Design and Implementation of On-board Electrical Power ... - OUFTI-1
Design and Implementation of On-board Electrical Power ... - OUFTI-1 Design and Implementation of On-board Electrical Power ... - OUFTI-1
sequently, the k was chosen above 0.4 in our two converters. The more limiting factor wasthe series resistance, which growths quickly when the size of the inductor is decreased (thesection of the wire is reduced).5.3 Design of input filtersAn undesirable feature of switch-mode power converters is their generation of conducted andradiated electromagnetic interference (EMI) on their input at the switching frequency andits harmonics. The power source and other systems supplied by the same power sourcescan be corrupted by EMI currents, if not filtered. Without filtering, EMI will be radiatedby the input line and interference with the operation of near equipments, especially radioequipments.5.3.1 Type of filterThe datasheets of the converters used in section 5.2(TPS63001, TPS61087, and LTC3528)recommend using one 4.7µF to 10µF ceramic capacitor close to the input of the converter.With the resistance of the source and the input line, this makes up a first-order low-pass filter.This may be sufficient in most of the usual applications for these converters, when they aredirectly connected to a battery.In our application, the converters are connected to the batteries bus. The impedanceof the bus seen from a converter is the impedance of the line plus the impedance of all thesystems connected to this bus in parallel (two batteries, five solar panels, two other converters,the EPS2...). As a consequence, the impedance may be quite low and it is difficult to knowthe cut-off frequency of the input filter if it only consists in a capacitor.Therefore, second-order low-pass filters will be used for the converter inputs. Second-orderfilters offer a better attenuation per decade of EMI and their cut-off frequencies can be chosenwith precision.5.3.2 Stability problemAs explained in Chapter 10 of [16], a converter is designed to have an input-to-output transferfunction G vg (s) (the ”audiosusceptibility”) sufficiently small over a wide frequency range. Theoutput voltage is regulated in spite of variations in the input voltage. The introduction of aninput filter will change the dynamics of the converter, often in a manner that degrades theregulator performance. The audiosusceptibility is degraded and there are conditions underwhich the system may even go unstable.The input power of a converter is more or less constant with the input voltage (P in =P out /η), thus one can writeP in = V in I in ⇒ V in = P inI in⇒ dV indI in= −P inI 2 in= −V inI in. (5.35)70
Equation 5.35 is a simplification. In actuality, the control loop impacts the frequencyresponse of the input impedance. As a consequence, the input impedance is not constantwith frequency. The important thing to observe is that the slope of the voltage-current curve,which defines the dynamic impedance of the power supply, is negative (Fig. 5.18).Figure 5.18: V-I curve of a converter input (from [24]).If the source impedance and the dynamic input impedance of the converter have equalvalues but opposite signs at a given frequency, the voltage tends to infinity. The system isthen unstable. The circuit corresponding to this situation is shown in Fig. 5.19.Figure 5.19: The negative impedance can result in oscillations (from [24]).A good solution to have a stable system is to ensure that the magnitude of the outputimpedance of the source is always smaller than the magnitude of the input impedance of theconverter. The impedance of the converter must be considered when it is minimal, i.e. withthe lowest input voltage and the highest load.Details on this stability problem are found in [24] and [16].5.3.3 Middlebrook’s criterionThe Middlebrook’s extra element theorem can be employed to determine how the addition ofan input filter affects the control-to-output transfer function [16]. The modified control-tooutputtransfer function can be expressed as follows1 + Zo(s)ZG 2 = G N (s)11 − Zo(s) ,Z D (s)71
- Page 20 and 21: Chapter 3Design of EPS architecture
- Page 22 and 23: • Voltage (4) and current (5) at
- Page 24 and 25: Figure 3.6: The equivalent circuit
- Page 26 and 27: of our Lithium-Polymer batteries va
- Page 28 and 29: Figure 3.12: I-V curve of a solar p
- Page 30 and 31: 3.3.3 CapacityA important value to
- Page 32 and 33: Parameter SLPB723870H4 SLPB554374HN
- Page 34 and 35: of the batteries is kept between -2
- Page 36 and 37: Over Charge Prohibition 4.275 ± 0.
