The invention relates to a DC-DC converter, according to the preamble of claim 1, in particular to a charge pump for high power applications, in particular comprising inductive elements combined with capacitors which allow zero current switching processes and consequently a minimization of the losses.

Efficient and lightweight DC to DC voltage conversion is required in a widespread range of applications which involve the most disparate power levels from small supplies with output powers in the mW order to large MW plants.

To obtain an efficient DC to DC conversion there are several possible topologies. However, the basic principle behind DC-DC conversion is always the same. Any topology presents an input DC link, an output DC link and one (or more) storage element. In a first step some energy is transferred from the input DC Link to the storage element. In a second step energy is transferred from the storage to the output DC link. The sequence is then repeated.

The quasi-totality of high power converters employs inductive elements as storage elements. In very low power applications it is also common to employ capacitors as storage elements often with circuits known as charge pumps.

For high power applications, charge pumps (e.g. : capacitors as storage elements) are practically never used.

Charge pumps allow extremely poor control possibilities concerning the current flow and/or the output voltage. However, the utilization of capacitors as storage elements involves significant benefits in terms of costs weight and manufacturing process. A specific type of DC to DC power converters and a method of conversion is disclosed in the US 6429632 Bl, provided using high-frequency switched capacitors where the switches are implemented by CMOS transistors or diodes on an integrated-circuit chip and using inductors to limit charging current. High efficiency is achieved using inductors to reduce energy losses in circuit capacitors by high frequency switching when inductor current is zero and capacitor voltage is maximized. The high-frequency (100 MHz or greater) operation of the converter circuit permits the use of inductors with a low inductance value on the order of 100 nH (100x10-9 Henrys) capable of fabrication directly on an integrated-circuit (IC) chip. The use of CMOS integrated components allows the entire converter to be formed on a single IC chip, saving significant space within the portable system. Output voltage and current is high enough to permit EEPROM programming. In addition, fluctuations in the output voltage (ripple voltage) are substantially eliminated when several of the converter circuits are used in parallel. While this type of converter is adapted for the use inside integrated circuits and operating with high currents, it is not suitable for high voltages and high power applications as for instance in hybrid or electric vehicles.

The object of the present invention is a charge pump topology suited for high power applications which does not require an accurate output voltage control.

This object is achieved by the characterizing features of independent claim 1. Further embodiments and advantageous variants of the inventive concept are given in the dependent claims, the following specification and the accompanying drawings.

Compared to classical and widespread charge pump topologies (designed for low power applications) the here proposed designs includes inductive elements which allow a zero current switch and a resonant operation of the converter, with the special resonant waveform generated with only one coil and one capacitor. In addition, the necessary winding goods can be reduced considerably, thereby making the new converter design more cost-effective and more lightweight.

The working frequency is typically below 500 kHz, preferably spans the range from about 20 to about 200 kHz. The rated input voltage is typically in the range of about 400 V and can be boosted up to 800 V, allowing for instance to charge 800 V batteries at 400 V stations or allowing the supply of 400 V systems by such 800 V batteries. Typical power values are between 20 kW up to 200 kW, but can even be lower or higher.

The list of reference numerals is part of the disclosure. The drawing figures are described in correlation and jointly. Same reference numerals are used for same parts. Reference numerals with different indices are used for functionally identical or similar parts.

Fig. 1 shows a charge pump circuit which doubles the input voltage, and the corresponding qualitative waveforms in the main components. Fig. 2 depicts a charge pump circuit which halves the input voltage, and the corresponding qualitative waveforms in the main components.

Fig. 3 shows an embodiment of charge pump circuit, which allows a bidirectional energy flow.

Fig. 4 depicts a charge pump circuit with pre-charge elements, which allow significant a reduction of the inrush currents when the device is turned on.

Fig. 5 shows a cascaded embodiment which generates an voltage multiplication by a factor of 1.5 using a two stages concept.

Fig. 6 depicts a single step embodiment which generates a voltage multiplication by a factor of 1.5, and the corresponding waveforms in the main components. Fig . 7 depicts a single step embodiment which generates a voltage multiplication by a factor of 1.5, and the corresponding waveforms in the main components.

Fig. 8 shows a charge pump circuit operating at a reduced frequency to minimize switching and gate drive losses at low load.

Fig. 1 shows a charge pump circuit which doubles the input voltage. By turning on the charging switch 9 the storage capacitor 2 is charged through the inductive element 3. In steady state condition, the process starts when the voltage on the capacitor 9 is slightly slower compared to the input DC- link voltage 1. Due to an initial voltage difference between the input DC- link 1 and the storage capacitor 2 an oscillation process with the inductance 3 takes place. A sinusoidal current flows through the charging switch 9 and the charging diode 7. After half of the oscillation period the diode 7 avoids a current flow in the opposite direction and the oscillation is consequently stopped. The voltage on the storage capacitor remains therefore slightly higher than the voltage of the input DC-link 1. At that moment, after a short "dead time ' the discharging switch 8 is turned on. Due to a voltage difference between the storage capacitor 2 and the upper DC-Link capacitor 5, which is constantly charged at the voltage L n, an oscillation between the storage capacitor 2 and the inductive element 3 takes places. The current flow succeeds through the discharging switch 8 and the discharging diode 6. After half of the oscillation period the diode 6 avoids a current flow in the opposite direction and the oscillation is therefore stopped.

