Electronics Guide

Smart Building Sensors

Building automation relies on wireless sensors for temperature, humidity, occupancy, light level, and carbon dioxide concentration. Self-powered building sensors are a mature product category rather than a laboratory demonstration: the EnOcean radio standard, published as ISO/IEC 14543-3-10, defines a wireless short-packet protocol designed specifically around the energy budget of a harvesting node, and equipment built to it is installed in a large and well-documented base of commercial buildings. A typical node needs 10 to 100 microwatts of average power for periodic sensing and transmission, because a short sub-gigahertz telegram costs only tens of microjoules and reporting once per minute therefore averages well under a microwatt for the radio itself. Sensor conditioning, timekeeping, and leakage usually dominate the budget.

Indoor light is the primary source, and its weakness relative to sunlight sets the scale of the problem. Office illumination of 200 to 500 lux corresponds to only about 60 to 150 microwatts per square centimeter of incident optical power from typical lamps, against roughly 100 milliwatts per square centimeter in full sun. Cell chemistry matters more indoors than outdoors: organic and amorphous silicon cells designed for low-level artificial light produce on the order of 5 to 10 microwatts per square centimeter at 200 lux and above 20 microwatts per square centimeter at 600 lux, while crystalline silicon optimized for the solar spectrum falls to a few microwatts per square centimeter or less under the same lamps. Covering a 10 to 100 microwatt load therefore calls for a cell of a few square centimeters, sized with margin for lamp aging, luminaire spacing, and the possibility that the sensor faces away from the light.

Pairing indoor photovoltaics with a thermoelectric generator across the difference between supply air and room air is attractive, since a duct-to-room difference of roughly ten kelvin is common while a building is conditioned, and a small module with adequate heat sinking converts that into tens to a few hundred microwatts. The complementarity is real but narrower than it first appears. Commercial buildings run night setback, so the hours when the lights are off are frequently the same hours when airflow is reduced and the duct gradient collapses. The thermal source cannot generally be relied on to cover the dark period; what carries the node overnight is storage sized against the longest expected unoccupied stretch, typically a weekend or a holiday closure. The thermal channel earns its place by shortening recovery time and by covering deep-plan spaces where lighting is sparse, not by acting as a night-time substitute for light.

Agricultural Monitoring

Precision agriculture distributes sensors for soil moisture, nutrient status, leaf wetness, and microclimate across areas where wired power is impractical and where a service visit to change a battery costs more than the node. Outdoor solar is the anchor source, and it is generous: even a small panel of a few square centimeters yields milliwatts in daylight, orders of magnitude above the node's duty-cycled demand. The design problem is not average energy but the worst-case sequence of overcast, short winter days, soiling, and canopy shading as the crop grows over the sensor.

Soil-to-air thermoelectric harvesting is often proposed as the complement, and it deserves a realistic assessment. The diurnal temperature difference between shallow soil and the air above it is typically only a few kelvin, which yields at best tens of microwatts even with heat exchangers far larger than the electronics they serve. It also reverses sign twice a day as the ground alternately lags and leads the air, so the converter must accept either input polarity and must start from a very low voltage. The channel is a trickle that sustains standby functions; wind-driven micro-turbines or flow harvesting from irrigation lines are the more effective complements where the site permits them.

Seasonal alignment works in the designer's favor. Solar energy peaks in the growing season, when sensing demand and reporting frequency are highest, and falls in winter, when a dormant field can be polled far less often. Adaptive duty cycling that tracks the harvest budget is usually cheaper than adding a second transducer.

Smart City Infrastructure

Urban monitoring for traffic counts, parking occupancy, air quality, and noise benefits from the density of ambient sources in cities, where a single pole may offer light, wind, traffic-induced vibration, and strong radio-frequency fields. Solar remains the primary source for anything mounted above street level; vibration and radio-frequency channels are supplements.

Vibration from passing vehicles is less abundant than intuition suggests. Roadway and pole vibration is concentrated below about twenty hertz at accelerations well under 0.1 g, which is a poor match for resonant harvesters, whose output falls steeply as the tuning frequency drops and whose proof mass must grow to compensate. Frequency up-conversion and non-resonant architectures address this, at the cost of volume. Deployments that require deep shade tolerance, such as sensors set into the roadway or mounted under a deck, are where the second source genuinely decides feasibility.

Fitness and Health Monitors

The human body offers several harvestable sources at once: metabolic heat, the kinetic energy of gait and gesture, and whatever ambient light reaches the device. Body heat is the most consistent. Measured wearable thermoelectric generators produce roughly 15 to 30 microwatts per square centimeter at rest, rising toward 100 microwatts per square centimeter during walking, when forced convection over the cold side steepens the gradient. The wrist is among the weaker sites, because its curvature prevents full contact and its perfusion is lower than the torso or upper arm, and reported figures fall further once boost conversion from the generator's tens of millivolts is accounted for.

