Table of contents 1 . introduction 1. 1 project summary 1. 2 objective 2. Theory 2. 1 single-phase full-bridge vsi 2. 2 pulse width modulation 2. 3 comparison between scr and mosfet based inverter 3. Model description 4. Working 5. Matlab simulink 1 . introduction 1 . 1 project summary This project presents a matlab simulink model of a dc-to-ac voltage source inverter. Load voltage rms feedback control and openloop control are used to compare the standard matlab simulink and power system blocksets used for the inverter model design. Simulation and experimental results using linear and non-linear loads are used to validate the accuracy of the model developed. Model present a high frequency transformer isolated, scr and mosfet based three-phase, voltage source inverter (vsi) for ac motor drives and uninterruptible power supply applications. The proposed converter produces a 3-phase sinusoidal ac output voltage from alternative energy sources (converted to a 12 v dc voltage source) while output voltage magnitude and frequency can both be controlled using scr and mosfet. 1 . 2 objective • to compare the pulse width modulation of scr based and mosfet based vsi for analyzing the harmonic distortion, voltage regulation, stability. • to provide such a converter with a fixed duty cycle selected so as to minimize third harmonic content in the signal supplied to the load. • to provide a two stage dc-to-ac converter that draws a generally constant current from a source of dc, meeting peak load demands with energy stored in the converter. • to provide separate voltage and current feedback loops to control and regulate the conversion process. 2 . theory 2 . 1 single-phase full-bridge vsi. Fig. 3 shows the power topology of a full-bridge vsi. This inverter is similar to the half-bridge inverter; however, a second leg provides the neutral point to the load. As expected, both switches s1+ and s1- (or s2+ and s2-) cannot be on simultaneously because a short circuit across the dc link voltage source vi would be produced. There are four defined (states 1, 2, 3, and 4) and one undefined (state 5) switch states as shown in table 2. The undefined condition should be avoided so as to be always capable of defining the ac output voltage. It can be observed that the ac output voltage can take values up to the dc link value vi, which is twice that obtained with half-bridge vsi topologies. Several modulating techniques have been developed that are applicable to full-bridge vsis. Among them are the pwm (bipolar and unipolar) techniques. Fig. 3: single-phase full-bridge vsi. Table 2: switch states for a full-bridge single-phase vsi Scr with voltage source inverter Commonly called six-step drives, they use scrs (silicon - controlled rectifiers) in their converter front-ends. Scr converters control the dc link voltage by switching on (or "gating") current flow for a portion of the applied sine wave and switching off at the zero crossing points. Unlike mosfet, scrs require control circuits for gate firing. Commutation notches Scr switching or commutation is such that there are brief moments when two phases will both be "on. " this causes what is in effect a momentary short circuit that tends to collapse the line voltage. This shows up as "notches" on the voltage waveform. These notches cause both high vthd and transients. The solution is to place a reactor coil or isolation transformer in series with the drive's front end to clean up both problems. 2 . 2 pulse width modulation 2 . 3 comparison of scr and mosfet based inverter Scrs at high temperature An scr is simply a p-n-p-n structure with a gate terminal (fig. 1). We can break the structure down as back-to-back transistors, one p-n-p, the other n-p-n. With that simplification, we can see that temperature analysis of the bipolar transistor extends logically to the scr structure. It is a well known empirical fact that leakage current approximately doubles with every 10° c increase in temperature. 1 in a bipolar transistor, this increase in leakage is accentuated by the "transistor action" of the device. This can be explained by using an n-p-n transistor as an example. As we increase the temperature, more and more electrons are able to jump the barrier from the emitter to the base. This further biases the base region with respect to the emitter and collector, causing an increase in collector current. In fact, a transistor can be turned on simply by applying high temperature - sufficient leakage current can be generated to trigger the transistor action. This discussion extends to the scr, which is nothing more than two bipolar transistors driving each other. Any effect felt by the bipolar transistor is only magnified when discussing the scr. The effect is not additive, it is multiplicative. There is another temperature-related phenomenon we must point out: as we increase temperature, diode voltage decreases at an approximate rate of 2 mv/°c. 2 therefore, a transistor in the on state will have a tendency to not only stay on at high temperature, but to conduct even more fully; i. e . , the barrier between p- and n-type regions is reduced even more. The result of these two phenomena is that the bipolar transistor has a negative temperature coefficient; the higher the temperature, the higher the collector current at a given base drive. Figure 1: scr structure In a dc application, once the scr is turned on, there is no way of turning it off. Under dc, the scr never experiences the reverse voltage condition across its terminals necessary to prevent conduction. An scr in the off state will tend to turn on and stay on (latch) at high temperatures. But the other will tend to turn on even without an input signal because of the above considerations. Mosfets at high temperature Under no gate bias, a mosfet can be thought of as a pair of back-to-back p-n diodes (fig. 2), from source-to-bulk and bulk-to-drain. Again, we point to the empirical fact that leakage current approximately doubles with every 10° c increase in temperature. However, as we increase temperature, we can think of the leakage current from each diode cancelling each other out, resulting in no net change in current. The dominant temperature-related mechanism for the mosfet is the reduction in carrier mobility. This reduction exists because of the increase of scattering a carrier experiences due to the increased excitation Figure 2: mosfet structure Of lattice sites in silicon at high temperature. The mobility of carriers in the channel of a mosfet behaves according to the following empirical expression3: µ(t) = µ(300 k)/(t/300)a - where t is measured in kelvins and a is 1. 