Again, after a short "dead time" the charging switch 9 is turned on and the whole process is repeated. Each of the described switching processes can be considered as zero current transitions, and therefore the switching losses of the described system are very low. The switches are all activated and deactivated, respectively, for about 50% of the duty cycle.

For a correct operation, it is important that the resonance frequency between the components 3 and 2 is equal or large than the switching frequency. Nevertheless, to minimize the dead time, and thus reducing the peak current in the components, the switching frequency should possibly equal the resonance frequency.

In addition, to avoid oscillations on the output voltage, the DC-link capacitors 4, 5 must be significantly larger than the storage capacitor 2.

The whole process described above can be summarized as follows: In a first step, energy is transferred from the bottom DC link capacitor 4 (which is connected to the input DC-Link) to the storage capacitor 2. In a second step, energy is transmitted to the top DC-link capacitor 5. An embodiment which allows to halve the input voltage is shown Fig. 2. The basic principle is very similar to the one described above. The turning on of the charging switch 9 triggers a half period oscillation which transfers energy from the top DC-Link capacitor to the storage capacitor. The activation of the discharge switch 8 triggers a half period oscillation which transfer energy to the bottom DC-link capacitor.

The combination of the two principles described above results in a bidirectional embodiment which is depicted in Fig.3. If the left side is used as input DC-Link, the charge pump works as a voltage doubler. In this case, the active semiconductor components are 8, 9, 7, 6. Instead, if the right side of the figure is used as input DC link, the charge pump works as a voltage divider. In this case the active components are 12, 13, 14, 15. The diode-switch combinations 12-9, 13-8, 14-7, 15-6 must not necessarily be separate components. Especially if MOSFETs are used, it is convenient to employ the body diodes of the semiconductor. The utilization of body diodes has an additional advantage for bidirectional embodiments.

With a bidirectional configuration it is also possible to implement the so called active rectification e.g. activate the switch parallel to the conducting diode to reduce losses. However, it is very important that the switch is turned off before the end of the half period oscillation. If the switch is not deactivated, the diode cannot stop the oscillation after a half of the period and the operation of the circuit is significantly degraded.

Fig. 4 proposes solutions to avoid high inrush currents while the charge pump is turned on. First of all, it is important to guarantee that the two DC link capacitors 4, 5 are charged at the same voltage. This can be guaranteed with two identical large resistors 18. Before any switching event it is also crucial to guarantee a pre-charge in the storage capacitor. This can be guaranteed with a resistor in positions 17 and/or 17a. If a pre-charge resistor is placed in position 17, the capacitor is charged through the diode 7 by the bottom DC-Link capacitance 4. If a pre-charge resistor is placed in position 17a the capacitor is charged through the diode 13 by the top DC- Link capacitor 5.

Alternatively to charging resistors, it is also possible to place highly resistive switches 16 parallel to one or more of the main switches. In this way during the first switching events only the highly resistive paths are activated thus avoiding high currents to charge the capacitors 2, 4, 5. Once the capacitors are charged the system can operate normally and the main switches 8, 9, 14, 15 can be activated.

Another option to avoid inrush currents is to operate one or more of the main switches 8, 9, 14, 15 with very short on pulses until the capacitances of the system 4, 5, 2 are charged.

With a bidirectional topology as for example the one shown in Fig. 3, to balance the voltage between the DC-link capacitors, it is also possible to alternatively work in buck (voltage divider) and boost (voltage doubter) modus. In the case of a voltage divider, some charge is transferred from the top DC-link capacitor 5 to the bottom DC-link capacitor 4. Instead, in the case of a voltage doubler, some charge is transferred from the bottom DC-link capacitor 4)to the top DC-link capacitor 5. Therefore, if the top capacitor 5 voltage it too high a buck modus operation is useful to equalize the voltage between the two DC-link capacitances. Respectively, if the bottom capacitor 4 voltage it too high a boost modus operation is useful to equalize the voltage between the two DC-link capacitances.

A two steps "cascading" charge pump embodiment is disclosed in Fig.5. The voltage of the top DC-iink capacitor 5 is halved thus charging the output capacitor 5a with a half of the input Voltage U_in. In a completely analogous way, using two steps, it is possible to obtain following U_out/U_in ratios: 0.25, 0.75, 1.5.