Those numbers set the ceiling on ambition. A watch-sized module of about four square centimeters yields tens of microwatts at rest, which comfortably supports a duty-cycled sensor and an occasional Bluetooth Low Energy advertisement but not a continuously lit display or continuous optical heart-rate sensing. Hybrid designs consequently pair thermoelectric and photovoltaic elements for the standby budget and retain a small cell for burst loads. Because body-mounted sources are variable and largely uncorrelated with one another, multi-input power conditioning with independent maximum-power-point tracking per channel is the norm rather than a refinement.

Integration, not conversion efficiency, is usually the binding constraint. The harvester must survive perspiration and skin contact, remain comfortable across a wide range of body shapes and motions, and avoid the thermal insulation that would defeat its own gradient. Flexible and stretchable constructions accommodate body contours, but every layer of encapsulation added for durability adds thermal resistance to the very path the generator depends on.

Smart Textiles

Electronic textiles distribute harvesting across the garment rather than concentrating it in one module, combining fiber-based photovoltaics, triboelectric generators driven by fabric-on-fabric and fabric-on-skin motion, and thermoelectric elements placed where heat flux is highest, such as the upper back and chest. Spreading the transducers over a large area is the strategy: a garment offers hundreds of square centimeters where a wristband offers a few.

Triboelectric elements complicate the power stage. They behave as high-impedance sources producing high voltage at very low current in irregular bursts tied to motion, so they require rectification and charge management quite unlike the low-voltage, quasi-continuous output of a thermoelectric generator. A textile hybrid therefore carries two dissimilar input paths into the same storage element, and the conversion losses on the triboelectric path frequently exceed those on the thermal one.

Laundering is the durability requirement that has no analogue elsewhere in harvesting. Encapsulation must exclude water and detergent, tolerate mechanical agitation and elevated temperature over dozens of standardized wash cycles, and do so without stiffening the fabric to the point of discomfort. Interconnects between distributed elements, which see repeated flexing and stretching at fixed anchor points, remain the most common failure site.

Hearing Aids and Personal Devices

Miniature personal electronics such as hearing aids, smart glasses, and earbuds sit awkwardly against harvesting arithmetic. A hearing aid draws on the order of a milliwatt continuously for its microphones, signal processing, and receiver, which is one to two orders of magnitude above what a few square centimeters of on-body photovoltaic or thermoelectric harvesting can supply. Harvesting in this class therefore extends runtime, trims the size of the cell, or maintains standby and housekeeping functions between charges; it does not eliminate the battery.

The exception is the low-duty-cycle periphery of these products. Charging-case telemetry, fit and wear detection, asset location beacons, and sleep-mode timekeeping all sit in the microwatt range where an ear-to-ambient thermal gradient or a small solar cell on a spectacle temple is genuinely sufficient. Volume overhead is the hard limit: in devices this small, the boost converter, storage capacitor, and protection circuitry can occupy more space than the transducer.

Bridge and Infrastructure Monitoring

Civil structures are monitored on the assumption of decades of service with minimal intervention, and access for battery replacement on a cable stay or a box girder interior is costly enough to justify substantial harvesting hardware. Nodes serving strain gauges, accelerometers, tilt sensors, and corrosion probes span a wide range of demand: a node reporting a temperature and a strain reading every hour lives on tens of microwatts, while one capturing ambient vibration at kilohertz sample rates for modal analysis needs milliwatts while it is awake.

Solar is the primary source wherever the deck or the tower offers exposure, with vibration harvesting as the complement for locations that are permanently shaded, such as bearing seats, girder interiors, and underdeck positions. As with street furniture, traffic-induced deck motion is low in frequency and modest in amplitude, and the harvester must be designed for a broad, drifting spectrum rather than a single resonance, since a bridge's dominant modes shift with temperature, traffic loading, and the structural degradation the system exists to detect.

Storage sizing, not harvesting, usually determines whether such a system survives its first winter. The design case is the longest credible run of low-light, low-traffic conditions, and the storage technology must retain capacity through the full temperature range without a maintenance visit. Supercapacitors tolerate the cycle count and the cold better than most cells but leak, which matters when the recovery source is a trickle; hybrid storage that pairs a supercapacitor buffer with a low-leakage secondary cell is common in this domain.

