Power electronics CAD: from space applications to industrial applications

Size: px

Start display at page:

Download "Power electronics CAD: from space applications to industrial applications"

Ralph Sharp
5 years ago
Views:

1 93

Power electronics CAD: from space applications to industrial applications Fig.3 Globalstar boost pre-regulator schematic. Power electronics CAD: from space applications to industrial applications P.G. Maranesi and M.

Its applications span over several areas: power semiconductor devices, power supply, control of power networks, HV dc systems, automotive, aircraft electric power systems, spacecraft power systems,

Space applications are most demanding in terms of reliability, efficiency and radiation hardness for the apparatus.

distribution apparatus aboard satellites and the International Space Station and also of HV supplies for ionic propulsion.

2 Power electronics CAD: from space applications to industrial applications Fig.3 Globalstar boost pre-regulator schematic. Power electronics CAD: from space applications to industrial applications P.G. Maranesi and M. Riva 1 Electronic devices era either employed to process the information associated to electric signals or to process the electric power. The latter role is that of Power. Its applications span over several areas: power semiconductor devices, power supply, control of power networks, HV dc systems, automotive, aircraft electric power systems, spacecraft power systems, biomedical instrumentation, etc. Constraints and critical aspects are related to each specific application. Space applications are most demanding in terms of reliability, efficiency and radiation hardness for the apparatus. The electronic section of the Department of Physics in the University of Milan has been involved in the last years in electronic power applications for space: the design of power converters and distribution apparatus aboard satellites and the International Space Station and also of HV supplies for ionic propulsion. In the figure 1a) a DC-DC converter payload supply circuit, proposed by the power electronics laboratory of the Department of Physics and installed aboard the International Space Station, (ISS) is shown; in 1b) the schematic of a HV generator for Radiofrequency Ion Thruster (RIT), developed in the same lab, is given. Fig. 4 FREDOMSIM forecasts vs experimental data of the frequency responses of input admittance a) and audio-susceptibility b) transfer functions. 94 Examples of power electronics converter; a) a phase-shifted full-bridge applied aboard the ISS and b) HV generators for ion thruster drive. Another field of the activities of the lab is the development of automatic CAD tools for the analysis and the design optimization of electronic power systems. The main product of this activity is an innovative CAD tool (FREDOMSIM) for the characterization in the frequency domain of the dynamic behaviors of the switched converters. Starting from the schematic entry according to the SPICE simulator of the circuit considered (in fig. 2 an example is represented), FRE- DOMSIM provides forecasts of the frequency responses of the transfer functions, characterizing the circuit with the feedback loops open. The high degree of compliance with real life can be evaluated in the fig. 3, where the plots are relevant to the boost converter in the fig. 2. The application of FREDOMSIM is user friendly (its visual interface is shown in the fig. 4) and it supports the designer in the optimization of the inner and external feedback loops. Even though the automatic tool was devoted to space applications, it demonstrated validity for the design of electronic circuits for terrestrial application too. The fig. 5 shows the block scheme of a distributed power assembly developed for industrial applications in cooperation with a semiconductor manufacturer. The high efficiency reached at 500kHz of switching frequency has been made possible by the technologies and the automatic tool originally developed for the design aid of space power circuits. Fig. 5 Typical scheme of a distributed power assembly. 95 Fig. 6 Example of the input admittance frequency responses of a closed-loop buck converter. Some of the FREDOMSIM features.

