**6. On-Track Testing**

The developed DC PETT prototype for the railway applications has been assembled as shown in Figure 4 in Section 2. Due to the direct availability of the 3 kV DC railway network on the railway siding, a number of experiments were conducted with the developed DC PETT prototype mounted on the roof of the EN81 series electric passenger railcar shown in Figure 2 in the introductory section. During the tests, only one drive set of the traction motor powered by the developed 3kV DC PETT was running, which made it possible to obtain measurement results specifically for one complete drive system. The second traction motor of the EN81 series electric passenger railcar was not running during the tests. Due to the limitation of the track length, the maximum achievable speed during the experiments was, however, limited and did not reach the operating speed of 120 km/h given by the manufacturer of the rail vehicle. The route of the test runs with a length of 600 m on the Bydgoszcz-Towarowa railway siding used during on-track testing is shown in Figure 19.

**Figure 19.** The route of test runs with a length of 600 m on the Bydgoszcz-Towarowa railway siding.

Specifications of the DKLBZ 0910-04 type traction motor of the EN81 series electric passenger railcar are listed in Table 6.


**Table 6.** DKLBZ 0910-04 type traction motor specification.

The results of the track test run on 600 m railway siding are shown in Figure 20. The recorded waveforms were saved on a PC connected to the DSP-based MASTER controller card via USB.

Figure 20 illustrates recorded waveforms during accelerating of the train with the DC PETT based asynchronous traction drive to the speed of 45 km/h and immediate braking. Before the connection of the DC PETT to the overhead traction line voltage there is no energy stored in the DC PETT and all DC links are empty. Similar to AC PETT reported in [19], initial charging is performed from the DC side utilizing the startup resistor that is bypassed later on. Moreover, the developed DC PETT has a second mode of pre-charging the intermediary circuits from the on-board battery bank using a set of DC-DC converters. The latter mode enables the DC PETT to be connected to the overhead traction line without any inrush current.

Referring to Figure 20, after the train pantograph is on, the uncontrolled voltage on each of the primary DC links equals the overhead line voltage divided by nine. Then, after starting DC PETT, the primary DC-link voltage is controlled to 520 V and the secondary DC-link voltage is controlled to be equal primary DC-link voltage. During DC PETT operation the input current drawn from the overhead traction line is controlled according to the control scheme shown in Figure 14. In all experiments, the power recovered during braking was transferred back to the traction network. As can be seen from Figure 20, the delivery of braking power of the order of 200 kW to the traction network did not significantly change the voltage of the traction network.

Figure 21 shows the characteristic traction motor stator current and the PWM output voltage waveforms obtained during three different operation modes of the DC PETT. Figure 21a shows the stator current waveform during start-up and the acceleration of the traction motor (50 A/div; 1 s/div). Thanks to the use of high-performance torque control, mentioned in Section 5, the stator current magnitude is limited at the desired value and its amplitude does not exceed 100 A at all times.

As can be seen, there are no undesirable oscillations in the current waveform. Figure 21b shows the stator current waveform during the final braking phase of the train from 15 km/h until the train stops (50 A/div; 1 s/div). Figure 21c shows the stator current and the PWM voltage at the DC PETT output in the steady-state operation of the traction motor (50 A/div; 1 kV/div; 10 ms/div). As it is shown in Figure 21c, the applied three-phase, seven-level CHB topology provides an almost sinusoidal PWM voltage, which has not been demonstrated in any other 3kV DC rail traction inverter so far. By using CHB technology, the instantaneous maximum value of the voltage switched by the SiC MOSFET transistors does not exceed a few hundred volts. At high switching speeds of SiC MOSFET transistors, this will significantly facilitate compliance with the stringent requirements of railway electromagnetic compatibility (EMC) standards, which will be the focus of the planned continuation of the work carried out by the authors.

**Figure 20.** Track test results: Recorded waveforms during start-up of the train with the DC PETT based asynchronous traction drive to a speed of 45 km/h and immediate braking. Top to bottom: (1) train speed; (2) traction motor torque; (3) DC PETT input power; (4) DC PETT input current; (5) traction DC voltage; (6) primary DC bus voltage of a single 4QC-DAB-DC/AC power electronic cell.

The efficiency of the SiC-based 3kV DC PETT prototype has been measured. As in [19], the power consumption of the auxiliary converters supporting the cooling system and the control system has not been included in the efficiency calculations. The plots of efficiencies versus output power are shown in Figure 22.