Aircraft and Aerospace Structures

Airframe monitoring for crack growth, delamination, and bonded-joint integrity is attractive precisely because wiring is what harvesting displaces: on a large aircraft, a wired sensor channel costs far more in installed harness mass and routing than the sensor. Two ambient sources are available in flight. Airframe vibration and acoustic excitation provide broadband mechanical energy, and the fuselage carries a large thermal gradient, since ambient temperature at typical cruise altitudes is around fifty degrees Celsius below zero while the cabin is held near twenty degrees Celsius.

Both sources are available only in flight, so the system must ride through ground time on storage, and both are strongly correlated with the same flight phases, which limits how much genuine complementarity the pairing provides. The dominant constraints are certification rather than physics. Airborne equipment is qualified against RTCA DO-160 environmental conditions and test procedures, covering temperature and altitude, vibration, humidity, and electromagnetic effects, and any structurally mounted device must be shown not to initiate or propagate damage in the structure it is bonded to. Mass budgets are severe enough that a harvester justifies itself only when it displaces more harness mass than it adds.

Pipeline and Industrial Infrastructure

Pipelines, transmission infrastructure, and process plants need distributed monitoring for leak detection, cathodic protection status, corrosion, and vibration across sites that are remote, hazardous, or both. This is the one structural domain where the thermal source is abundant rather than marginal: a hot process line at 80 to 200 degrees Celsius against ambient air presents a gradient of a hundred kelvin or more, and commercial thermoelectric generators clamped to such piping produce output at the watt level, enough to power not merely a sensor but a radio link with real range. Solar complements it for lines that are unheated, buried, or seasonally idle.

Hazardous-area classification shapes every part of the design. Equipment for explosive atmospheres is commonly certified to the intrinsic safety requirements of IEC 60079-11, which caps the stored and switchable energy in the circuit so that no spark or hot surface can ignite the surrounding atmosphere. That cap works directly against harvesting practice, since it limits the size of the storage capacitor or cell the system may carry, and it constrains the surface temperature a thermoelectric hot-side interface may reach. Intrinsically safe harvesting nodes are consequently designed around small, tightly bounded energy reservoirs and correspondingly aggressive duty cycling.

Wildlife and Ecosystem Monitoring

Ecological research deploys instruments in places chosen for their remoteness, where a service visit may cost more than the instrument and may itself disturb the system under study. Hybrid harvesters pairing photovoltaics with wind or stream-flow generators extend deployments from a season to a multi-year record. Avoiding the disposal of primary lithium cells in protected habitat is a genuine secondary benefit, not merely a rhetorical one.

The environment dictates the pairing. Under a closed forest canopy, understory light is a small fraction of the level above it and is heavily filtered, so wind, stream flow, or kinetic sources carry the system while the solar channel contributes only in gaps and in the leaf-off season. Open tundra and grassland invert the arrangement. Practical failure modes in this domain are rarely electrical: animal interference, snow and leaf-litter burial of panels, insect and bird nesting inside turbine housings, and mechanical damage from falling limbs account for more lost deployments than harvester degradation.

Oceanographic and Marine Sensors

Ocean instruments face corrosion, biofouling, pressure, and access costs measured in ship time. Surface and near-surface platforms pair photovoltaics with wave-driven mechanical harvesting, an unusually favorable combination because the sea state that suppresses light through cloud and spray is generally the sea state that raises wave energy. Wave harvesting is also, by ambient-source standards, powerful: surface floats routinely support instrument loads that would be inconceivable from thermal or vibration sources.

Thermal harvesting from ocean stratification is often listed as a third channel and rarely earns a place. The temperature difference across the thermocline is substantial in absolute terms but is distributed over tens of meters of depth, so extracting it requires either a long thermally conducting path or a phase-change working fluid circulated between depths, both of which add drag, cost, and failure modes. Biofouling sets the practical service life of any submerged harvester by fouling optical surfaces within weeks and by adding mass and drag to moving elements, and antifouling coatings, copper shutters, and wipers are as central to the design as the transducer.

Glacier and Polar Research

Polar deployments combine the harshest available conditions with the highest access cost. Solar is not merely reduced in winter but absent for weeks or months at high latitude, so a polar system is defined by what carries it through the dark season. Wind is the usual answer, since it is available year-round and often strongest in winter, making the solar-wind pairing about as strongly complementary as any in practice. Waste heat from co-located instrumentation offers a small supplementary thermal channel where such equipment exists.

Cold degrades the storage element more than the harvester. Conventional lithium-ion chemistry loses substantial usable capacity well above the temperatures these sites reach and must not be charged below freezing without damage, so polar systems favor chemistries rated for low temperature, insulated and self-heated enclosures, or supercapacitor buffers that tolerate the cold directly. Mechanical elements need lubricants specified for the temperature range, and rime icing on turbine blades and panel surfaces is a routine rather than exceptional condition.

