A Magnetic Trap for Evaporative Cooling of Rb Atoms Benjamin Deissler Schifferstadt, Germany Vordiplom, Universität Kaiserslautern, 2001 A Thesis presented to the Graduate Faculty of the University of Virginia in Candidacy for the Degree of Master of Science Department of Physics University of Virginia August 2003 i A man will be imprisoned in a room with a door that’s unlocked and opens inwards; as long as it does not occur to him to pull rather than push. Ludwig Wittgenstein, Culture and Value ii Acknowledgements I realize that I am indebted to a number of people, and I would like to use this opportunity to mention and thank them. First of all, I would like to thank the German Academic Exchange Service (DAAD) for its generous financial support, without which my year here at the University of Virginia would not have been possible. I would also like to thank my high-school physics teacher, Mr. Zschernitz, for showing me the beauty of physics, as well as my professors at the University of Kaiserslautern, Prof. Hotop and Prof. Fleischhauer, for keeping my fascination alive and kindling my interest in AMO physics. A big nod of appreciation also to my teachers here at the University of Virginia – Prof. Fowler, Prof. Fishbane and Prof. Poon. I cannot begin to thank my advisor, Prof. Sackett, for all that he has done for me. His support and helpfulness stand unmatched, I can only hope to have the pleasure of working with more such people in my career. Thanks to Prof. Jones for the helpful comments and for being on my committee. Jessica and Ofir also deserve appreciation for their help in the lab, and the same goes to Ken, Jeramy, Jessica, Patrick and Patipan, who were here in the summer. Were I to list all of the people I wish to thank in the Ballroom Dance Club at UVa, this section would probably be longer than my thesis. So here’s a general thank-you to all of you for keeping me sane and for the wonderful times we’ve had together – the last year would not have been as fun without you! My girlfriend Kristen deserves my gratitude for her interest, understanding, and patience in hearing me complain when my project wasn’t working (again) and for believing in me that it would work. Finally, my parents have always been there for me, supported me in all possible ways and helped me in ways I cannot express in these few lines. Thank you. Contents 1 Introduction 1.1 Magnetic Traps . . . . . . . . . . . . . . 1.1.1 Quadrupole Trap . . . . . . . . . 1.1.2 Ioffe-Pritchard Trap . . . . . . . 1.2 Evaporative Cooling . . . . . . . . . . . 1.2.1 A Model for Evaporative Cooling 1.2.2 Cooling Sequence . . . . . . . . . 1.3 Goals . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 2 4 5 5 7 8 . . . . . 10 10 12 13 14 15 3 Computer Simulation 3.1 Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Effects of Varistor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 19 21 22 4 Technical Details 4.1 Making the Coils . . . 4.2 Subcircuits . . . . . . 4.2.1 Debouncer . . 4.2.2 LEDs . . . . . 4.3 Connection Diagrams 4.3.1 Front Panel . . 4.3.2 Board 1 . . . . 4.3.3 Board 2 . . . . . . . . . . . . 24 24 27 27 27 28 28 29 29 5 Performance 5.1 Measurements with “Dummy” Coil . . . . . . . . . . . . . . . . . . . . . . . 5.2 Measurements with Quadrupole Coils . . . . . . . . . . . . . . . . . . . . . 5.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 34 35 39 2 Description of Circuit 2.1 Main Circuit . . . . 2.1.1 Switching Off 2.1.2 Switching On 2.1.3 MOT2 Field 2.2 Control Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv A MATLAB code A.1 sc.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.2 V.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 40 41 B Part Specifications B.1 Main Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.2 Board 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3 Board 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 42 42 43 List of Figures 1.1 1.2 Coil configuration in quadrupole trap . . . . . . . . . . . . . . . . . . . . . . Coil configuration in Ioffe-Pritchard trap . . . . . . . . . . . . . . . . . . . . 3 4 2.1 2.2 2.3 Main circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Control circuit generating V1 , V 1 and V2 . . . . . . . . . . . . . . . . . . . . Control circuit generating Vcontrol . . . . . . . . . . . . . . . . . . . . . . . . 11 16 17 3.1 3.2 3.3 3.4 3.5 Simulation circuit . . . . . . . . Current vs. time . . . . . . . . Voltage vs. time . . . . . . . . Current vs. time with varistor . Voltage vs. time with varistor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 21 22 23 23 4.1 4.2 4.3 4.4 4.5 4.6 Magnetic field of the coils . . . Switch debouncer . . . . . . . . Subcircuit to drive LEDs . . . Connections on the front panel Connection diagram of board 1 Connection diagram of board 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 27 28 29 30 31 5.1 5.2 5.3 5.4 Turning Turning Turning Turning with “dummy” coil with “dummy” coil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 36 37 38 on off on off the the the the magnetic magnetic magnetic magnetic field field field field v List of Tables 2.1 2.2 Components of main circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . Switching sequence for Vcontrol . . . . . . . . . . . . . . . . . . . . . . . . . 12 17 4.1 4.2 Connections on first board . . . . . . . . . . . . . . . . . . . . . . . . . . . . Connections on second card . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 33 vi Chapter 1 Introduction The experimental achievement of Bose-Einstein Condensation (BEC) was one of the holy grails of physics before successful observation of BEC in the mid 1990’s in rapid succession by three groups at JILA [1], MIT [2], and Rice [3]. We are attempting to produce a BoseEinstein Condensate for use in atom interferometry experiments [4]. For example, since the phase shift in a Sagnac type interferometer is given in general by [5] φrot = 4π ΩA , λv (1.1) where λ is the wavelength of the interfering field, v the velocity, Ω the rotation frequency and A the area enclosed by the interferometer beams, using atoms instead of light gives a factor mc2 h̄ω greater sensitivity dφ dΩ [4]. Using a BEC allows the area enclosed to be increased, giving an additional boost in sensitivity. In our experiment, we use a magneto-optical trap (MOT) to capture and pre-cool 87 Rb atoms. However, the temperature and density that can be reached by optical traps are limited by the effects of spontaneous emission heating, radiation trapping effects and inelastic excited-state collisions [6], [7]. To reach lower temperatures and higher densities, we therefore use evaporative cooling. Making the magnetic trap for evaporative cooling is therefore a crucial part of the process of producing a Bose-Einstein Condensate. 