3 High dynamic range low-noise preamplification of nuclear signals Architecture of the time-variant preamplifier. The connection cable and the ADC are also shown. 2 INFN A. Pullia 1,2, F. Zocca 1,2, C. Boiano 2, R. Bassini 2, S. Riboldi 1,2, and D. Maiocchi 1,2 We have developed and realized a new mixed continuous-pulsed reset preamplifier purposely conceived for advanced multi-channel gamma-ray sensors. The continuous-reset mode is used for typical signal amplitudes within the ADC input range, or from a few kev to a few MeV. The pulsed-reset mode works on larger signal amplitudes, up to 100 MeV caused by either individual events or by bursts of piled-up events. Such operation mode maximizes the live time in the typical energy range and minimizes the dead time caused by ADC saturation on large individual signals or at high count rates. The amplitude of large individual signals can also be estimated from the reset time still achieving a high resolution. Using this technique and a bulky HPGe detector we got a continuous-reset range of 2 MeV with a resolution of 2.02 kev fwhm at MeV, and an extended pulsed-reset range of 27 MeV, with a typical resolution of 0.3% at 27 MeV. 1. Introduction The development of advanced segmented detectors for the new generation of nuclear physics experiments with exotic beams (AGATA, GRETA) [1,2] poses challenging requirements to the front-end electronics, including an extremely high dynamic range and an ultra low noise over the whole range. The typical energy range of a digitized preamplifier for γ-ray spectroscopy is from a few kev to a few MeV. Individual highly energetic events which exceed this range or bursts of piled-up events at high count rates can easily cause ADC saturation and introduce a significant system dead time. We suggest that a mixed reset technique for the preamplifier can address these issues. The idea consists of using a simple continuous reset in the typical amplitude range and a fast pulsed reset for larger signals. When the amplitude of the preamplifier output signal is within the ADC input range, the circuit works in the classic continuous-reset mode, i.e. with the typical resistive continuous discharge of the feedback capacitance. But when the preamplifier output signal exceeds the ADC voltage range, the pulsed-reset mode intervenes which yields a fast recovery (in a few µs) of the output voltage to its floor level. The amplitude of a large individual signal can be indirectly estimated from the reset time, when the ADC overflow condition would give no possibility to make a direct measurement of the pulse height. Using this technique, we were able to measure highly energetic signals up to 30 MeV, still achieving a high resolution. As a result the dynamic range of a digitized preamplifier is substantially increased, as given by a continuous-reset range of a few MeV and a pulsed-reset range which can reach 30 MeV or more. Such a large energy range is of interest for the last generation of nuclear physics experiments with bulky 4π germanium detectors. In fact, Montecarlo simulations show that a 9 cm germanium layer still has a detection efficiency of 9 to 14% for γ-rays photons of 20 to 30 MeV. This large dynamic range is required for a class of experiments where giant-dipole resonances can produce γ photons of energies up to 30 MeV. (a) Large signal (H) upon a tail (h) at the PZ stage output. (b) Schmitt-trigger comparator signal. 3. Large signal measurement technique In a we show the typical reset transient seen at the P/Z amplifier output, when a large individual signal (H) occurs upon a tail of a regular signal (h). Figure 2b shows the corresponding waveform seen at the Schmitt-trigger comparator output, which delivers a rectangular pulse of width T, corresponding to the pulsed-reset time. The regular signal which provides the tail h does not exceed the threshold corresponding to ADC saturation and has the typical exponential decay time of about 50 µs. The large signal H is instead over threshold and puts the preamplifer into the pulsed reset mode. In the figure the ADC saturation at a voltage level of about 1V is not shown, as the signal has been acquired through a digital oscilloscope. But in the real acquisition chain, the preamplifier output is directly sampled by the ADC module and so it would not be possible to directly estimate the amplitude of the over-threshold signal. However, the energy information can be indirectly obtained. By accurately calculating the expression of the reset transient at the P/Z stage output, we can derive the exponential relation between the reset time T and the sum of the energy E S of the large event and the energy E C of the tail: REFERENCES 1. Technical Proposal for an Advanced Gamma Tracking Array for the European Gamma spectroscopy Community, 2. I. Y. Lee, Gamma-ray tracking detectors, Nucl. Instrum. and Meth., vol. A422, pp , Architecture of the time-variant charge preamplifier The architecture of the proposed time-variant preamplifier is shown in. Continuous reset is ensured through the feedback resistance R F, while pulsed reset is realized through an add-on switchable device we called Large-Signal Manager (LSM). It consists of an inverting Schmitt-trigger comparator and a temperature stabilized switchable current sink. The LSM senses the amplitude of the preamplifier output signals. On small-amplitude signals the current sink is in the off condition. In this case the circuit behaves like a classic charge preamplifier, with resistive continuous reset of the detector charge stored on the feedback capacitance. But when a large signal arrives, which exceeds the comparator threshold, the LSM switches the current sink on and puts the preamplifier into a pulsed-reset mode, so as to swiftly desaturate the ADC. As shown in Fig.1, in this phase the current sink continuously absorbs charge from capacitance C of the P/Z stage until the preamplifier output voltage is brought back to its DC level. At this point the comparator automatically turns over and switches the current sink off so completing the fast reset procedure. E S + E C = A (exp (T/τ) 1), (1) where A and τ are coefficients depending on circuit parameters. As the reset time T results substantially shorter than the time constant τ, the relation can be linearized by expanding the exponential term with no loss of accuracy. We can then express the large signal energy E S as a function of the reset time T and the energy of the tail E C, obtaining E S = b 1 T + b 2 T 2 E C, (2) where b 1 and b 2 are the first and second order coefficients. The reset time T can be obtained through a time over threshold measurement performed on the waveform acquired at the comparator output (see b), while the tail energy E C can be obtained from the difference of the pre and post large event voltage levels evaluated on the output signal of the preamplifier. So Eq. (2) can be written as E S = b 1 T + b 2 T 2 (V 1 V 2 ) /G + E 0, (3) where (V 1 V 2 ) is the voltage amplitude of the tail, G is the energy-to-voltage gain of the preamplifier and E 0 is an offset term. We can conclude that (3) permits to indirectly derive the large signal energy E S. In this relation T, V 1, and V 2 must be evaluated any time an over-threshold signal occurs, while b 1, b 2, G, and E 0 are parameters which must be obtained through preliminary calibrations.