Cardiac Pacemakers and Defibrillators

Implantable cardiac devices run from primary lithium cells whose exhaustion eventually requires a replacement procedure, and eliminating that procedure is the clearest clinical case for harvesting anywhere in electronics. The arithmetic is unusually favorable, because pacing is a genuinely microwatt-scale function. A leadless pacemaker consumes only a few microwatts in continuous pacing, with published figures around two microwatts for a commercially available device and roughly four microwatts for continuous pacing at sixty beats per minute.

Harvesters have reached that range. Inertial piezoelectric harvesters tuned to cardiac wall motion, and miniature electromagnetic harvesters intended for endocardial placement, have demonstrated single-digit to low tens of microwatts in animal studies and bench models driven by recorded cardiac motion. The obstacle is no longer raw output but the combination of constraints: the harvester must fit inside a capsule already sized for delivery through a catheter, deliver that output reliably across the full range of heart rates and patient activity levels including the low-motion states where pacing demand is highest, and do so for the device's full service life without maintenance. No harvester has yet been integrated into an approved cardiac device.

Hybrid approaches address the reliability rather than the magnitude problem. Pairing a motion harvester with a thermoelectric element operating across the small core-to-surface gradient, and retaining a reduced primary or rechargeable cell as backing, gives redundancy that no single transducer can offer. Certification follows the ISO 14708 series for active implantable medical devices, with biological evaluation to the ISO 10993 series, and both frameworks demand that any new energy source be characterized for failure behavior as rigorously as for output.

Neural Implants and Brain-Computer Interfaces

Multichannel neural recording and stimulation sit well above the harvesting range. Amplifying, digitizing, and wirelessly transmitting many channels of neural data consumes milliwatts, which is two to three orders of magnitude beyond what physiological sources can supply from the volume available inside a cranial implant. Primary power for these devices comes from inductive or ultrasonic transfer, or from a rechargeable cell replenished by an external coil.

Harvesting nonetheless has a defined role. A physiological source that covers standby, timekeeping, and safety supervision between charging sessions relaxes the recharge interval and permits a smaller receiving coil, which is a meaningful gain when the implant volume is fixed by anatomy. Glucose biofuel cells are the most studied candidate, with reported power densities on the order of tens of microwatts per square centimeter in laboratory conditions; long-term stability in vivo, where enzyme degradation and the foreign-body response both act against the electrode, remains the unresolved obstacle.

Drug Delivery Systems

Implantable and ingestible drug delivery devices split cleanly into two power domains. Actuation, whether by pump, valve, or electrochemical gas generation, is a burst load far beyond harvesting and is supplied from storage. Sensing, dose logging, timekeeping, and short-range telemetry are microwatt functions that a thermoelectric element across the body-to-ambient gradient or a piezoelectric element recovering part of the actuation stroke can plausibly cover.

The value of partial autonomy is disproportionate here, because it is the standby drain over months between refills, rather than the energy of the doses themselves, that frequently sets the size of the cell and therefore the size of the implant. Ingestible devices, which need power only for hours to days, are a different case: their constraint is safety of materials on transit rather than service life, and harvesting from gastric fluid as an electrolyte has been demonstrated at the scale such devices require.

Rotating Machinery Monitoring

Vibration and temperature monitoring of motors, pumps, fans, and gearboxes underpins predictive maintenance, and industrial machinery is the most energy-rich environment ambient harvesting encounters. A motor casing running at 40 to 80 degrees Celsius against plant ambient offers a gradient of tens of kelvin, from which a thermoelectric module with a modest heat sink yields milliwatts. Machine vibration is likewise strong and, in constant-speed equipment, narrowband and predictable, which suits resonant harvesters far better than the low, drifting spectra of civil structures.

That abundance changes what harvesting is for. Rather than merely reporting a scalar temperature once an hour, a self-powered node in this environment can afford to capture a full vibration waveform and compute spectral features locally, transmitting only diagnostic results. Nodes typically feed established industrial wireless networks, principally WirelessHART, standardized as IEC 62591, and ISA100.11a, standardized as IEC 62734, whose mesh routing duty adds a load the energy budget must account for, since a node that relays traffic for its neighbors consumes considerably more than one that only reports.

The pairing of thermal and vibration sources is deliberately redundant rather than complementary: both vanish when the machine stops. That is acceptable, and often desirable, because a stopped machine needs no condition monitoring, and the harvester's collapse is itself a valid state signal. What the second source buys is tolerance to variable-speed operation, where vibration amplitude and frequency swing widely while casing temperature stays comparatively stable.

