4.1. Simulation
We adopted two simulations one is used to identify the effect of PRU alone and the other is to identify the PRU function inside of the resonant WPT with class E amplifier. The first simulation is shown in
Figure 13, which applied the circuit parameters as shown in
Table 2. Both the battery capacitor and the stabilization capacitor
are initialized with the voltage 4V. The instability current can be observed in
Figure 14 that the charging current toward he capacitor ramps up quickly after 100us simulation time. The circuit becomes unstable which goes to Mega-watt charging to the battery in 300us simulation time. It was due to the PRU AC-DC output capacitance
(scale 10:1; in gold color line) is charged with high enough voltage to feed into the PRU stabilization capacitance
(in blue color line). The battery charging current (in green line) dramatically rises when the voltage in
reaches above the battery voltage plus the diode threshold voltage, which simultaneously brings up the input current (scale 1000:1; in red line). Both PRU resonant capacitor and the AC voltage divider capacitance voltages rises up due to the current resonance. Taking a closed examine on the phase change of the voltage and current with different color markers on the same circuit as shown in
Figure 15, we zoom into the time duration between 63us to 65us before the power burst was happening. Compared to the time (phase) basis is on the input 5VAC (in purple line), the voltage
(scale 1000:1; in green line) on the primary winding varies from a
phase lead at 63us to nearly
phase lead at 65us. The voltage
(scale
:1; in red line) on the primary winding varies from a
phase lag at 63us to nearly
phase lag at 65us. It is also when the voltage
and the voltage
comes into a
phase, yielding a resonance condition satisfying equation (5) and (6) with the voltage gain
nearly 1000 times.
Although this simulation is unreal in the wireless power transfer as the voltages cannot be as high as 1Mega-volt in real application when most of capacitor endures only thousands of volts, and also the voltage gain
cannot be 1000 for long distance power transfer with low coupling coefficient
, it is still very valuable to understand the resonance mechanism as well as the proper function of the PRU circuit from this simulation example. It can be observed from
Figure 14 that the voltage in the capacitor
is oscillating in a subharmonic frequency around 10kHz. The current ramp is in a sense of exponential growth form which is known as the instability. The battery voltage
is 4V in the simulation because it takes very long simulation time to reach the instability of the current when a larger battery voltage is applied. In the experiments we will use high battery voltage to attain the similar instability current effect.
The second simulation is based on the class E amplifier instead of an AC voltage source, which can easily deliver high power with high voltage in high frequency around 4MHz. The class E amplifier was similar to the boost converter except the output is a LC tank as shown in
Figure 16. In the circuit, all parameters used for PRU are identical to the previous simulation example except for the resonance capacitance
is micro adjusted to 392pF for the resonance and also the stabilization capacitance is omitted from the circuit. The resonance occurs immediately after the circuit is turned on in the simulation, which does not need a stabilization capacitance to assist. The reason why the circuit can some into the resonance much faster than that of using the AC voltage source may be due to the precise matching of the resonant frequency or the class E amplifier is self-adjusting to the resonance. The corresponding results is shown in
Figure 17. The AC-DC output capacitor
first oscillates a while before reaching the voltage higher than the battery voltage, i.e. 4V and soon after the charging current rose up and reaches a 4A level. The charging current carries still some high frequency ripple same as indicated in the wide-band filter result of equation (20). The input current comes after the rising charging current. Since the battery voltage is set to 4V, the output power is calculated as 4A
4V = 16W. There was no negative resistance effect of PRU occurred in previous example because of the steady state exists for the class E amplifier circuit. Taking a closer look in the time duration between 122us to 125 us before the charging current ramped up as shown in
Figure 18, we observed that the ZVS is gradually forming from the wave form of transistor
(scale 100:1; in gold line) when it is compared to the
(in red line) turn-off time. It is also observed that the phase difference between the voltage
(scale 1000: 1, in green line) of the primary winding and the voltage
(scale 100: 1, in green line) of the secondary winding is becoming
which matches the resonance condition in equation (6). The current
on inductor
rises up during the transistor turn on time and the current phase angle
is around
and the corresponding voltage magnification
is also around 3.28 as expected in equation (2) of section 2.1 for the steady state. It shows that the class E amplifier can work just fine with the proposed PRU module. In this simulation, the switching frequency is 1000/242 = 4.1MHz and the duty is 120/242 = 49.6%. The voltage gain
calculated from
Figure 18 is 120V/500V = 0.24 which is achievable in the real resonant WPT application with quite a distance between PTU and PRU in the measure of diameter of the PYU coil.