A fractional voltage ratio can also be obtained with a single step. An embodiment which boosts the voltage by a factor 1.5 is shown in Fig. 6. The charging process succeeds with two in series connected storage capacitances 2 and 2b through the activation of the switches 9 and 19. The average voltage on the single storage capacitance is therefore only 0.5 x UJn. With the activation of 8 and 8b, the discharge process succeeds with parallel connected storage capacitors 2, 2b. For this reason, the upper DC- link capacitor (5) is charged with 0.5 x UJn and consequently the output voltage 11 is 1.5 x UJn.

The serially connected circuit 20, which is in Fig. 6 formed by two LC oscillating circuits 2, 2b, 3, 3b, can be also composed by more than two of these elements in series. In this way, it is possible to obtain lower voltage ratios of 1.33 (if three elements are employed), 1.25 (4 elements), 1.2 (5 elements) and so on.

It is important to notice that the embodiment of Fig. 6 can also work as a voltage doubler (if the switch 19 is constantly on and the switch 8 constantly off), de facto allowing a charge pump with selectable voltage ratio, (in this specific case 2 or 1.5). Depending on the activated switches one and the same circuit allows for a plurality of different voltage ratios.

Another fractional voltage ratio can be obtained with the embodiment shown in Fig. 7 which multiplies the voltage by a factor of 0.33. Also in this case, the charging process succeeds with two in series connected storage capacitances (2 and 2b) through the activation of the switches 9 and 19. The average voltage on the single storage capacitance is therefore only 0.33 x Ujn. With the activation of 8 and 8b the discharge process succeeds with parallel connected storage capacitors 2, 2b. For this reason, the bottom DC- link capacitor 5 is charged with 0.33 x U_in.

The serially connected circuit 19 which is in Fig. 6 composed by two LC oscillating circuits 2, 2b, 3, 3b can be also composed by more than two of these elements in series. In this way it is possible to obtain lower voltage ratios of 0.25 (if three elements are employed), 0.2 (4 elements), 0.166 (5 elements) and so on.

As shown in Fig. 8 the operation of the charge pump is also guaranteed for lower switching frequencies. If the charge pump work at very low load, gate drive and switching losses are dominant. For this reason, it is convenient to reduce the operating frequency. A reduction of the operation frequency can but must not necessarily involve a reduction of the duty cycle. On the other hand, high frequency switching is of advantage at the beginning of the operation of the DC-DC converter to pre-charge the capacitors 2, 4, 5.

To obtain a lower voltage ripple on the DC-link capacitors, it is convenient to design circuits composed by two or more charge pumps (phases) which operate preferably at the same frequency. To minimize the voltage ripple it is wiser to set the phase delay at 360°/n where n is the number of the operating phases.

The circuit in a preferred embodiment relies on the principle of balancing of the top and bottom capacitor 4, 5 and alternating buck and boost operation and comprises two DC link capacitors 4, 5, the sum voltage of both capacitors being the sum of both capacitor voltages. The top and the bottom capacitor 4, 5 have about the same voltage. During boost mode the top capacitor 5 is lightly charged and the bottom capacitor 4 is lightly discharged, while in buck mode the top capacitor 5 is lightly discharged and the bottom capacitor 4 is lightly charged. By alternating the two modi of operation the equality of both capacitor voltages can be assured.

According to a further preferred embodiment, two or more DC capacitors can be switched on each other in the DC link.

List of reference numerals

1 Input DC-Link

2 Storage capacitor

2a Storage capacitor of a second step charge pump circuit

2b Storage capacitor in series

3 Inductive element

3a Inductive element of a second step charge pump circuit

3b Inductive element in series

4 DC-Link bottom capacitor

5 DC-Link top capacitor

6 Diode discharging the storage capacitor

6a Diode discharging the storage capacitor of a second step circuit

6b Diode discharging the storage capacitor connected in parallel 7 Diode charging the storage capacitor

7a Diode charging the storage capacitor of a second step circuit

7b Diode charging the storage capacitor connected in parallel

8 Switch discharging the storage capacitor

8a Switch discharging the storage capacitor of a second step circuit 8b Switch discharging the storage capacitor connected in parallel

9 Switch charging the storage capacitor

9a Switch charging the storage capacitor of a second step circuit

10 Load resistor

11 Output DC-Link, Voltage

12 Diode discharging the storage capacitor for an opposite energy flow

13 Diode charging the storage capacitor for an opposite energy flow

14 Switch discharging the storage capacitor for an opposite energy flow

15 Switch charging the storage capacitor for an opposite energy flow Highly resistive switch for an active capacitor pre-charge Resistor to pre-charge the storage capacitor

Resistor to pre-charge the storage capacitor (for voltage divider) Resistors to balance the DC-Link capacitors

Additional switch ailowing a series connection of storage capacitors Storage capacitor bank connected in series