*Energy harvesting devices such as piezoelectric elements and antennas capture ambient energy, but this energy has to be converted to DC for load and energy storage devices. For this high-efficiency power harvesting circuits are required. This article discusses important design aspects of energy harvesting circuits and their implementation.*

Recent developments in ultra-low-power circuit designs have reduced the power consumption of portable electronic devices and low-power wireless sensors significantly. The power now required has been reduced to just picowatts in the standby mode and nanowatts in the active mode, with the devices operating on an ultra-low duty cycle. These breakthroughs harvest energy from ambient vibrations, radio frequency, static electricity, and other ambient sources, making the best use of energy harvesting solutions.

The energy harvesters make it possible to achieve battery-less operation in low-power devices. The previous article on battery-less solutions discussed some popular energy harvesting techniques that produce enough power to run specific low-power devices. But after capturing the power from source, it needs to be rectified and conditioned before it is fit for use.

A typical ambient energy harvesting circuit has a transducer that translates the ambient energy into electrical energy. In the case of vibrational energy harvesters, these transducers are piezoelectric elements, mechanical oscillators, electromagnetic vibrational harvesters, etc. In the case of RF energy harvesting, the transducer is the antenna. Likewise, there are various other transducers for different energy sources.

The energy converted by the transducers is rectified to generate a DC voltage suitable for load and supercapacitors (for energy storage). The main aim while designing energy harvesting circuits is to maintain high efficiency at every stage. To ensure that, we need an impedance matching circuit between the transducer and the rectifier for maximum power transfer.

After rectification, the voltage may be multiplied by voltage multiplier circuits for compatibility. Now, most of the battery-less low-power devices employ supercapacitors for energy storage. To facilitate such storage, a switch-mode DC-DC converter circuit can be used. Some sophisticated systems use adaptive energy storage systems that compensate for the shortcomings of single energy storage devices.

For ensuring smooth output and high efficiency, feedback circuits can be used to manage the converter circuits. The goal is not only to optimise the power harvested but also to maximise the energy conversion efficiency.

With this, we start discussing the very first stage that ensures maximum power transfer from the transducer to the rectifier circuit—an impedance matching circuit.

Impedance matching

Impedance matching circuit is crucial in optimising the performance of the energy harvesting system. Mismatch in the impedance of transducers and rectifiers degrades the power conversion efficiency of the energy harvester circuit.

Let us consider an example of RF energy harvesting system, where, to increase the conversion efficiency, the designed circuit should resonate at the targeted frequency. A matching circuit has to be designed to match the impedance of the multiplier circuit to the standard 50-ohm antenna for allowing maximum power transfer from the antenna to the circuit.

To match the impedance, it is necessary to first identify the input impedance of the rectifier circuit, which generally consists of nonlinear devices such as diodes and transistors. Small-signal modeling techniques for approximating the behaviour of these nonlinear devices cannot be used due to the large input signal and absence of an input DC bias. The large input signals without any DC bias results in the rectifying device to operate in different regions of operation within the same input cycle. This leads to variation in the input impedance of the rectifier circuit.

For such systems, designing impedance matching circuits is complicated. You need to identify input impedance under all the operating regions such as the subthreshold region, ohmic region, and saturated region according to different DC-bias points, and then design multiple matching circuits accordingly.

However, small variation in input signal guarantees that small-signal parameters of the transistors remain relatively constant, and the calculated input impedance accurately predicts the behavior of the circuit during the normal operation. A fixed impedance matching circuit can then be designed.

There are two popular methods by which impedance matching can be executed, one by fixed LC (inductor-capacitor) impedance matching circuits, and the other by using a transformer to achieve impedance matching. Transformer based matching has an advantage of lower die area and robustness compared to the LC matching method.

Let us first have a look at the LC based matching method. We start with simple topology consisting of an inductor and a capacitor. The difference between the impedances of the rectifier and the transducer dictates the quality factor and the values of capacitor and inductor.

There are four different possible versions of LC circuit, two of them would be low-pass and the other two would be high-pass versions. The low-pass versions are generally employed in RF (radio frequency) circuits since they attenuate harmonics, noise, and other undesired signals. But in our case, we need energy sources from the maximum bandwidth possible. Therefore, the topology shown in Fig. 1 works best in our case.

The circuit in Fig. 1 works when the input impedance of the rectifier is greater than the output impedance of the transducer. For identifying the values of inductor and capacitor, you need both the aforementioned impedances and a desired frequency. Then you need to plugin the numbers in the following set of equations:

**X _{L}=√(R_{S}×(R_{L}-R_{S}))**

**X**

_{C}=(R_{S}×R_{L})/XLwhere RS is the output resistance of the transducer and RL is the input impedance of the rectifier circuit. XL and XC are the reactances from which you can get the values of L and C.

The quality factor of these circuits can only be changed by changing the values of L and C. The quality factor determines the steepness of the frequency response. Low quality factor will be very broadband and will not filter out undesired frequencies, which is desired in our case. There are many other versions of impedance matching circuits like T or pi-network where we can control the quality factor, but in this case low quality factor is desirable and can be easily achieved.