Process Control Sensors

Chemical processing, refining, and manufacturing plants deploy distributed measurement of temperature, pressure, flow, and composition across areas where conduit installation costs dominate. Process equipment provides steady thermal gradients that make thermoelectric generation the natural primary source, with vibration from pumps and compressors as a secondary channel for unheated lines.

Harvesting is favored in hazardous areas for a reason beyond installation cost: it removes a wiring penetration and its associated ignition risk from the classified zone. The same intrinsic safety limits described for pipeline infrastructure apply, bounding stored energy and constraining surface temperatures, and they must be designed in from the start rather than certified after the fact.

Logistics and Asset Tracking

Supply chain visibility requires trackers on containers, pallets, and increasingly individual packages, moving through environments that change hourly: a lit yard, a dark trailer, a cold store, a warehouse aisle. No single source is present throughout, which is the strongest possible argument for a hybrid front end, and the low cost target for a package-level tag is the strongest possible argument against adding transducers.

The radio-frequency channel in this application is widely misunderstood and deserves precision. Ambient radio-frequency energy from broadcast and cellular infrastructure is very weak, typically nanowatts to about one microwatt per square centimeter even in dense urban settings, and cannot sustain a tracker. The field near a deliberate transmitter is a different matter entirely: an ultra-high-frequency RFID interrogator radiates at the watt level of effective isotropic radiated power and readily powers a passive tag within a few meters. That energy is available only for the seconds a tag spends in a reader portal, so the realistic architecture harvests opportunistic bursts at read points into storage, while vibration from transport and light during outdoor transit supply the between-times trickle. Distinguishing the dedicated field from the ambient one is the difference between a workable design and a specification that cannot be met.

Representative Load Budgets

Selecting sources begins with knowing what the load actually demands. The following average figures are representative planning values for the device classes discussed above, assuming the duty cycling normal to each.

• Battery-free switch or telegram sensor: a few microwatts, dominated by leakage and timekeeping rather than by the radio.
• Duty-cycled environmental or building node: roughly 10 to 100 microwatts.
• Leadless cardiac pacemaker: approximately 2 to 4 microwatts in continuous pacing.
• Condition-monitoring node capturing vibration waveforms: hundreds of microwatts to several milliwatts, depending on sampling rate and analysis burden.
• Hearing aid: on the order of one milliwatt, continuous.
• Multichannel neural implant: milliwatts, beyond the reach of physiological harvesting.

Comparing these against the harvested power densities cited earlier separates the applications where harvesting can supply the whole load from those where it can only supplement a stored source. Both outcomes are legitimate; conflating them is the most common error in application planning.

Power Budget Matching

Duty cycling and event-driven operation reduce average demand by orders of magnitude and are almost always cheaper than adding a transducer. A node that wakes for ten milliseconds each minute spends less than 0.02 percent of its life awake, so sleep-mode current and the quiescent draw of the power conditioning chain, not the active load, determine whether the design closes. Harvested power must exceed consumed power with margin for source degradation, soiling, aging, and the difference between the survey conditions and the worst month of the year.

Cold start deserves separate analysis. A system that has fully depleted its storage must restart from a source that may be delivering millivolts, and the converter's minimum start-up voltage, rather than its steady-state efficiency, decides whether the node ever recovers. Thermoelectric front ends are the usual difficulty here, since tens of millivolts across a small module is a common operating point.

Environmental Characterization

Every deployment site presents a distinct combination of sources with characteristic magnitudes and temporal patterns, and published averages are a poor substitute for measurement. Site surveys with instrumented prototypes over a period long enough to capture the relevant cycles, whether diurnal, weekly, or seasonal, inform both source selection and system sizing. What matters for sizing is not the mean but the tail: the longest interval of simultaneous source scarcity, which sets the storage capacity, and the frequency with which that interval recurs.

Reliability and Lifetime

Applications demanding decades of unattended operation must consider how the system degrades, not merely whether it works when new. Photovoltaic output declines with soiling and encapsulant yellowing, mechanical harvesters accumulate bearing and fatigue damage, thermal interfaces dry out and increase in resistance, and storage capacity fades with cycling and temperature. Sizing against beginning-of-life performance guarantees a mid-life failure.

Redundant sources give graceful degradation, allowing the system to fall back to reduced reporting rates rather than stopping, and the harvested-power trend is itself a diagnostic worth telemetering. Where the deployment is genuinely inaccessible, the design should assume that no calibration, cleaning, or adjustment will ever occur.