15°C. t o t o In practice this also turned out to be all right. 9. The F ast Pulse g This was designed to enable a pulse with the full peak current of 1. 6 x 10 amp to be taken from the minimum generator speed, so that if anything went wrong, there would be a minimum of energy available to be dissipated. This was the situation at the time of the test load short during the initial quarter machine tests, using NaK on 7th June, 1962 (reference 1). A one million amp peak was being aimed at on that occasion. (The electrolyte conductivity at 18°C was 4.2 ohm cm., initial speed was 255 r.p.m. , the fast pulse cam being used.) The velocities and accelerations provided by this cam are: a. A high initial acceleration is used to get the plates into the electrolyte as soon as possible, thus providing minimum lag time between the pulse initiation sig nal and the pulse beginning. The velocity accelerated to is 21. 6 in. sec“-*-. This takes 0.15 sec during which time the tank moves 1. 62 in. b. The constant velocity of 21. 6 in. sec-'*' lasts 0.15 sec during which time the plate tongues become fully immersed, i. e ., tank travel during this time is 3 -1 /4 in. c. During the period (0. 09 sec and 2.4 in.) during which the transition area of the plates is being immersed, the tank is accelerated at 115 in. sec-2 to a velocity of 32 in. sec“-*-. d. This velocity is maintained for 6-3/4 in., the time for this constant velocity being 0.16 sec. a Coodnts n Clsia Pus Calculations ulse P lassical C and o-ordinates C Cam Table 3: Calculation of Cam Co-ordinates, starting from full speed and 240 volts and using an electrolyte resistivity of 32 ohm.cm. — n—H~ — o « **> r~ 5 o ‘T»O ly CO * . t —i ^ ^ v * b . : ...... , 4 — , —ff V v *>• *v> ^ cO * Co • . y ( > t v r _ ^ jj o _ V—H to 4- Table 5: Effect of Using Classical Pulse Cam with input volts and conducti 22 the same as for the fast pulse. The Fast Pulse 23 -2 e. The tank is then decelerated^ 188 in. sec for 0. 3 sec. This leaves the tank falling back down at 24 in. sec . The total tank rise to the peak height from the point where the plates were just entering is 10. 8 in. f. This velocity downwards of 24 in. sec 1 is held to the transition re gion, a distance of 7.9 in., and a time of 0. 33 sec. g. Over the transition region, the downward velocity is decelerated at 90 in .s e c “^ to 12 in .s e c - *. h. This velocity is held until the tank is lowered clear of the plates. This cam was designed to be used with electrolyte conductivity of 4. 2 ohm cm at 18°C to give a 1.6 meg-amp pulse from V0 = 100, i. e ., from 375 r.p. m. The calculated pulse co-ordinates for this fast pulse are given in Table 4. TABLE 4 F ast Pulse t(sec) 0 .1 . 2 .3 .4 .5 .6 .7 .8 .9 1.0 1.1 1.2 1.3 1.4 I(megamp) 0 .28 .55 1.21 1.60 1.57 1.25 .91 .60 .36 .19 .09 .06 .05 .01 V(volt) 100 98 92 80 64 49 33 25 20 18 16 15 15 15 15 The effect of using the classical pulse cam but with the other variables (initial volts and conductivity) the same as for the fast pulse are shown in Table 5. This gives a pulse almost identical to the fast pulse, except that the slow initial rate of rise of cur rent is missing and also the final volts approach zero. The various pulses are plotted in Figure 12, in which is also shown the ori ginal classical pulse for comparison. 10. The Emergency Brake Pulse This was designed to slow the generator from 500 r. p. m. to the minimum speed possible in 3. 8 seconds and with a maximum current of 1/2 million amps. The current shape for this was rising to 1/2 megamp in 0. 3 sec., this current being then maintained for a further 1. 6 sec., at which time the current decayed exponentially to 0.13 megamp. The tank was then lowered clear at 30 in. sec-* in a further 0. 7 sec., making the pulse length a total of 3. 8 sec. The Emergency Brake Pulse 24 The electrolyte resistivity for this was 25 ohm cm. The final volts were calculated to be about 10, i. e ., a reduction of 13:1. T »C OflC-iAJAL CIA Si/ (4L S . 1 Tune ^(iNDOCT/veload) Figure 12. Comparison of Various Pulse Shapes. 