- Page 38 and 39: supplied in 5V. The circuit will be
- Page 40 and 41: Chapter 4The Power Budget4.1 Introd
- Page 42 and 43: Figure 4.1: P-V curve of a solar pa
- Page 44 and 45: 4.3.2 Efficiency of convertersTo at
- Page 46 and 47: Figure 4.3: Consumptions in % in me
- Page 48 and 49: Chapter 5Electrical Design of EPS5.
- Page 50 and 51: V outV in= D. (5.1)Since D ≤ 1, t
- Page 52 and 53: The power losses in the inductor ar
- Page 54 and 55: ∆i L,1 + ∆i L,2 = 0, (5.16)V in
- Page 56 and 57: Using the value of ∆i L given by
- Page 58 and 59: There is no data about the case to
- Page 60 and 61: Capacitor selectionFour 10µF ceram
- Page 62 and 63: • Output voltage: 5V.• Maximum
- Page 64 and 65: Figure 5.12: Burst mode operation (
- Page 66 and 67: Figure 5.14: Simplified schematics
- Page 68 and 69: Figure 5.15: Worksheet for 3.3V con
- Page 72 and 73: where G 1 is the initial control-to
- Page 74 and 75: Figure 5.21: Measured Bode diagram
- Page 76 and 77: Figure 5.26: Equivalence between th
- Page 78 and 79: C f =12πf f R 0f,L f = R 2 0f C f
- Page 80 and 81: Figure 5.37: Schematics of the firs
- Page 82 and 83: R KR >1.45V100mA − 1.3A35= 23.07
- Page 84 and 85: The schematics is shown on figure 5
- Page 86 and 87: A commercial model meets all requir
- Page 88 and 89: Figure 5.45: Schematics of the heat
- Page 90 and 91: PrefixX7X5Y5Z5SuffixTemperature ran
- Page 92 and 93: 6.2.1 The second dissipation system
- Page 94 and 95: • The antenna deployment system.
- Page 96 and 97: 6.3.3 TestsThe engineering model of
- Page 98 and 99: 7.3 ActivitiesAs OUFTI-1 is designe
- Page 100 and 101: 8.1.2 DesignA model of Li-Po batter
- Page 102 and 103: [15] Fabien Jordan, Phase B Electri
- Page 104 and 105: TaK = 273 + TaC;%Photo-current ther
- Page 106 and 107: Appendix BPower budget worksheetIn
- Page 108 and 109: 108
- Page 110 and 111: Appendix CPictures of the prototype
- Page 112 and 113: Appendix DSchematics of the enginee
- Page 114 and 115: 876543213V3 CONVERTER AND INPUT FIL
- Page 116: 87654321ANTENNA DEPLOYMENT CIRCUITB
Equation 5.35 is a simplification. In actuality, the control loop impacts the frequencyresponse <strong>of</strong> the input impedance. As a consequence, the input impedance is not constantwith frequency. The important thing to observe is that the slope <strong>of</strong> the voltage-current curve,which defines the dynamic impedance <strong>of</strong> the power supply, is negative (Fig. 5.18).Figure 5.18: V-I curve <strong>of</strong> a converter input (from [24]).If the source impedance <strong>and</strong> the dynamic input impedance <strong>of</strong> the converter have equalvalues but opposite signs at a given frequency, the voltage tends to infinity. The system isthen unstable. The circuit corresponding to this situation is shown in Fig. 5.19.Figure 5.19: The negative impedance can result in oscillations (from [24]).A good solution to have a stable system is to ensure that the magnitude <strong>of</strong> the outputimpedance <strong>of</strong> the source is always smaller than the magnitude <strong>of</strong> the input impedance <strong>of</strong> theconverter. The impedance <strong>of</strong> the converter must be considered when it is minimal, i.e. withthe lowest input voltage <strong>and</strong> the highest load.Details on this stability problem are found in [24] <strong>and</strong> [16].5.3.3 Middlebrook’s criterionThe Middlebrook’s extra element theorem can be employed to determine how the addition <strong>of</strong>an input filter affects the control-to-output transfer function [16]. The modified control-tooutputtransfer function can be expressed as follows1 + Zo(s)ZG 2 = G N (s)11 − Zo(s) ,Z D (s)71