1 1 Introduction 1.1 2 Magnetic Traps Magnetic traps take advantage of the magnetic moments of neutral atoms and the forces exerted on these by inhomogeneous magnetic fields. The energy of an atomic level with angular momentum F~ and magnetic quantum number mF in a magnetic field of strength B along the z-axis is [7] E(mF ) = gµB mF B ≡ µm B , (1.2) where g is the gyromagnetic moment and µB the Bohr magneton. This shows that the potential that an atom sees will be proportional to the strength of the magnetic field. If we use this to trap and cool atoms, the depth of the trap will be on the order of T = µm (Bmax − Bmin ) , kB (1.3) where T 1K for magnetic fields that are conveniently generated1 . The atoms must therefore be pre-cooled to µK temperatures by the MOT in order to be loaded into a magnetic trap for further cooling. From equation (1.2), we see that if gmF > 0, that we have a weak-field seeking state that needs a minimum of B in order to be trapped, while for gmF < 0, we have a strong-field seeking state, being trapped at a maximum of B. It can be shown that no local maxima of B are possible, so only weak-field seeking states can be trapped [8],[9]. Two main field configurations are used for magnetic traps – the quadrupole trap and the Ioffe-Pritchard trap [9]. 1.1.1 Quadrupole Trap The quadrupole trap, originally suggested by Wolfgang Paul [10] and first used in 1985 to trap neutral atoms [11], consists of two identical coaxial coils carrying currents in opposite directions (Anti-Helmholtz coils) (Fig. 1.1). A simple configuration is an axially symmetric trap, which gives a linear potential with x̂ · 1 ~ ~ ~ dB dB dB = ŷ · = −2ẑ · , dx dy dz The depth of our trap is estimated in section 4.1. (1.4) 1 Introduction 3 I I Figure 1.1: Coil configuration in quadrupole trap where the condition ∇ · B = 0 from Maxwell’s equations produces the factor of 2 in the ẑ term. A magnetic field B(x, y, z) = B 0 [xx̂ + y ŷ − 2z ẑ] (1.5) satisfies this condition and provides the desired trap. Such a trap has equal depth in the radial and longitudinal directions when the separation of the coils is 1.25 times their radius [12]. Though the implementation of this kind of trap is extremely simple, it does have one major problem. The field at the center of the trap is zero, so when the atoms reach this point, they can flip their spin in a Majorana transition [13] and thus fall out of the trap. Simply adding a spatially uniform time-independent bias field to this arrangement does not eliminate the zero-point, but simply shifts it, so a more elaborate setup is needed. This can be acheived by adding a time-dependent bias field Bbias (t) = B0 [x̂ cos Ωt + ŷ sin Ωt] , where the rotation frequency Ω is faster than the atomic orbital frequency ωtrap = with M being the mass of the atoms and rD = 2 B0 B0 (1.6) q µm ∇B M rD the radius of the “circle of death”2 . For The “circle of death” is the trajectory of the zero-point of the magnetic field. If trapped atoms reach 1 Introduction 4 a typical setup, ωtrap /2π ≈ 20Hz [6]. We therefore choose Ω/2π ≈ 20kHz. The zero-point will then move continuously, and the atoms will seek the “hole” but not reach it. Adding this additional field changes the field configuration, so that in the time average the potential is now a so-called Time-Orbiting Potential (TOP), where3 U = µm B0 + Bρ00 ρ2 + Bz00 z 2 with ρ2 = x2 + y 2 and Bρ00 = B 02 2B0 = Bz00 4 (1.7) [9]. We will use this trap configuration for our experiment. 1.1.2 Ioffe-Pritchard Trap (b) (a) I - + + - I x z y Figure 1.2: Coil configuration in Ioffe-Pritchard trap. (a) side view (b) front view An alternative coil configuration where the minimum of the magnetic field is not zero is the Ioffe-Pritchard trap. This consists of four current-carrying bars plus two pinch coils (see Fig. 1.2). Near the origin, this gives the magnetic field 0 −xz x 00 0 ~ B B(x, y, z) = B0 −yz 0 + B −y + 2 1 0 z 2 − 12 (x2 + y 2 ) , the zero-point, they can fall out of the trap as explained above. 3 Ref. [9] neglects the constant part of the field B0 . Compare [14], Eq. 21 and [17], Eq. 2. (1.8) 1 Introduction 5 or in cylindrical coordinates −zρ 0 cos 2φ 00 B ~ φ, z) = B0 0 + B 0 ρ − sin 2φ + B(ρ, 0 2 2 z 2 − ρ2 1 0 . (1.9) Here, the first term describes the uniform B-field along the z-axis, the second term is a quadrupole field with a minimum at the origin in the ρ − φ-plane while the last term is a “bottle” magnetic field with a minimum at the origin along z and a maximum at the origin in the ρ − φ-plane. Altogether, for kB T |µm |B, we get a harmonic potential4 U≈ with Bρ00 = 1.2 B 02 B0 − µm 2B0 + Bρ00 ρ2 + B 00 z 2 2 (1.10) B 00 2 . Evaporative Cooling Evaporative cooling, like cooling in a cup of coffee, consists of continuously removing the high-energy tail of the thermal distribution. In doing this, more than the average energy per atom is removed, so the remaining atoms, after rethermalization, are cooled. 1.2.1 A Model for Evaporative Cooling In this section, we use a simple model (described in [6], [15]) to understand how evaporative cooling works. The model assumes that elastic collisions dominate over inelastic ones, that thermalization is much faster than the rate of cooling, and that escaping atoms do not interact with the atoms in the trap. The cooling process is here described as happening in one step, although in reality it is a continuous process. We consider a trapping potential s2 s3 s1 x y z U = 1 + 2 + 3 a1 a2 a3 4 Ref. [9] again neglects the constant part of the field 2B0 . (1.11) 1 Introduction 6 where the ai are some characteristic length and si the power in a given direction i. Then the volume scales as V ∝ T ξ with ξ≡ 1 1 1 + + . s1 s2 s3 (1.12) The model then describes evaporative cooling with only two parameters, ν≡ N0 , N (1.13) which is the fraction of the atoms remaining in the trap after cooling, and γ≡ log(T 0 /T ) log(T 0 /T ) = , log(N 0 /N ) log ν (1.14) so that the temperature changes as T 0 = T ν γ . It can be shown that the average reduced energy ¯ = Ē kB T before truncation of the high-energy tail is given by ¯ = ξ + 3 . 