0-1. 5 This relation shows that mobility may decrease as much as 40% for a 100° c temperature increase. Decreased mobility leads directly to an increase in on-resistance. Thus, we can say that the mosfet's on-resistance has a positive temperature coefficient. On the other hand, vt, the threshold voltage of a mosfet, typically decreases by approximately 1. 5 mv/°c. 4 the reason for this effect is that at high temperature, we can expect an increase in thermally generated carriers (leakage). With more n-type carriers available in the conducting channel, we need less of a gate voltage to achieve the same amount of conduction. This effect would tend to counter the decrease in mobility, however, the effect is relatively weak. Vt decreases by approximately 0. 15 v for a 100° c increase in temperature - this is relatively insignificant. We would expect the reduction in mobility to be the dominant mechanism at high temperature. The reduction in vt is further offset by the reduction in the voltage of the gate driver circuit. In sso's mosfet-output devices, the driver consists of 14 series diodes. These diodes generate sufficient voltage to drive the gate of the output mosfet, allowing conduction. As mentioned above in the scr discussion, diode voltage drops at higher temperature. This in turn reduces the drive voltage to the gate of the mosfet. Empirical results have shown that the drive voltage is reduced by approximately the same amount as the reduction in vt, thereby virtually cancelling the effect. A mosfet in the off state will not turn on when exposed only to high temperature. This is because high temperature alone will not be able to create the inversion layer beneath the gate necessary for conduction. A gate voltage is required to do this. Furthermore, a mosfet in the on state will tend to conduct less and ultimately shut off at high temperatures due to the reduction in carrier mobility. 3 . model description A system model showing the physical components of the single-phase vcvsi modeled using matlab. This inverter uses a low-voltage dc bus (24vdc), which is stepped up to 240vac using a step-up transformer (tx). The transformer provides galvanic isolation and is a simple solution for the stepping up of a low-voltage dc bus. The dc bus in the model comprises of the battery (vbatt), lead wire and battery resistance (rbatt), and dc filter capacitor (cdc). The full-bridge uses mosfet switching devices with the full-bridge output filtered using a low-pass lc filter (lf and cf). The inductor filter resistance is represented as rlf with the lc filter-damping resistor being rcf. The load connected to the inverter (zl) is considered arbitrary (linear and/or non-linear). The pwm generator provides the switching signals for the full-bridge with the load voltage rms value used to regulate the load voltage. The rms controller is a simple and standard controller used for inverters only requiring load voltage rms regulation. The dc input and filterthe power of matlab simulink provided a suitable development tool for this application. Working A typical dc-to-ac inverter energized from a 12-volt dc input signal uses a single stage inverter circuit to produce a quasi-sine wave output signal. The peak output signal amplitude for conventional converters of this type often varies over a wide range, e. g . From 110 volts peak to over 200 volts peak for a nominal specified output of 120 volts rms. Third and other odd order harmonics in the output of these devices are usually quite high, and because the duty cycle of the output signal typically varies with load, the harmonic content is difficult to control. Variations in the power required by the load is directly reflected back to the dc source since the conventional dc-to-ac inverter does not include adequate provision for storing energy to meet even short term peak current demands. As a result, the ratio of peak to average current demand on the source can be quite high, causing the overall efficiency of the conventional inverter to be relatively low. Ideally, the input current to a inverter should remain constant during short term variations in the load. Most conventional inverters draw current from the dc source in a quasi-sine wave pattern that is similar to their output voltage waveform. The average current from the source for such devices is significantly higher than if the current were supplied at a substantially constant level. Due to their inefficiencies, the typical inverter tends to be relatively heavy, requiring a larger transformer than would be necessary for providing a given load current with a generally constant supply current. Matlab simulink The design of inverters can be improved using software packages suitable for this application such as matlab simulink and psim. This can provide insight into the inverter performance and allows for the analysis of the design before it is implemented in hardware and software, which can lead to improved performance and reduced development and production costs. In this paper, matlab simulink is used to model a 2kva single-phase full-bridge vsi. This software package is designed for modelling, simulating and analysing dynamic systems. It supports linear and nonlinear systems modelled in continuous time, sampled time or a combination of both. Therefore it is well suited to modelling and simulating inverters and controllers in the analogue and digital domains. Load voltage rms feedback control and openloop control are used to compare the standard matlab simulink and power system blocksets used for the inverter model design. Simulation and experimental results using linear and non-linear loads are used to validate the accuracy of the model developed