AgriTech
AgriTech Hardware Development
Raonebytes develops AgriTech hardware for farms and field operations: soil and crop sensors, irrigation controllers, weather stations, livestock monitors, long-range gateways and farm robotics. Agricultural hardware is deployed once and then largely forgotten about for months, so the design decisions that matter are the ones about power, sealing, radio range and physical abuse, not the dashboard.
The short answer
AgriTech hardware development covers the field sensors, controllers, radios, power systems and enclosures needed to run unattended across a growing season, plus the firmware and backend that turn readings into decisions. A rugged agricultural IoT node or irrigation controller prototype typically costs $8,000 to $20,000 and takes 8 to 14 weeks, with most of the engineering going into power budgeting, weather sealing and long-range connectivity rather than the app.
Common builds
Soil and crop sensing
Irrigation control
Weather stations
Livestock monitoring
Farm robotics and survey drones
What Raonebytes delivers
Rugged sensor nodes
Low-power PCB and enclosure design for soil, canopy and weather sensing, built around a solar and battery budget sized for the worst month of the year rather than the average.
Long-range connectivity
LoRaWAN, NB-IoT, LTE-M, BLE and private sub-GHz links, with gateway placement and link budget modelled against the actual field layout instead of assumed cellular coverage.
Firmware and data
Sampling, calibration handling, local logging through outages, store-and-forward telemetry, alerting and a rollback-safe over-the-air update path that works over a slow rural link.
Field and machine integration
Enclosures, glands, connectors and mounting designed for washdown and weather, plus CAN and ISOBUS-aware interfaces where the device has to sit alongside a tractor or implement.
What makes this hard
The constraints that decide whether a design survives contact with the real world.
Rural connectivity is not what the coverage map says
Farms sit at the edge of cellular cells, and NB-IoT and LTE-M are deployed unevenly across rural areas even where LTE voice coverage exists. LoRaWAN with a farm-owned gateway is often the only link that survives, and choosing between it and cellular is a decision that has to be made from a real site survey rather than a carrier map.
Growing crops break the radio link you validated in spring
A sub-GHz link tested over bare soil degrades badly once a canopy closes over it, because wet foliage attenuates the signal and the crop intrudes into the Fresnel zone. Antenna height, node placement and link margin have to be specified for the tallest, wettest condition of the season, not for the day of installation.
Unattended power across a full season
A node placed in a field has no technician and no mains, so the energy budget has to cover the lowest-insolation weeks with a panel and cell sized accordingly, allowing for panel soiling from dust and bird fouling. Lithium cells also cannot be charged safely below 0 degrees C, which means the charger needs a low-temperature cut-off if the device overwinters outdoors.
Animals, pests and machinery treat hardware as an obstacle
Rodents chew cable jackets and gland seals, insects nest inside vents, and livestock rub against, chew and submerge anything within reach. This is a mechanical design constraint on par with weather sealing: armoured or conduit-run cabling, protected vents, and collar or ear tag housings with no edges an animal can get a purchase on.
Standards and protocols that shape the design
Identifying these early is what keeps them from forcing a redesign later.
- ISO 11783 (ISOBUS)
- The interoperability standard that lets a tractor and an implement from different manufacturers share a control terminal, built on the J1939 CAN application layer running at 250 kbit/s. It defines the universal terminal, task controller and diagnostics behaviour, and the AEF runs the conformance testing and database that the industry treats as the mark of real compatibility.
- IEC 60529 ingress protection
- IP66 covers powerful water jets, which is the realistic test for equipment that gets pressure-washed alongside a machine, and IP67 covers temporary immersion for anything that can end up in standing water or a trough. The rating drives gland selection, seal groove design and vent choice, and it has to be set before the enclosure is laid out.
- ISO 25119 and ISO 18497
- ISO 25119 is the functional safety standard for safety-related control systems on agricultural and forestry machinery, the sector equivalent of ISO 13849, using agricultural performance levels (AgPL). ISO 18497 addresses the safety of highly automated agricultural machines and is the reference point for anything that moves through a field without an operator on board.
- Radio equipment approvals (FCC Part 15, EU RED)
- Sub-GHz operation sits in different bands by region, 902 to 928 MHz under FCC Part 15.247 in the US and 863 to 870 MHz under the EU Radio Equipment Directive, so a single hardware variant rarely covers both. Antenna, matching network and duty cycle behaviour all form part of the approval, which means the radio design has to be frozen before testing rather than tuned afterwards.
- Drone operating rules (FAA Part 107 and Part 137, EASA)
- In the US, survey flights fall under Part 107, while dispensing a substance from an aircraft, which includes spraying, additionally requires a Part 137 agricultural aircraft operator certificate, and aircraft above 55 lb need a separate exemption. In Europe the equivalent sits in the EASA specific category with a SORA risk assessment. These certificates are held by the operator, so we design the aircraft and its logging around the evidence those approvals require.
Typical technology stack
Soil and crop sensing
Capacitive probes are inexpensive and low-power but respond to soil salinity and temperature as well as water, so they need a soil-specific calibration and periodic drift checking. TDR probes infer volumetric water content from the travel time of a pulse and are markedly more stable across soil types, at a higher unit cost that only pays off where absolute accuracy matters.