2 (1.15) For each evaporated atom, the energy carried away is 3 1 − ν γ+1 , out = ξ + 2 1−ν (1.16) so for ν ≈ 1, γ= out −1, ξ + 32 (1.17) which is just the excess energy above the average energy. Choosing the speed for evaporation is a trade-off between efficiency and speed. If the atoms are cooled too quickly, then thermalization will not be completed and cooling will not be efficient. On the other hand, if cooling is too slow, then the loss of particles through inelastic collisions becomes important. Finding the speed of evaporation uses the principle of detailed balance to give dN = −N f ( > η)kel = −N ηe−η nσv̄ , dt (1.18) where if we lower the trap depth to a value ηkB T , f ( > η) is the fraction of atoms in the thermal distribution with > η, and kel = nσv is the elastic collision rate 5 . 5 The elastic scattering cross-section is given by σ = 8πa2 with a = 5.4nm for 87 Rb [16]. 1 Introduction 1.2.2 7 Cooling Sequence In our experiment, we hope to achieve Bose-Einstein Condensation with the following steps. 1. Transfer pre-cooled atoms from MOT1 to MOT2. The MOTs both have magnetic G field gradients of B 0 ≈ 10 cm . 2. Laser cool the atoms by turning off the magnetic field and lowering the intensity of the lasers for about 2ms (this is the same cooling procedure as in MOT1). 3. Optical pumping for about 1ms. We want the atoms to be in the correct (weakfield seeking) states for evaporative cooling, these are the |F = 1, mF = −1i and |F = 2, mF = 1, 2i states in our case. We will pump to one of these states. 4. Turn the magnetic trap on at a fraction of full strength, matching the trap depth to the temperature of the atoms. This gives us a potential with its width on the order of the size of the atom cloud6 . 5. Adiabatically increase the field strength, leading to a decrease in the trap width. This causes the atom density as well as the temperature to increase. 6. Evaporative cooling We have two possibilities for the actual evaporative cooling (a) rf-induced evaporation This type of evaporation uses radio frequency radiation to flip the atomic spin, so that the attractive potential becomes a repulsive one [17]. If the rf-radiation is of frequency ωrf , then a resonance will occur at |g|µB B = h̄ωrf . In a trapping potential U = mF gµB (B(r) − B(0)), only atoms with E > h̄|mF | (ωrf − ω0 ) will be removed, where ω0 is the frequency that induces spin-flips at the bottom of the potential. However, at the origin, the magnetic field is zero, so the atoms 6 We can estimate at what fraction of its maximum value the field is turned on when we know the geometry of the coils. This is done in section 4.1. 1 Introduction 8 can flip their spin randomly, causing them to leave the trap (Majorana transition [13]) 7 . (b) time-dependent bias field We can apply a rotating magnetic field so that the zero-point of the magnetic field rotates [9]. This creates a “circle of death” since if the atoms reach the zero-point, they can flip their spin and fall out of the trap. If the zero-point is not too far outside of the cloud of atoms, the hottest atoms can leave by getting to the B = 0 spot. This setup was explained in more detail in section 1.1.1. We are thinking about using the rf-induced evaporation for a while, then switching to the time-dependent bias field. We think that the density at the origin is not so high at the beginning of the cooling as to allow many atoms to leak out. 7. Turn the magnetic trap off, transferring the atoms to a smaller magnetic trap with a weak potential. The density will decrease, which is what we desire in order to ultimately do the planned interferometry experiments. We need to do this very carefully in order to make sure that the atoms remain in a condensate. 1.3 Goals In designing the magnetic trap and the circuit that goes along with it, the main problem arises when switching the trap on and off. We want to switch extremely quickly so that the atomic cloud does not expand too far or fall under the influence of gravity. If the magnetic field does not turn on or off fast enough, these influences will disturb the cloud and our experiment. When turning the field on, we want the atoms to move much less than the size of the cloud during the switching time ts . We can estimate the temperature of the atoms before loading into the magnetic trap to be approximately 50µK, their velocity is thus about 10 cm s 7 For details on rf-induced evaporation on Rb atoms, see Ref. [17]. 1 Introduction 9 and the size of the atom cloud 1mm. So we want d = vts 1mm = L ts 0.1cm = 10ms . 10cm/s At a time of 500µs, the atoms will have moved d L (1.19) ≈ 5% of the cloud size. We want to preserve the velocity information of the atoms, e.g. for imaging, when turning the field off. Non-adiabatic processes preserve this information, so we want the field to turn off fast compared to the period of trap oscillations. For a frequency of about 20Hz, this means that the switching must take less than 50ms. A problem arises from extremely fast switching times – a quickly changing magnetic field gives a large voltage according to the law of induction V = −L dI , dt (1.20) where the magnetic field B is proportional to the current I. According to Lenz’s rule, this voltage works to keep the magnetic field from decreasing when switching off and from increasing when turning on. The large expected voltage can also damage components in our circuit. We thus aim to design a circuit controlling our magnetic trap with the following properties: • fast switching times less than 1ms • ability to switch between field configurations for magnetic trap, MOT, and no field • ability to control the strength of the magnetic trap • an emergency shut off if the temperature of certain components gets too high It is beneficial to generate the magnetic field needed for the magneto-optical trap (MOT) with the same pair of coils. This has the advantage that the MOT and the magnetic trap are already aligned as we need them to be. Chapter 2 Description of Circuit The main task in achieving our goals is to design a fitting circuit. We purchased a 750A/20V power supply and then had to build a switch that accomplishes the goals we have described above. This was done, and the result will be explained in two parts: the “main circuit” and the “control circuit”, which provides inputs to the IGBT switches of the main circuit and to the power supply. 