**Figure 21.** Traction motor stator current and output PWM voltage waveforms obtained during three operation modes of the DC PETT: start-up and acceleration mode (**a**); braking (**b**); phase voltage and phase current of the traction motor steady-state operation (**c**). <sup>310</sup>

**Figure 22.** Measured efficiency versus output power of the three power stages and the entire 3kV DC PETT prototype.

As can be seen from Figure 22, the three-phase seven-level CHB traction inverter (7L CHB), considered separately, characterizes peak efficiency around 99%, which is comparable to the corresponding SiC based MV multilevel CHB inverters constructed from low-voltage SiC MOSFET transistors, presented in the literature [22]. The efficiency of the 4QC input stage is slightly higher than that of the seven-level CHB traction inverter, because when cascaded H-bridges operate with a constant voltage of 3kV DC overhead traction line, actually only two SiC MOSFETs work in each H-bridge. However, thanks to the use of the full H-bridges in the DC PETT input stage instead of half bridges, it is possible that the developed SiC-based DC PETT can also work with an AC input voltage, as is the case with multi-system locomotives. However, the operation of the proposed SiC-based DC PETT in an AC voltage system is beyond the scope of this paper and will be the subject of future publications. As can be seen in Figure 22, the isolation stage ensured by nine DAB DC-DC converters has the greatest impact on the efficiency of the entire SiC-based DC PETT. The SiC-based DC PETT prototype has an efficiency of around 96% in a wide range of output power and the peak efficiency around 96.5%.

#### **7. Conclusions**

The design and development of the SiC-based DC PETT intended for EMUs operated in 3 kV DC rail traction have been presented in this paper. The developed DC PETT has been implemented into the PESA 308 EN81 series electric passenger railcar that operates in Polish regional passenger rail transport. The conducted experimental tests during train runs on the trial confirm the full functionality of the developed device.

As with the MV PETT for AC traction [19], the proposed DC PETT offers a number of advantages that make it very attractive for rolling stock operating in 3 kV DC traction. First of all, in the era of widespread striving to design highly efficient and ultra-quiet drive converters from SiC semiconductor devices, the proposed solution has a number of advantages if one compares it to high-voltage SiC traction inverters with a classic design being currently at the stage of analyzes and preliminary tests. Conventional two-level voltage source traction inverters of the working voltage in the catenary 3 kV DC would contain SiC MOSFET transistors with a voltage blocking of 6 kV, and the conventional three-level inverter voltage would contain SiC MOSFET transistors with a voltage blocking of 3 kV. It is already known that obtaining high voltage switching frequencies SiC MOSFET above 5 kHz is energy inefficient and the management of electromagnetic disturbances at such high switched energies is quite a challenge. The component modules of the proposed DC PETT, in the form of nine 4QC-DAB-DC/AC power electronic cells, are made with the use of low-voltage SiC technology (1200 V). The applied high switching frequency: 30 kHz to 1.2 kV SiC MOSFETs used in DAB DC-DC converters and 20 kHz to 1.2 kV SiC MOSFETs used in SiC MOSFET H-bridges of the

input and output stage, do not cause as significant energy losses as it would be in the case of high voltage (>3 kV) SiC MOSFET technology. The applied high switching frequency allows for favorable elimination of noise from the converter operation. Moreover, the use of multi-level topology made it possible to follow the command voltage from the control system with very high precision and, therefore, enables the application of the high precision control of the traction motor. Moreover, compared to classical topologies, the applied active input stage with the regulator of the current drawn from the overhead line enables the minimization of the input LC filter and, thus, minimization of the total volume and cost, which the authors intend to make the subject of detailed analysis in future publications.

The MFT design path, discussed in detail in the article, shows that the important factors influencing the power density of the developed transformers are the provision of appropriate insulation gaps to ensure galvanic isolation at the level of 9 kV DC and the provision of structural gaps between the windings to obtain the desired transformer leakage inductance. In the case of the traction drive investigated in the paper, with a relatively small power of 325 kVA (500 kVA in peak), the power density of the designed 38 kW MFT was 3.5 kW/dm<sup>3</sup> (≈5 kW/dm3 peak). This allows the authors to reasonably hope that for a higher power MFT, the power density obtained will also be higher. At the present stage, it is difficult to compare the power density of the developed 3kV DC PETT prototype with a built-in lightweight LCL input traction filter with 3 kV DC roof-mounted traction inverters available on the market because the solutions known to the authors have a heavy external traction filter mounted in a separate container, which is not taken into account by manufacturers to estimating the power density of the traction converter. The developed 3kV SiC-based DC PETT prototype, thanks to the built-in 4QC power input stage, is immediately ready for cooperation with the AC traction network in a multi-system EMUs. The proposed modular DC PETT structure, composed of the same repeatable power electronic cells, could ensure lower maintenance costs, short inspection and repair times for potential faults, and thus high availability - required in the rolling stock.

**Author Contributions:** Conceptualization, M.A.; software, J.S.; validation, M.A., J.S.; investigation, M.A., J.S.; writing—original draft preparation, M.A.; writing—review and editing, M.A., J.S.; visualization, M.A.; All authors carried out the theoretical analysis and contributed to writing the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors gratefully acknowledge financial support from the European Union under the Polish Operational Programme Smart Growth, The National Centre for Research and Development, Grant no. POIR.01.02.00-00-0193/16-00. The APC was funded by Gda ´nsk University of Technology.

**Acknowledgments:** The authors acknowledge the commitment and resources provided by MMB Drives Ltd., H.Cegielski-Energocentrum Ltd. and PESA Bydgoszcz S.A. during the research. The authors would like to thank J ˛edrzej Pietryka, Sebastian Giziewski and Patryk Kruk for their valuable support in writing this article.

**Conflicts of Interest:** The authors declare no conflict of interest.