Connectivity
LoRaWAN for wide-area low-rate links where the farm can own the gateway, NB-IoT or LTE-M where a carrier genuinely covers the site and the operator does not want to run infrastructure, and BLE for handheld commissioning and calibration in the field. Dual-path designs that fall back from cellular to LoRa are worth the cost on sites where coverage is marginal rather than absent.
Positioning
Standalone GNSS is adequate for asset location and field boundaries, but guidance, row-level mapping and autonomous machine work need RTK to reach centimetre-level accuracy. RTK depends on a correction stream from a local base or an NTRIP service, and accuracy degrades as the baseline to that reference grows, so the correction link is part of the system design and not an accessory.
Power
Monocrystalline panels sized against the lowest-insolation month, LiFePO4 or lithium packs chosen for temperature range and cycle life, and MPPT or bq-class harvesting front ends rather than a bare diode. Duty cycle is the real lever: dropping from continuous sampling to a scheduled wake often changes the required panel and cell by an order of magnitude.
Enclosures and sealing
UV-stabilised polycarbonate or glass-filled nylon, EPDM or silicone gaskets, and cable glands rated for the same IP class as the box, since a cheap gland is the usual point of failure. Breathable membrane vents let the enclosure equalise through daily thermal cycling without drawing in moisture, which is what prevents the condensation that otherwise kills sealed field electronics.
Machine and vehicle control
CAN is the backbone in agricultural machinery, with ISOBUS layered on top for tractor and implement interoperability, so a device that has to speak to a machine needs a proper CAN transceiver, isolation and bus loading discipline. For autonomous field platforms we work with ROS-based navigation, LiDAR and camera fusion, and encoder odometry corrected by RTK.
Why this matters
Agricultural hardware is judged on whether it still works in month eight. The failures that matter are unglamorous: a gland that let water in, a panel that was sized on an annual average, a radio link that worked before the crop grew, a cable a rat found interesting. Raonebytes designs the electronics, firmware, power system and enclosure together so those interactions surface during the prototype rather than after a season of deployment. Our autonomous robot platform work covers the chassis, multi-sensor fusion and SLAM navigation side of farm robotics, and our quadcopter UAV build covers GPS waypoint missions, autopilot tuning and return-to-home failsafe behaviour, which are the same foundations survey and spraying aircraft are built on.
Frequently asked questions
What does an AgriTech hardware prototype cost?
A rugged sensor node, irrigation controller or livestock monitor typically runs $8,000 to $20,000 for a working prototype, covering custom PCB design, firmware, enclosure and a telemetry link. A simple enclosed single-sensor device can start near $5,000. Farm robotics and drones sit higher, roughly $15,000 to $35,000 for a drone, $25,000 to $45,000 for a robot, and $40,000 to $75,000 and up for an autonomous field vehicle.
How long does an AgriTech build take?
Most prototypes take 8 to 14 weeks from kickoff to a tested unit. Schematic and layout usually take 3 to 4 weeks, fabrication and assembly 2 to 3 weeks, and firmware plus integration the remainder. Anything involving soil sensor calibration or a season-dependent field trial needs additional calendar time that does not compress, so we plan bench-equivalent tests to keep the build moving.
Should I use LoRaWAN or cellular for a farm deployment?
It depends on coverage and who owns the infrastructure. LoRaWAN with a farm-owned gateway gives predictable range and no per-node data cost, which suits dense sensor deployments across a property. NB-IoT or LTE-M avoids running a gateway but only works where the carrier has genuinely deployed it, and rural NB-IoT coverage is frequently thinner than the published map suggests, so we recommend a site survey before committing.
How long will a solar-powered field sensor run without service?
A properly budgeted node should run for multiple seasons without intervention, but the number depends entirely on duty cycle, latitude and panel siting. We size the panel and cell against the lowest-insolation month with margin for soiling, then set the sampling and transmit schedule to fit that budget rather than the other way round. Realistic figures come from a measured current profile on the actual hardware, not a datasheet sum.
Can you make hardware that talks to tractors and implements?
Yes. Agricultural machinery runs on CAN, with ISOBUS (ISO 11783) layered on top for tractor and implement interoperability including the universal terminal and task controller. We design the CAN interface, isolation and message handling for that environment. Formal ISOBUS compatibility is established through AEF conformance testing, which is a separate certification step the product owner arranges once the design is stable.
How accurate are soil moisture sensors in practice?
Accuracy is dominated by calibration, not by the sensor datasheet. Capacitive probes are affected by soil salinity, texture and temperature, so they need a soil-specific calibration and will drift as salts accumulate around the probe. TDR is substantially more stable across soil types because it measures dielectric permittivity directly, and for most decision-support use the practical answer is to calibrate against gravimetric samples and track relative change rather than trusting an absolute figure.
Related capabilities
Build with Raonebytes
Send your requirements and get a fixed quote, typically within 24 hours. NDA on every project.