Next, let us take a look at transformer based impedance matching. This technique is preferred when robustness is required. Moreover, a transformer offers a close ideal method for making one impedance look like another. It does this by using the turns ratio (N), which is NS (turns on the secondary side of the transformer) divided by NP (turn on the primary side of the transformer).

The relationship of the impedances can be calculated as:

**Z _{S}÷Z_{P}=(N_{S}÷N_{P})2**

where ZP represents the output impedance of the transducer and ZS represents the input impedance of the rectifier circuit.

Most of the energy harvesting circuits use an inductor based matching circuit as it is easier to implement on-chip, but transformer based matching can be considered for increased reliability and compactness. However, an ultra-efficient transformer will be needed to ensure high efficiency of the circuit.

Let us now take a look at high-efficiency rectifier circuits required to convert noise-like power to DC.

## Rectification

The rectifier circuit is the heart of an energy harvesting circuit. It is responsible for converting the received energy from the transducer to DC voltage, sufficiently enough to feed battery-less electronics. Signal rectification exists in many forms including single-phase and three-phase, controlled (synchronous) and uncontrolled, and full-wave or half-wave. In most cases, non-synchronous, uncontrolled, and single-phase rectification is needed. Full-wave vs half-wave rectification is a decision that can be made by the designer based on losses, expected input voltage, and desired output voltage.

One common method of rectification is Schottky diode based voltage multiplier circuit, which not only rectifies the signal but also amplifies it. Schottky diodes present low-voltage switching, lower switching loss, and high efficiency, so they are preferred for low-power applications.

There are various topologies for voltage multipliers but among them Dickson voltage multiplier is widely used, as it offers a good trade-off between simplicity and performance. Fig. 3 depicts an n-stage Dickson rectifier. The diodes here can be implemented by MOSFETs, and the load resistance represents the input impedance of the circuit to be powered (microcontroller, sensor, etc).

The working of this rectifier is quite simple. Let us consider a single-stage rectifier consisting of D1, D2, C1, and C2. When the input voltage is negative, capacitor C1 stores the charge through diode D1. When the input becomes positive, capacitance C2 stores the charges resulting from the sum of the input voltage and the voltage across C1 via D2. In case of ideal diodes, the voltage across C2 without any load is given as:

**2×(√2VRF,in–Vth)**

The capacitors C1 and C2 can be implemented as stray capacitance on the chip, and the diodes as diode-connected MOSFETs that feature low ‘on’ resistance.

## Adaptive power management

The output of the rectifier can be directly connected to the supercapacitor or the battery for charging. However, the variations of the piezoelectric voltage amplitude at the AC side of a rectifier, while the DC side voltage is retained fixed, can significantly reduce the amount of extracted power from the transducer. Various research suggest that based on the efficiencies achieved, the adaptive control circuits can enhance the efficiency of the rectifier circuit using maximum power point tracking like approach. They describe an adaptive energy harvester circuit from vibrational energy using a piezoelectric element as a transducer.

The system senses voltage from the transducer and at the output of the rectifier for controlling the output DC-DC converter that charges the battery. The aim of the analogue controller is to keep the output voltage of the rectifier and the open circuit voltage constant, when no load is applied.

Also, there might be multiple energy storage elements, one can be primary and the other secondary. Such an approach increases the reliability, because if the primary supercapacitor or battery is completely discharged, the system powers the load from the secondary storage device. Many applications ranging from low-power devices to electric vehicles use multiple supercapacitors/batteries for this kind of adaptive energy storage system.

Thus, if you are looking to implement an energy harvesting circuit on a chip, the above-mentioned circuits sum it up. However, you do not want a dedicated circuit on chip to realise the harvesting circuit in certain applications. There are specialised power management integrated chips (ICs) that integrate rectifiers, ultra-low power LDOs (low dropouts), and DC-DC converters for low-power energy for such applications and for those where integration of energy harvesting circuits in a chip is not required. These ICs, moreover, feature low-power start-up because the power that is harvested is usually very small.

One such power management integrated circuit (PMIC) is e-peas’s AEM40940 RF energy harvesting IC. It can start-up at RF input power from -19.5dBm to 10dBm. It features ultra-low-power boost converter, LDOs, and flexible energy management system that has selectable over-charge and over-discharge protection, suiting both rechargeable batteries and supercapacitors. The IC has provision for two energy storage devices, a primary and a secondary that automatically switch between each other when one of them is exhausted.

The adaptive control circuit involves open-circuit voltage sensing for maximum power point tracking (MPPT) every 0.33 second, and configurable MPPT. Impedance matching is also provided internally.

There are various other PMICs by e-peas intended for such energy-harvesting applications. S6AE10xA from Infineon is a solar-optimised energy harvesting solution and, likewise, there are many other such solutions for low-power management.

## Conclusion

Battery-less solutions are flooding the market, and to support the energy harvesting technologies available today, there is an increasing demand for harvesting circuits that can not only make the power desirable for load but also maintain high efficiency while doing that. To achieve high efficiency, circuits like impedance matching and adaptive control systems are very critical. We have seen the PMICs available in the market that integrate all of the functionality. So, if integration on-chip is not needed, those can be preferred.