2.1 Main Circuit Fig. 2.1 shows an overview of the main circuit. Table 2.1 gives the values of the components. For power and voltage ratings, see Appendix B. IGBT1 works as the main switch to send the current from S1 through the coils, it is controlled by the gate voltage V1 . We want the current source S1 to be on when no current is flowing through the coils (or rather, have it shut off slowly so that the shunt resistors R1 do not overheat), so we need an alternate path for the current to flow when IGBT1 is not conducting. This is implemented by applying a conjugate voltage V 1 to IGBT2, with a shunt resistance R1 equal to the resistance RL of the coils in front of the switch. The rest of the main circuit can itself be subdivided into three different parts, each having a different task. 10 S1 D1 V1 Figure 2.1: Main circuit IGBT2 2x R1 6x V1 S3 Varistor IGBT1 3x L + V2 R8 R6 C1 R2 R7 R5 T1 Switch off C3 C4 R3 IGBT3 MOT 2 D2 C2 5x R4 S4 S2 T2 Switch on + 2 Description of Circuit 11 2 Description of Circuit 12 Power Supplies S1 20V / 750A S2 500V / 25mA S3 12V battery S4 20V / 40A Resistors R1 16.7mΩ R2 8Ω R3 10kΩ R4 20kΩ R5 500Ω R6 1kΩ R7 100kΩ R8 100kΩ Inductance L 83.3µH, 17.1mΩ Capacitors C1 10µF C2 250µF C3 1µF C4 0.33µF Table 2.1: Components of main circuit 2.1.1 Switching Off The main problem in switching the magnetic field off is that the inductance tries to keep the field up according to Lenz’s rule. With V = −L dI dt (2.1) 2 Description of Circuit 13 and the fact that the magnetic field B is proportional to the current through the coils I, we see that a quickly changing field will cause a large voltage. Our goal is to limit the voltage to below the voltage rating of the IGBTs, 600V. We achieve this with two devices: • A varistor, which works like a bi-directional Zener diode in that it lets a current flow when the voltage reaches a breakdown voltage characteristic of the varistor – in our case, this is 400V. • The C1 - R2 combination, which damps once the varistor stops conducting. There will be oscillations in the voltage after switching off, and we would like the resistor R2 to dissipate the energy in these. The capacitor C1 works as a DC block – we do not want to dissipate power when we are sending current through the coils. In addition, the capacitor C3 increases the time for the anticipated voltage spike to reach its maximum after switching off. 2.1.2 Switching On Lenz’s rule also hinders quickly switching on the magnetic field. The induced current will try to prevent the magnetic field from building up. To counteract this, we want to apply a large voltage when the switch is closed. The source S2 is used to charge the capacitor C2 . We can estimate the charge as Q = C · U = 250µF × 500V = 0.125C. We can now estimate the maximum current that flows during the discharge. The voltage will oscillate as V = V0 cos ωt with ω = √1 LC ≈ 6.9 × 103 rad s . Then with I= dQ dV =C = CV0 ω sin ωt dt dt it follows that the maximum current will flow at a time t = (2.2) π 2ω ≈ 225µs. The maximum current is then r Imax = V0 C L s = 500V · 250µF ≈ 860A . 83.3µH (2.3) 2 Description of Circuit 14 When the switch (IGBT3) is closed, a high voltage is applied, causing diode D1 to be reverse biased, thus not allowing the high current of source S1 through. Since it is again related by V = −L dI dt , a high voltage allows a fast buildup of current to flow through the coils. As the strength of the current rises, the voltage will drop, and at some point the diode will open up, allowing the high current to get through and flow through the coils. We need IGBT3 to switch shortly after IGBTs 1 and 2 switch, otherwise the capacitors will discharge through the shunt resistor R1 , which is not desired. Our switch will be IGBT3, which needs a constant difference between gate and emitter voltages. This is not given in our case because the voltage is changing when the capacitors discharge. To fix this problem, we use the transistor T1 with its gate controlled by a signal from the optocoupler. Since its collector is at 12V, the gate of IGBT3 will be 12V higher than the emitter when transistor T1 is switched on. The Zener diodes were placed between base and emitter to prevent voltage spikes above 15V to cause damage to IGBT3. 2.1.3 MOT2 Field We want to be able to run a current backwards through the coils to generate the field necessary for the magneto-optical trap (MOT). If the transistor T2 , working as a switch, is closed, then the current generated by power source S4 just flows off to ground. When T2 is open and the 750A current from source S1 is not flowing through the coils, then the 40A current from S4 can flow through the coils and generate the desired magnetic field. In the switching process, the voltage is negative for part of the time. In this case, the transistor works as a diode, since the emitter of the transistor is at a lower voltage than the collector, leaving the voltage above the transistor to be a diode drop (about 0.5V) above ground. Were the voltage to drop below ground, the current would flow through the diode/transistor, an undesired result. 