Energy Harvesting Circuit Design Guide
An energy harvesting circuit takes a source that produces microwatts or milliwatts and turns it into something a sensor node can use. The source is weak, intermittent and has a high internal resistance, so the design is about matching, storing and budgeting rather than about efficiency in the usual sense.
The Sources and Their Behaviour
A photovoltaic cell produces a voltage that is nearly constant and a current proportional to the light. Its output impedance is low in bright light and rises as the light falls, and its maximum power point moves with the illumination.
A thermoelectric generator produces a voltage proportional to the temperature difference across it and has a very low output resistance, which means that a small temperature difference produces a small voltage. A piezoelectric harvester produces bursts of charge from vibration and needs rectification and storage.
An electromagnetic harvester from motion or from a current carrying conductor behaves like a current source in many cases. The common feature of all of them is that the available power is proportional to the source strength and that the voltage at which it is delivered varies widely.
start up voltage and Cold Start
start up voltage is the input level at which the converter begins to operate. A boost converter needs enough voltage to start its oscillator and its control circuit, and below that it does nothing even though the source is producing power.
A cold start from a dead storage element is the difficult case. Many harvesting circuits include a separate low voltage start up path or a mechanical switch to give the first boost, and the datasheet figure for the cold start voltage is separate from the running voltage.
Once running, the circuit often continues to work down to a much lower input voltage, because the control circuit is now powered from the output. That hysteresis between the start up and the running voltage is used deliberately in many designs.
maximum power point and Impedance Matching
maximum power point tracking adjusts the load presented to the source so that it draws power at the point where the product of voltage and current is greatest. For a solar cell that is a characteristic voltage that moves with illumination and temperature.
A simple implementation measures the open circuit voltage periodically and sets the operating point to a fixed fraction of it, usually around seventy to eighty percent. A more accurate one perturbs the operating point and observes the resulting change in power.
For a thermoelectric source the tracking is nearly trivial, because the source resistance is low and a fixed input voltage close to half the open circuit value is adequate. Complexity should match the value of the extra energy recovered.
<img src="https://www.gopcba.com/wp-content/uploads/2023/05/pexels-kateryna-babaieva-2760343.png" alt="Energy harvesting module with a solar cell and a storage capacitor” />
Storage and Its Management
storage capacitor holds the energy between the bursts of harvest and the periods of transmission. A large capacitor has a low leakage but takes a long time to charge, while a small one charges quickly and cannot support a radio transmission.
The storage element is usually a capacitor or a thin film cell rather than a battery, because the number of charge cycles is very large and the energy per cycle is small. A supercapacitor with a low leakage current is the usual choice for a node that must last for years.
The energy needed for one transmission is the design driver. A radio that draws twenty milliamps for ten milliseconds needs two hundred microcoulombs, and the storage has to supply that without dropping below the minimum operating voltage of the radio.
Power Budgeting
The budget starts from the average power the source delivers, which is much lower than its peak. A solar cell indoors may deliver a few tens of microwatts on average, and the node has to work within that.
The node then adds up its own consumption: the sleep current of the processor, the leakage of the storage, the quiescent current of the harvesting circuit and the energy of each measurement and transmission. The quiescent current is often the largest term, and a part that draws a microamp continuously can consume the whole harvest.
The result is the interval between transmissions, which is the number that defines the product. A node that can transmit once every ten minutes is a different product from one that transmits every second, and no amount of circuit cleverness changes the energy balance.
Conversion and Efficiency at Low Power
The efficiency of a converter at microwatts is far below its efficiency at rated power, because the switching and control losses do not scale down. Datasheet efficiency curves are quoted at milliamp loads and are meaningless at the operating point of a harvesting node.
Designing for the actual operating point means measuring the converter at the current it will see, including the long periods at very low output. A converter that is efficient in the datasheet and poor at a few microamps will disappoint.
Where the source voltage is above the required output, a buck converter is used and the same reasoning applies. Some designs use a charge pump for a small voltage step up, which has no inductor and is easy to integrate but a limited input range.
Sensing and Control Overhead
The monitoring of the source and the storage must not consume a significant part of the harvest. A resistor divider across the storage capacitor that draws a microamp continuously is the same order as the entire budget of some nodes.
Switched measurement, using a high value resistor switched in only during the measurement, reduces the overhead to almost nothing. The measurement itself must be short and the resistor large, and the input leakage of the measuring channel then becomes the limit.
Reporting the state of the storage is useful for maintenance. A node that reports its storage voltage alongside the measurement lets the operator see a failing source before the node stops working altogether.
Verification and Faults
Verify the design with a source that can be set to the weakest condition expected, and record the time to charge from empty and the number of transmissions that follows. That measurement answers the only question that matters, which is how often the node can send.
A node that works on the bench and fails in the field usually has a source that is weaker than assumed. Measuring the actual source in the installation for a week settles the question and is far cheaper than redesigning the electronics.
A node that stops working after a few months often has a storage element whose leakage has risen. The release checks that keep such a board consistent are collected in our PCB design release checklist, and the assembly points are listed in judging PCB quality. The leakage of the board surface itself matters at these currents, and the practices that keep it low are described in our guide to conformal coating board protection.
FAQ
Why does my harvesting node not start in dim light? The source is below the cold start voltage of the converter. Use a part with a lower start up voltage or add a kick start path.
Should I use a supercapacitor or a battery? A supercapacitor tolerates a very large number of cycles and has no charge termination problem. A battery stores more energy per unit volume.
How often can the node transmit? Divide the average harvested power by the energy of one transmission. That is the limit, and the quiescent current of the circuit is usually what sets it.