The control system will become unstable when the resonance with very high voltage gain
can be achieved by adjusting the PRU resonance capacitance.
Figure 19 shows the instability response when the resonance capacitance
is micro adjusted to 292.1pF. With slight variation of the resonance capacitance
, the LC tank changes from the stable circuit become an unstable circuit. An on-line monitoring of the PRU is therefore needed in practical application, the PTU may be controlled to avoid the instability. Both simulations above were using the same coupling coefficient
since the instability phenomenon will still happen with different coupling coefficients only the simulation time to reach instability will increase due to smaller value of coupling coefficient
. When we examine the output wave form (red curve trace) of the charging current flowing into the battery, we observe the subharmonics
around 100kHz and other lower frequencies.
4.2. Experiment
The experimental layout includes a PTU consisting of a class E amplifier and a transmitter coil, a PRU consisting of the PRU circuit and a receiving coil, a DC power supply, and a function generator to control the PTU switch as shown in
Figure 20. The back-iron plate is isolated from all devices. The two coils are placed 50cm away from each other measured from the center of the coils. The PTU switch integrates three 20mm GaN devices connected in parallel which are all D-mode GaN HEMT driven by the charge pump gate drive. The resonant capacitor is in series of a ceramic capacitor with an adjustable capacitor which uses to tune the resonance capacitance to match with the switching frequency of the PTU switch. Each of the 20mm GaN device is able to conduct 6A DC current. In the class E amplifier, the maximum current flowing on the switch is 2.5 times as the DC current read from the power supply when the duty cycle of switching is 50%. Therefore, the maximum power allowed to deliver from the PTU under VDD = 90V input is 3 * 6 / 2.5 * 90 = 600W. Considering the switching loss among the GaN HEMT switches, we have to confine the experiment the maximum output power from the DC power supply within 350W by setting the maximum current output of the DC power supply into 3.5A. The battery string we used in the experiments, consists of two 48V Lithium battery packs and three 12V Lead-acid batteries which are all for different EV application. The batteries were connected in series, which together shows 132V as terminal voltage. The purpose of this experiment is demonstrating the EV battery charging with 1A current using the resonant WPT techniques with the PRU.
Figure 21 shows the oscilloscope screen dump at one instance during the battery charging. The dark blue color trace shows the
of one of the D-mode GaN HEMT switch which is with a threshold voltage of -7V for turning on the switch, which indicates also the duty of switching is 0.5, i.e. there is 50% of the time
is below -7V. The cyan color trace shows the
of one of the D-mode GaN HEMT switch which is with a maximum
as 500V, which is actually as high as 700V before the current limiting is reached. The green trace shows the battery voltage
which is oscillating between 200V to 80V during the battery charging. The purple trace shows the battery current
flowing through the battery string, which shows the average charging current of 992mA and no discharging current flowing through the battery even when the battery voltage is below 132V. From the positive charging current point of view, it may be concluded that the battery voltage oscillation may be due to the measurement of connection wire between the batteries via the high frequency magnetic flux. The red color trace shows the calculation done by the oscilloscope internally, which shows an average of 141.5W going into the battery charging.
Figure 22 shows the power delivery to load (PDL) starting from the time we turn on the power supply until the time when the power supply reaches its current limit 3.5A. It took about 21 seconds for the PTU and PRU to form the perfect resonance. The PDL becomes unstable after 21 seconds, which ramps up until the current limit of the DC power supply is reached. The instability phenomenon agreed with both the simulation results as well as the theory of negative impedance converter via the sub-harmonic oscillation. The subharmonic oscillation had successfully converted the low frequency current back into the transmitting coil as a current feedback to the closed-loop of the resonant WPT. Although the detailed physics are still left behind to be explored, the experimental result matched the simulation results for instability current surge. The current limiting function is activated at the time of 34 seconds, after that the voltage of the DC power supply dropped and the power output will still be increasing for several seconds before it dropped down to 60% of the maximum PDL.
Figure 23 shows the power transfer efficiency (PTE) of the resonant WPT. The PTE grows simultaneously when the instability current shows, which is showing 50% of efficiency in this experiment. In other experiments, there is 65% PTE can be delivered with lower battery voltage which is with lower PDL that extends the time when the current limiting from the DC power supply. The PTE remained at 50% when the current limiting is reached. The PTE rises simultaneously with the PDL rising that is favorable to higher power transfer when more D-mode GaN HEMT transistors can be connected in parallel in the future.