2 Description of Circuit 2.2 15 Control Circuit We want our circuit to have three different configurations. “On” means the current is flowing through the coil, “Off” means the current source S1 is off, and “Ready” means that the current source is on, the current flows through the shunt resistor. Starting with the current source off (“Off”), we first turn the source on, letting the current flow through the shunt resistor (“Ready”). Next, we let the current flow through the coils (“On”), turning the current source off some time later (“Off*”). Finally, we can turn “Ready” off, not changing anything (“Off”). These transitions need to be realized in our control circuit. “Switch On” has the capacitor C2 discharge through the coils, the high current from S1 following thereafter. After 10ms, we want to start recharging C2 . V1 −→ 15V V 1 −→ 0V V2 −→ 15V, 10ms delay then V2 −→ 0V “Switch Off” redirects the high current through the shunt resistor R1 and turns off the power supply by turning off the control voltage. V1 −→ 0V V 1 −→ 15V Vcontrol −→ 0V Finally, “Get Ready” lets the power supply warm up while the main switch is still off, so no current flows through the coils. If V1 = 0V, the current flows through the shunt resistor R1 . Vcontrol −→ Vin The control circuit consists of two parts. The first part uses the digital input on/off to generate the outputs V1 , V 1 and V2 . The second part uses digital inputs on/off and ready and analog input Vin to generate the analog output Vcontrol , which controls the 2 Description of Circuit 16 15V 15V 15V 2k IC7a ON/OFF 2.5V 20 V1 15V 15V 15V IC7b 2k 20 V1 2.5V 15V 15V IC8 IC7c IC9 2.5V 3.3k IC10 V2 Figure 2.2: Control circuit generating V1 , V 1 and V2 2 Description of Circuit 17 IC2 Vcontrol Vin IC1 READY ON/OFF J CP K Q _ Q IC3 IC4a IC5a IC6 IC4b IC5b Figure 2.3: Control circuit generating Vcontrol power supply S1 . Vin is a voltage 0 . . . 10V which corresponds to a current 0 . . . 750A put out by the current source. In the second part, the switching sequence at the flip-flop will follow Table 2.2. J (ready) K (on/off) Q XOR (clk) ON 1 1 1 0 OFF* 1 0 0 1 OFF 0 0 0 0 READY 1 0 1 1 ON 1 1 1 0 Table 2.2: Switching sequence for Vcontrol A deviation from the given switching sequence makes it possible for Q to be in the opposite state of what we want. We therefore installed a LED to show the state of Q. If we are in the wrong state, we need to switch on/off twice to get into the desired state. 2 Description of Circuit 18 We put a delay in before the J and K inputs using a one-shot, because we want to be sure that the J and K signals arrive at the flip-flop after the clk input. Both are determined only by the ready and on/off inputs, the clk being the exclusive OR of these two. For technical details on the circuit, see chapter 4. Chapter 3 Computer Simulation In order to find the theoretical behavior of the circuit, we did a computer simulation with MATLAB. We were especially interested in the behavior when switching off the magnetic field since we expected the highest voltages to occur here and we wanted to make sure the chosen values for C1 , C3 and R2 were optimal. 3.1 Differential Equations We need to consider only the part of the circuit that is relevant to the process of switching off. The effective circuit therefore looks like Fig. 3.1. L I1 RL VC R2 C1 Ceff I2 Figure 3.1: Simulation circuit 19 3 Computer Simulation 20 We do need to remember that the IGBTs as well as the Varistor have some intrinsic capacitance. They are given by Cvar ≈ 16nF and CIGBT ≈ 160nF for each IGBT. The effective capacitance Cef f is then given by Cef f = Cvar + 3 · CIGBT + C3 . Looking at the voltages in various parts of the circuit gives us a system of differential equations: V̇C V̇C VC I1 + R2 I˙1 C1 I2 = Cef f = V0 − L I˙1 + I˙2 − RL (I1 + I2 ) . = (3.1) (3.2) (3.3) Inserting (3.2) into (3.3) gives us: VC = −LI˙1 − LCef f V̈C − RL I1 − RL Cef f V̇C , (3.4) and setting the time derivative of this equal to (3.1) gives: ... I1 −LI¨1 − LCef f V C − RL I˙1 − RL Cef f V̈C = + R2 I˙1 . C1 (3.5) Finally, inserting (3.1) into this and simplifying gives us an ordinary differential equation for I1 : ... Cef f −R2 LCef f I 1 − L + L + RL R2 Cef f I¨1 C1 Cef f 1 − RL + R L + R2 I˙1 − I1 = 0 . C1 C1 (3.6) To solve this differential equation numerically, we need to find the initial conditions. We assume that the voltage VC (0) is zero and so is the initial current I1 (0). The initial current I2 (0) through the IGBT is 750A. Using this, we equate (3.1) and (3.2) to get I˙1 (0): I˙1 (0) = 1 I2 (0) . R2 Cef f (3.7) We can find I¨1 (0) by equating the time derivatives of (3.1) and (3.2) and inserting I˙2 (0) from (3.3) to get: 1 I¨1 (0) = R2 ! V0 1 RL 1 − I2 (0) . 2 I2 (0) + R L I2 (0) − R C C LCef f R2 Cef 2 2 1 ef f f (3.8) 3 Computer Simulation 21 We can also find the voltage: VC 3.2 = V0 − L I˙1 + I˙2 − RL (I1 + I2 ) C ef f = V0 − R2 LCef f I¨1 − L + L + RL R2 Cef f I˙1 C1 Cef f − RL + RL I1 . C1 (3.9) Numerical Results Solving these differential equations with MATLAB (see Appendix A for program code) yields the results seen in Figs. 3.2 and 3.3. 400 300 I1 [A] 200 100 0 −100 −200 0 0.5 1 1.5 2 2.5 t [s] 3 3.5 4 4.5 5 −4 x 10 Figure 3.2: Current vs. time The switching time of 100µs for both current and voltage is very good. As can be seen in Fig. 3.3 however, the voltage rises to a maximum of 3500V, which is much higher than the 600V voltage rating of the IGBTs. We must therefore find a way to limit the peak voltage. 3 Computer Simulation 22 3500 3000 2500 VC [V] 2000 1500 1000 500 0 −500 −1000 0 0.5 1 1.5 2 2.5 t [s] 3 3.5 4 4.5 5 −4 x 10 Figure 3.3: Voltage vs. time 3.3 Effects of Varistor Now we want to examine the effect of the varistor, which will conduct current once the voltage reaches the threshold of 400V. To simulate this situation, we can assume that the voltage VC (0) is and always has been at 400V. Then V̇C (0) = V̈C (0) = 0 and I1 (0) = 0. From this and using VC (0) = I2 (0) Cef f and from V̇C (0) = 0, I¨1 (0) = it follows that I˙2 (0) = I¨2 (0) = 0. We also find I˙1 (0) = − VCL(0) RL L VC (0). The results can be seen in Figs. 3.4 and 3.5. The voltage is now limited to under 500V while the current reaches approximately 63A. The switching time for the voltage is still about 100µs. It has gotten slightly higher for the current (circa 250µs), but this is still acceptable. This time does not include the time in which the varistor conducts. 