11 . The Bus Bar Test Cam This cam was designed to completely immerse the plates at the full speed of 30 in. sec- , to hold this for 3 sec., then to drop the tank down again at 30 in. sec-^. It was used to switch 10, 000 amps at 1 volt from high current capacity lead acid accum ulators for doing tests on the bus bars. Cam Manufacture To convert cam co-ordinates of cam radius and time to polar co-ordinates, two further basic facts must be known, namely the rotational speed of the cam and the cam base circle, that is, the circle which is considered as radius equals zero. (It also has the property that if a cam were used that had its base circle as radius, the load plates would just be touching the electrolyte, the feed back chain length being ad justed to make this so.) The cam shaft rotates anti-clockwise at the rate of 1 revolu tion in 5. 38 seconds, giving a rotation of 66. 7° per second. The base circle is taken as 10-1/4 in. radius. Cam Manufacture 25 This could have been anywhere from 8 in. to 12 in. It cannot be much smal ler because then for a given rate of rise, the cam slope gets too steep. At 10 in. or so, the maximum slope is a few degrees greater than 45°. It cannot be much bigger because then the cam becomes too large for convenience. Another piece of necessary information is that the cam follower is 1-1/4 in. diameter. The procedure to adopt for making a cam is then as follows: The cam is made from 3/4 in. thick plywood and it is best to lay the cam out directly on this, thus eliminating errors of transposition. The following description applies to classical pulse cam for t = 0. 3 to 0.7 which is taken as an example to illustrate what is in volved. See Figure 13. The centre is chosen and marked on the ply and the base cir cle plus cam follower radius are drawn. Radial lines at angular intervals of 6. 67° are drawn, these representing 0.1 second time intervals. The appropriate value of r + A r is marked out on each radius. When there is a discontinuity in acceleration as at t = 0. 5, 0. 96, 1.0, 1.2, then both values of r + A r are marked. On each of these marks as centre a circle of 1-1/4 in. diameter is drawn, these representing the cam follower. Curves are then drawn which are tangents to these circles. The most con venient way of doing this is to use railway curves, picking their radius to touch three Figure 13. Test Load Cam Layout. Cam Manufacture 26 circles at a time. At the points where acceleration is discontinuous, the cam profile can be discontinuous, giving shapes as at point A in Figure 13. (This at first sight might not seem right but the point to remember is that the oil pressure under the ram is discontinuous at the same time as is the acceleration. The only way to get such oil pressure jumps is to make the control valve jump. This requires sudden changes in cam profile.) The co-ordinates given in Table 2 are for the active part of the cycle only and co-ordinates must be calculated for the remainder. In this case the starting posi tion has been taken with the plates 3 in. from the surface of the electrolyte and an ini tial acceleration of 150 in. sec“^ is used to bring the tank velocity up to the required 30 in. sec“l in that distance. At the end of the pulse, the tank is lowered an extra 2 in., so that the plates come clear of the electrolyte surface, which is raised with respect to the tank, because of the gas bubbles formed in it during the pulse. A de- celeration of 90 in. sec is used over this 5 in. The cams have two brass trigger strips which screw on to their backs which operate a microswitch. This in turn stops the cam motor. One of these is positioned to stop the cam after the final deceleration after the pulse is over. The other stops the cam at the "start” point for the next pulse, namely with the cam follower just at the toe of the initial acceleration region of the cam. Because there is a difference in level of the tank before and after a pulse of 2 in., the tank must be raised this amount between pulses. Thus a convenient rise is put on the cam between these two stop positions. Any convenient accelerations and velocities less than the maxima may be used for this. With these marked out on the plywood, the cam may be drawn and cut out on a band saw and the rim sanded reasonably smooth. The mounting holes and holes for the trigger strips are then drilled. These may be marked off from an existing cam or from the hole template. The triggers are then screwed in place and the cam is ready for use. The last thing to be done to the feed-back system before pulsing is to rotate the cam to the "start" position and to adjust the length of the feed back chain to bring the electrolyte surface to the correct distance below the plates - in this case 3 in. 27 13. R eferences 1. BLAMEY, J.W ., P.O. CARDEN, L.U. HIBBARD, E.K. INALL, R.A. MARSHALL, and M. OLIPHANT: "The Large Homopolar Generator at Canberra, " Nature, 1962, 195, pp 113-114. 