4 4. Gamma spectroscopy with mixed continuous and pulsed reset: experimental results We tested the mixed-reset technique by using a custom-made preamplifier equipped with the add-on LSM device sketched in, connected to a state-of-the-art closed-end cylindrical HPGe detector. We used 152 Eu and 60 Co calibration sources placed at different distances from the detector, obtaining overall counting rates in the range 1 khz to 32 khz. We could also artificially simulate energetic events in the range 5 MeV to 27 MeV by providing large signals at the preamplifier input with a spectroscopy-grade pulser and a calibrated test capacitance. The pulser signals were large enough to trigger the LSM device. The fast-reset transients were superposed to the tails caused by the distribution of γ photons delivered by the mentioned radioactive sources. Fig. 6 Spectrum of 152 Eu with high-energy pulser lines. The average event rate is 6 khz. We then irradiated the detector with a 152 Eu source placed in such a way to get an event rate of 6 khz, and injected large pulser signals in the energy range from 5 to 27 MeV. All signals from the 152 Eu source were under threshold and have been acquired in the continuous-reset mode, while the pulser signals were over threshold, and have been acquired through the pulsed-reset technique, as explained before. As a result we got the overall spectrum shown in Fig. 6 in the extremely wide range from 10 kev to 27 MeV. All 152 Eu lines are well resolved, the MeV line having a fwhm of 2.02 kev. The high-energy pulser lines are also finely resolved, having a relative fwhm of the order of 3 perthousand. In Fig. 3 we show the result of the algorithm for the estimation of a ~ 10 MeV signal injected at the preamplifier input, when a background of 32 kcounts/s of 60 Co signals is present. As evident, if we try to reconstruct the large signal energy only by evaluating the reset time with no tail correction, the line completely disappears, as the stochastic distribution of tails due to 60 Co signals scrambles it out. When the tail correction from (3) is made, the line gets nicely reconstructed showing a relative fwhm of 0.56 ~ 10 MeV. As shown in Fig. 4, the resolution improves if the counting rate of the 60 Co regular signals is reduced, which diminishes the disturbing effect of the tails. In Fig. 5 the obtained resolution is shown in an energy range of 5 to 27 MeV. 98 Fig. 3 Large pulser line. A background of 32 kcounts/s of 60 Co signals is present which destroys the large signal resolution if no correction is made. 99 Fig. 4 Obtained energy resolution at 10 MeV for increasing counting rates of the 60 Co regular signals. Fig. 5 Obtained energy resolution at high energy for two different values of background event rate.

5 Ion thruster high voltage drive F. Belloni 1, P.G. Maranesi 1, and M. Riva 1 Electrostatic Radiofrequency Ion Thrusters (RIT) have been introduced in the last years for propulsion, orbit control, and orientation of space vehicles. This type of engines is characterized by the generation of a heavy ion plasma which is accelerated by the electric field of the high voltage supplies before ejection from the spacecraft. Accelerating forces in the order of several tens of millinewtons, as required by interplanetary missions, can be obtained by RITs of a few Kilowatts. The propellant is an inert gas, such as Argon, Krypton or most commonly Xenon that is ionized by a radiofrequency (RF generator) and then accelerated by two high voltage grids. The first (beam grid) is maintained at a positive voltage (Positive High Voltage Generator - PHVG) and the second one (extraction grid) at a negative level (Negative High Voltage Generator - NHVG). The schematic of the RIT, with HV and RF generators is shown in. 100 Short circuit between the positive and the negative voltage grids happens not rarely and the capacity of withstanding such occurrence is mandatory as well as of removing the short circuit. Due to the threshold effect of the positive grid, the beam current decreases by increasing the voltage generating a negative resistance behavior. Moreover, the plasma accelerated from the PHVG, constitutes a non-linear load. It s advisable that the high voltage cell consists of a transconductance power stage supplying current to the load and to the output capacitance with a single pole behavior. The proposed solution for a module PHVG module is a phase-shift PWM controlled converter switching at the frequency = 65.5 khz, rated power of 1.5kW and Vout = 1500V. The input DC voltage is 100V, the output is obtained by the connection in series of two 750V transformer coupled half bridges. The former is active and it uses MOS switches, the latter employs rectifying diodes. Zero voltage switch-on and voltage clamping characterize the switching transients. The IGBTs, in series to the output, avoid to overcome the load current limit in cases of short circuit or overcurrents and provide prompt switch-off of the output currents. The NHVG consists in a flyback topology, operating in discontinuous conduction mode, modified to achieve zero voltage transitions. The switching frequency is 65.5 khz, the rated power is 300W at maximum load and the output voltage can vary between -150V and -800V. The output voltage rectification is made by two passive single-ended rectifiers, in order to reduce the voltage stress on each diode. Schematics of both the PHVG and the NHVG are represented in. Structure of Matter 101 Short circuits between the two grids can happen due to impacts of positive ions on the grids and material settling. The protection implies current limitation and switch off. This inconvenience is removed by a dedicated operating procedure of the PHVG that aims to sublimate the short circuit material by Joule effect. The delay time is fixed at and the control of the current injected to the short circuit is done by varying the switching frequency. RIT and voltage generators schemes Electrical schemes of the PHVG and the NHVG

Gamma Ray Spectroscopy with NaI(Tl) and HPGe Detectors

Nuclear Physics #1 Gamma Ray Spectroscopy with NaI(Tl) and HPGe Detectors Introduction: In this experiment you will use both scintillation and semiconductor detectors to study γ- ray energy spectra. The