3 Computer Simulation 23 0 −10 −20 I1 [A] −30 −40 −50 −60 −70 0 0.5 1 1.5 2 2.5 t [s] 3 3.5 4 4.5 5 −4 x 10 Figure 3.4: Current vs. time with varistor 500 400 300 C V [V] 200 100 0 −100 −200 0 0.5 1 1.5 2 2.5 t [s] 3 3.5 4 Figure 3.5: Voltage vs. time with varistor 4.5 5 −4 x 10 Chapter 4 Technical Details 4.1 Making the Coils As part of the trap, we made two coils out of 1/4 x 1/4-inch square copper tubing. This somewhat unconventional tube shape allowed tight packing while maximizing the crosssection of the tube to get as low a resistance as possible. Coil 1 (see markings) was made first. The coil has about 17 turns, a resistance of 8.5mΩ and an inductance of 50µH. We made coil 2 later, trying to match these values as closely as possible. This coil also has approximately 17 turns, a resistance of 8.6mΩ and an inductance of 49.9µH. The inductance of the two coils together is approximately 83.3µH. We then measured the magnetic field between the coils at a current of 10A and a separation of the coils of about 68mm. This separation is close to how the experiment will be set up. The results of the measurement with the Hall probe are seen in Fig. 4.1. As can be seen in the figure, the zero point of the magnetic field is about 0.5mm from the center, which is good enough. The gradient in the region we are interested in is constant to a good approximation. Given the properties of the coil, we can now estimate the magnetic field gradient. In cylindrical coordinates, the components of the magnetic field generated by a quadrupole 24 4 Technical Details 25 2.0 1.5 1.0 0.244z - 8.393 B [G] 0.5 0.0 -0.5 -1.0 -1.5 -2.0 26 28 30 32 34 36 38 40 42 z [m m ] Figure 4.1: Magnetic field of the coils trap can be described by an expansion as given in [12]: X 3 3 Bz = bn Bzn = b1 z + b3 z − zρ + . . . 2 n X X Bρ = bn Bρn + ρn [cn cos((n + 1)φ) + dn sin((n + 1)φ)] n (4.1) n 1 3 2 3 2 = −b1 ρ − b3 ρz + ρ + . . . 2 2 8 X + ρn [cn cos((n + 1)φ) + dn sin((n + 1)φ)] , (4.2) n where b1 = 3µIAR2 (R2 +A2 )5/2 and b3 = 4(4A2 −3R2 ) b . 6(A2 +ρ2 )2 1 The coils are positioned at z = ±A, their radius is given by R and in vacuum, µ = µ0 = 4π × 10−7 in units A, m, T and µ0 = 4π 10 in units A, cm, G. We now consider the gradient in the direction of z. The expansion gives us dBz 3 2 2 = b1 + b3 3z − ρ + . . . , dz 2 (4.3) so for small values of z and ρ = 0, the gradient is just equal to N · b1 if our coils have N turns. Inserting the values of our coil geometry I = 750A, N = 17, A ≈ 3.5cm and 4 Technical Details 26 R ≈ 8cm gives us dBz 3µIAR2 G G ≈N 2 ≈ 211 . ≈ N · 14.3 5/2 2 dz cm cm (R + A ) (4.4) We can compare the measured value for the gradient with the calculated one. In Fig. 4.1, G the gradient is approximately 2.4 cm . Using Eq. 4.4 with a current I = 10A gives us a G . The difference can be attributed to the approximations made in the gradient of 2.8 cm calculation as well as the fact that the coils do not have a fixed radius or separation, but that these vary from turn to turn. Using the values for the coil geometry, we can now estimate the depth of our trap as well as the fraction of its maximum strength when turning the field on. For both estimates, we need to know the magnetic moment of the Rb atoms. For Rubidium, the contribution of the nuclear magnetic moment is negligible, so we use the values of the electronic gyromagnetic moment g ≈ 2 and the magnetic quantum number mS = 21 . We can therefore use µm = µB in Eq. 1.2. For the depth of the trap as described in section 1.1, given by the magnetic field at the edge of the smaller magnetic trap (radius z ≈ 1cm), we assume a linear potential with the gradient as given by Eq. 4.4 so that Eq. 1.3 becomes T = 1 dBz µB z ≈ 14mK . kB dz (4.5) We now estimate at what fraction of its maximum strength the field should be turned on in order to match the temperature and size of the atom cloud (cf. section 1.2.2). Assuming the cloud has a diameter of 1mm and a temperature of 50µK, we get dB kB T 50µK T G = ≈ ≈ 0.149 = 14.9 , dz µB z 0.6717KT−1 · 0.5mm m m which is about 7% of the maximum value given by Eq. 4.4. (4.6) 4 Technical Details 4.2 Subcircuits 4.2.1 Debouncer 27 The manual switches that we used for testing and manual control have the tendency to “bounce” when switching, i.e. they toggle between on and off several times before settling into the desired state. To prevent this, [18] suggest a debouncer as in Fig. 4.2(a). The debouncer can be realized with a J-K flip-flop, as can be seen in Fig. 4.2(b). We put these debouncers in front of the on/off and ready signals that come from the manual switches. The debouncer is on board 1. (a) (b) +5 A Q A GND B Q Q ____ C L R 1 _ V_ C_ C_ CLR2 J_ 1_ __ J 2 K1 K2 C_ LK1 _ P R 1 C _L _K _2 Q_1_ PR2 Q2 Q1 __ Q2 GND B +5 Figure 4.2: (a) Switch debouncer. (b) Connections on J-K flip-flop. 4.2.2 LEDs In order to see what state the computer inputs on/off and ready are, we want to include LEDs on the front panel that light up when the inputs are high. Since the computer cannot supply the current needed to drive the LEDs, we need to add a subcircuit that does this. 4 Technical Details 28 We would also like to make sure that the signal is not affected by noise on the line. We achieve this by adding two Schmitt trigger inverters. The part of the circuit that does this is seen in Fig. 4.3. Signal in Signal out 5V LED Figure 4.3: Subcircuit to drive LEDs The resistor in series with the LED must be chosen to give the current required for the LED. In the case of the green LEDs we used for the computer inputs on/off and ready, this current is I = 2mA. Given a voltage of U = 5V, the resistance must be R = U I = 2.5kΩ. We would also like to see the state of Vcontrol . Since the Schmitt trigger produces unsatisfactory results here, we use just the transistor and LED, driven by the output Q̄ of the J-K flip-flop on board 2. 4.3 Connection Diagrams On the following pages are connection diagrams for the front panel as well as for the two cards with the control circuit on them. The tables list the connections between them. 4.3.1 Front Panel See Fig. 4.4. 4 Technical Details 29 Board2 Z A Board1 22 22 1 1 Manual/ computer ready on/off +15V 5V -15V Connectors to/from coils Vcontrol V1 To battery IGBT3 out V1 battery in battery out ready Vin on/off Figure 4.4: Connections on the front panel 4.3.2 Board 1 See Fig. 4.5 and Table 4.3.2. 