2. MARSHALL, R.A.: "Tests with Solid Brushes on the Canberra Homopolar Generator, " Nature, 1964, 204, p 1079. 3. CARDEN, P.O .: "Features of the High Field Magnet Laboratory at The Australian National University, Canberra, " Australian National University, Department of Engineering Physics Pub lication EP-RR 19, January, 1967. 4. MARSHALL, R.A.: "The Design of Brushes for the Canberra Homo- polar Generator, " Australian National University, Department of Engineering Physics Publication EP-RR 3, April, 1967. 5. INALL, E.K.: "Proving Tests on the Canberra Homopolar Genera tor with the Two Rotors Connected in Series, " Australian National University, Department of Engineering Physics Pub lication EP-RR 7, April, 1967. RI Publications by Department of Engineering Physics F irs t No. Author Title Published Re-issued E P -R R 1 Hibbard, L. U. Cementing Rotors for the May, 1959 April, 1967 Canberra Homopolar G en erato r E P -R R 2 Carden, P. O. Limitations of Rate of Rise Sept., 1962 April, 1967 of Pulse Current Imposed by Skin Effect in Rotors E P -R R .3 Marshall, R. A. The Design of Brushes for Jan ., 1964 April, 1967 the Canberra Homopolar G en erato r E P -R R 4 Marshall, R. A. The Electrolytic Variable May, 1964 April, 1967 Resistance Test Load/Switch for the Canberra Homopolar G en erato r E P -R R 5 Inall, E.K. The Mark II Coupling and Oct. , 1964 April, 1967 Rotor Centering Registers for the Canberra Homopo- lar Generator E P -R R 6 Inall, E.K. A Review of the Specifica Nov.,1964 April, 1967 tions and Design of the Mark II Oil Lubricated Thrust and Centering Bearings of the Canberra Homopolar Generator E P -R R 7 Inall, E.K. Proving Tests on the F e b .,1966 April, 1967 Canberra Homopolar Gen erator with the Two Rotors Connected in Series E P -R R 8 B rady, T.W. Notes on Speed Balance M ar. ,1966 April, 1967 Controls on the Canberra Homopolar Generator E P -R R 9 Inall, E.K. Tests on the Canberra May, 1966 April, 1967 Homopolar Generator Arranged to Supply the 5 Megawatt Magnet Publications by Department of Engineering Physics (Cont.) R2 F irs t No. Author Title Published R e-issu ed EP-RR 10 Brady, T.W. A Study of the Performance June, 1966 April, 1967 of the 1000 kW Motor Gen erator Set Supplying the Canberra Homopolar Gen erator Field EP-RR 11 Macleod, I.D.G. Instrumentation and Control Oct., 1966 April, 1967 of the Canberra Homopolar Generator by On-Line Com p u te r EP-RR 12 Carden, P. O. Mechanical Stresses in an Jan., 1967 Infinitely Long Homogeneous Bitter Solenoid with Finite External Field EP-RR 13 Macleod, I.D.G. A Survey of Isolation Ampli- Feb., 1967 fier Circuits EP-RR 14 Inall, E.K. The Mark III Coupling for Feb. , 1967 the Rotors of the Canberra Homopolar Generator EP-RR 15 Bydder, E. L. On the Integration of Mar. ,1967 Liley, B.S. "Boltzmann-Like” Collision Integrals E P -R R 16 Vance, C.F. Simple Thyristor Circuits Mar. ,1967 to Pulse-Fire Ignitrons for Capacitor Discharge EP-RR 17 Bydder, E. L. On the Evaluation of Elastic Sept. ,1967 and Inelastic Collision Fre quencies for Hydrogenic-Like P la sm a s E P -R R 18 Stebbens, A. The Design of Brushes for M ar. ,1964 Sept., 1967 W ard, H. the Homopolar Generator at The Australian National U niversity Copies of this and other Publications (see list inside) of the Department of Engineering Physics may be obtained from: The Australian National University Press, P.O. Box 4, Canberra, A.C.T., 2600. Australia. Price: $ A 1.00 Copyright Note: Reproduction of this publication in whole or in part is not allowed without prior permission. It may however be quoted as a reference.