4.3.3 Board 2 See Fig. 4.6 and Table 4.3.3. The delays on the one-shot are set by the combination of a resistor and a capacitor. In front of the J and K̄ inputs of the flip-flop, we want a short delay. Here, the delay time is given by 0.7 tw = 0.28 × RT × Cext × 1 + RT (4.7) where tw is measured in ns, RT in kΩ and Cext in pF. Using RT = 5kΩ and Cext = 10nF 4 Technical Details 30 74ACT14 I1 O1 I2 O2 I3 O3 GND VCC I4 O4 I5 O5 I6 O6 ____ C L R 1 _ V_ C_ C_ CLR2 J_ 1_ __ J 2 K1 K2 C_ LK1 _ P R 1 C _L _K _2 Q_1_ PR2 Q2 Q1 __ Q2 GND B 1 Y 22 Figure 4.5: Connection diagram of board 1 gives a delay tw ≈ 16µs. Before the optocoupler, we want a pulse of about 10ms. In this case tw = 0.33 × RT × Cext (4.8) with the values in the units as before. If we use a 330nF capacitor and 100kΩ resistor, the 4 Technical Details 31 pulse time will be tw ≈ 10.9ms. IC5 74LS08N 74LS08N A1 B1 Y1 A2 B2 Y2 GND A1 B1 Y1 A2 B2 Y2 GND VCC B4 A4 Y4 B3 A3 Y3 VCC B4 A4 Y4 B3 A3 Y3 IC9 IC4 74LS123 74LS123 1A VCC 1_ B__ 1R/C 1_ CLR 1C _ 1Q 1Q _2 _Q _ 2Q 2C 2CLR 2R/C 2B GND 2A 1A VCC 1_ B__ 1R/C 1_ CLR 1C _ 1Q 1Q _2 _Q _ 2Q 2C 2CLR 2R/C 2B GND 2A IC2 IC8 ADG417BN S NC GND VDD D VSS IN VL IC7 ____ C L R 1 _ V_ C_ C_ CLR2 J_ 1_ __ J 2 K1 K2 C_ LK1 _ P R 1 C _L _K _2 Q_1_ PR2 Q2 Q1 __ Q2 GND LM339N O2 O1 V+ I1I1+ I2I2+ O3 O4 GND I4+ I4I3+ I3- IC1 411C BAL INV+15 IN+ OUT V-15 BAL IC10 CNY17 ANODE CATH IC6 4070BE A1 B1 Y1 Y2 B2 A2 VSS B C E Figure 4.6: Connection diagram of board 2 VDD B4 A4 Y4 Y3 B3 A3 IC3 4 Technical Details 32 A,Z ground green 1 computer on/off in brown 2 computer ready in blue 4 +5V yellow 6 manual on/off A brown 7 manual on/off B brown 8 manual on/off out brown 11 manual ready A blue 12 manual ready B blue 13 manual ready out blue 14 battery in (to +) purple 15 optocoupler in purple 17 battery out (to -) purple 18 IGBT3 out purple B computer on/off to board brown C computer ready to board blue D on/off LED brown E ready LED blue K shorted to L L shorted to K R LED Vcontrol (Q) out orange S LED Vcontrol (Q) in orange Table 4.1: Connections on first board 4 Technical Details 33 A,Z ground green 2 +15V red 3 battery in (to +) purple 4 +5V yellow 6 V1 (output for IGBT1) orange 8 V 1 (output for IGBT2) orange 10 V2 (optocoupler output) orange 12 Vcontrol orange 13 Vin gray 15 ready blue 18 on/off brown 21 -15V black 22 Q out orange Table 4.2: Connections on second card Chapter 5 Performance 5.1 Measurements with “Dummy” Coil The first test of the circuit was performed with a “dummy” coil, since the real coils were already in use as MOT coils in the experimental setup. This “dummy” coil had a similar inductance of around 100µH, but a larger resistance of R ≈ 0.6Ω. Since the resistance was higher and the flow rate of water through the coils lower, it was only possible to increase the current up to values of around 400A before the coil got too hot. In order to measure the changing magnetic field, we installed a probe coil with its axis parallel to the axis of the “dummy” coil and measured the voltage induced in it. Again because of the law of induction V = −L dI dt , this voltage is a measure of the change of the magnetic field. Fig. 5.1 shows the effect of turning on the magnetic field with the current from S1 at 400A and the voltage of S2 at 500V. As can be seen, VBE rises to about 11.5V, meaning that the switch of IBGT3 is now closed. Approximately 300µs later, the capacitors start to discharge and the probe coil measures a change in the magnetic field. The whole process of switching on is completed after about 700µs. In Fig. 5.2, the voltage at the collector of IGBT1, VC , is clamped by the varistor to about 350V and dies down after about 200µs. Similarly, the voltage in the probe coil peaks 34 5 Performance 35 500 Voltage [V] 400 300 200 100 0 -100 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Voltage [V] 10 0 t [ms] Figure 5.1: Turning on the magnetic field. The upper diagram shows the voltage at the capacitors. In the lower diagram, the solid line is VBE at IGBT3 and the dotted line the voltage at the probe coil divided by a factor of 5. sharply and then decays after 200µs. The total switching off time is thus about a third of the time for turning on. 5.2 Measurements with Quadrupole Coils The circuit was then connected to the actual quadrupole coils positioned on the optical table for the BEC experiment. These coils were able to handle a current of 750A through 5 Performance 36 350 300 Voltage [V] 250 200 150 100 50 0 -50 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Voltage [V] 0 -50 t [ms] Figure 5.2: Turning off the magnetic field. In the upper diagram, the solid line is VC at IGBT1 and the dashed line is VBE at IGBT3. The lower diagram shows the voltage at the probe coil. effective water cooling. The magnetic field was now switched on and off at a voltage of 500V at S2 and a current of 750A generated by S1 . The voltage at S1 when the current was at its maximum value was 17.2V. Since the coils were already in place around the chamber, it was not possible to install the probe coil with its axis parallel to the axis of the quadrupole coils. We therefore placed it with its axis perpendicular to that of the quadrupole coils and thus measured the change 5 Performance 37 500 Voltage [V] 400 300 200 100 0 -100 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 15 Voltage [V] 10 5 0 -5 t [m s] Figure 5.3: Turning on the magnetic field. The upper diagram shows the voltage at the capacitors. In the lower diagram, the solid line is VBE at IGBT3 and the dashed line the voltage at the probe coil. in Bρ instead of Bz . As we were primarily interested in the time of change and not the absolute values of the field, this was not a problem. Fig. 5.3 shows the measurement of switching on the current through the coils. Again, there is a delay of approximately 200µs between switching (IGBT1 switches on, IGBT2 switches off) and the capacitors discharging. The whole process of switching is again finished after 600µs. 5 Performance 38 400 350 300 Voltage [V] 250 200 150 100 50 0 -50 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Voltage [V] 15 10 5 0 -5 t [m s] Figure 5.4: Turning off the magnetic field. The upper diagram shows VC at IGBT1. In the lower diagram the solid line shows the voltage at the probe coil and the dashed line is VB at IGBT1. The measurement of turning off the magnetic field can be seen in Fig 5.4. The voltage is clamped by the varistor to the same value as before. Since the current is higher and thus the magnetic field is stronger, this clamped high voltage persists for a longer time, ca. 250µs. The change in the magnetic field is also on the order of 250µs. The switching times thus do not depend dramatically on the current flowing through the coils. 5 Performance 5.3 39 Conclusion We have implemented a strong magnetic field that can be used to trap neutral atoms for evaporative cooling. We have shown that the switching times of the circuit controlling the field are under 1ms when running at its maximum power, as desired and predicted by the computer simulation. Since the final plan calls for switching the field on at around 7% of its maximum strength, the time should be somewhat less in practice. An emergency interrupt has already been made to shut off the current source S1 if temperatures at critical points exceed 100◦ C. This could happen at the shunt resistors, 2 which dissipate P = RI 2 = 0.1Ω × 750A ≈ 1500W of power for short periods of time, 6 which is almost twice the rated power. Overheating these components would unquestionably lead to their destruction. In addition, the computer control for the current source S1 will be implemented, allowing us to vary the magnetic field such that it follows the optimum path of evaporation calculated by a computer simulation. As of now, current source S4 has not been integrated into the circuit and will need a separate control circuit controlling it, preferably also with computer inputs. When these activities are finished, the achievement of BEC and the onset of experiments with an atom interferometer should be only a matter of time. Appendix A MATLAB code A.1 sc.m % program sc % solves ODE of switching circuit clear; L = 100e-6; RL = 16.7e-3; R2 = 8; C1 = 10e-6; Ceff = 496e-9+1000e-9; % use only one set of these initial conditions % initial conditions I2 = 750; I1 = 0; dI1 = I2/(R2*Ceff); ddI1 = 1/R2*(1.2/Ceff/L - I2/R2/Ceff^2 + RL*I2/R2/L - I2/R2/C1/Ceff); % initial conditions after varistor shutdown V = 400; I1 = 0; dI1 = -V/L; ddI1 = RL/L^2*V; % solve ODE for current [T,Y] = ode45(’V’,[0 0.001],[I1;dI1;ddI1]); figure(1); plot(T,Y(:,1)); %title(’Current vs. time in switching circuit’); xlabel(’t [s]’); 40 A MATLAB code ylabel(’I_1 [A]’); % solve ODE for voltage VC = 1.2 - L*R2*Ceff.*Y(:,3) - (L+L*Ceff/C1+RL*R2*Ceff).*Y(:,2) - (RL+RL*Ceff/C1).*Y(:,1); figure(2); plot(T,VC); %title(’Voltage vs. time in switching circuit’); xlabel(’t [s]’); ylabel(’V_C [V]’); A.2 V.m function dy = V(t,y) L = 100e-6; RL = 16.7e-3; R2 = 8; C1 = 10e-6; Ceff = 496e-9+1000e-9; a = R2*L*Ceff; b = L+L*Ceff/C1+RL*R2*Ceff; c = RL+RL*Ceff/C1+R2; d = 1/C1; dy = [y(2); y(3); -b/a*y(3)-c/a*y(2)-d/a*y(1)]; 41 Appendix B Part Specifications B.1 Main Circuit • IGBT: Powerex CM600HU-12F max. values: ICE = 600A VCE = 600V VBE = 20V IGBT1: 3 in parallel; IGBT2: 2 in parallel; IGBT3: 1 • Diodes D1 : 6 diodes IRKE320-08 in parallel. Forward voltage drop per diode: ca. 0.8V at 80A. • Varistor: Littelfuse V151BA60 Clamping voltage: 400V Max. energy dissipated in single pulse (2ms): 530J • Resistors R1 : 6 resistors ISOTEK IRV800P in parallel at 0.1Ω each. Maximum power dissipated: 800W each. • Resistor R2 : RH-50 8Ω. Maximum power dissipated: 50W. • Capacitor C1 : GE 97F5300S, 10µF. Maximum voltage: 1000V • Capacitors C2 : 5 capacitors GE 97F5211S in parallel at 50µF each. Maximum voltage: 1000V. • Capacitor C3 : GE A28F5604, 1µF. Maximum voltage: 2000V B.2 Board 1 • J-K flip-flop: 74LS109 • Hex inverter with Schmitt trigger input: 74ACT14 42 B Part Specifications B.3 Board 2 • Quad 2-input AND gate: 74LS08 • Dual One-Shot: 74LS123 • Quad Comparator: LM339 • Optocoupler: CNY17 • J-K flip-flop: 74LS109 • Quad 2-input XOR gate: 4070 • Op Amp: LF411 • CMOS Switch: ADG417 43 Bibliography [1] M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E. Wieman, E. A. Cornell. Observation of Bose-Einstein Condensation in a Dilute Atomic Vapor. Science 269, 198(1995) [2] K. Davis, M-O. Mewes, M. Andrews, M. van Druten, D. Durfee, D. Kurn, W. Ketterle. Bose-Einstein Condensation in a Gas of Sodium Atoms. Phys. Rev. Lett. 75, 3969(1995) [3] C. C. Bradley, C. A. Sackett, J. J. Tollett, R. G. Hulet. Evidence of Bose-Einstein Condensation in an Atomic Gas with Attractive Interactions. Phys. Rev. Lett. 75, 1687(1995) [4] C. A. Sackett. A Condensate-Based Atom Interferometer for Ultra-Precise Gyroscopic Sensing. Grant proposal (2001). [5] G. E. Stedman. Ring-laser tests of fundamental physics and geophysics. Rep. Prog. Phys. 60, 615(1997) [6] H. J. Metcalf, P. van der Straten. Laser Cooling and Trapping. Springer, New York (1999) [7] W. Ketterle, D. S. Durfee, D. M. Stamper-Kurn. Making, probing and understanding Bose-Einstein condensates, in Proceedings of the International School of Physics ‘Enrico Fermi’ Course CXL. IOS Press, Amsterdam (1999) [8] W. H. Wing. On neutral particle trapping in quasistatic electromagnetic fields. Prog. Quant. Elec. 8, 181(1984) [9] P. Meystre. Atom Optics. Springer, New York (2001) [10] W. Paul. Electromagnetic traps for charged and neutral particles. Rev. Mod. Phys. 62, 531(1990) [11] A. L. Migdall, J. V. Prodan, W. D. Phillips, T. H. Bergeman, H. J. Metcalf. First Observation of Magnetically Trapped Neutral Atoms. Phys. Rev. Lett. 54, 2596(1985) [12] T. Bergeman, G. Erez, H. J. Metcalf. Magnetostatic trapping fields for neutral atoms. Phys. Rev. A 35, 1535(1987) [13] E. Majorana. Nuovo Cimento 9, 43(1932) 44 BIBLIOGRAPHY 45 [14] K. Helmerson, W. D. Phillips. Cooling, trapping and manipulation of neutral atoms and BEC by electromagnetic fields, in Proceedings of the International School of Physics ‘Enrico Fermi’ Course CXL. IOS Press, Amsterdam (1999) [15] W. Ketterle, N. J. van Druten. Evaporative Cooling of Trapped Atoms, in Advances in Atomic, Molecular, and Optical Physics Vol. 37. Academic Press, San Diego (1996) [16] J. M. Vogels, C. C. Tsai, R. S. Freeland, S. Kokkelmans, B. J. Verhaar, D. J. Heinzen. Prediction of Feshbach resonances in collisions of ultracold rubidium atoms. Phys. Rev. A 56, R1067(1997) [17] W. Petrich, M. H. Anderson, J. R. Ensher, E. A. Cornell. Stable, Tightly Confining Magnetic Trap for Evaporative Cooling of Neutral Atoms. Phys. Rev. Lett. 74, 3352(1995) [18] P. Horowitz, W. Hill. The Art of Electronics. Cambridge University Press, Cambridge (1989)