Be independent
Hardware and firmware, custom tailored for your needs. Own the technology you use. Keep building on it.
We build custom hardware and firmware around your needs—systems you own and control. We’ll give you the design files and source code so you can maintain and expand your own systems, or you can choose an ongoing support plan.
Hardware and firmware, custom tailored for your needs. Own the technology you use. Keep building on it.
Take your design files and source code with you. Choose who maintains or develops your system.
Components selected around your requirements. Hardware built around what you need.
You set the requirements. We agree on the specifications, deliverables, and physical tests together before building.
We built our engineering process around a powerful combination: vertically integrated development, specialized automation, and direct accountability. Hardware, firmware, and physical validation come together in one coordinated process, powered by a custom AI harness and driven by a human engineer. That integration accelerates development, while our lean operation keeps overhead low. You work directly with the engineer responsible for the finished system.
You establish what your system needs to do and the constraints it must work within. Tell us where it will be used, what it connects to, and what needs to happen when it receives an input.
We work through those requirements with you, including size, power, operating conditions, and compatibility with existing equipment and software. Drawings, interface information, and examples of the current equipment help turn the intended behavior into requirements we can design and test against.
Before design begins, we agree the specifications, deliverables, and physical validation criteria. We define the test conditions, acceptance limits, and evidence needed to demonstrate a passing result.
We also settle prototype quantities, the files and documentation to be delivered, the price, and the deadline before accepting the project. You know what you’ll receive and how we’ll determine whether the work is complete. You pay upfront.
We develop the hardware and firmware, then integrate and bring up the prototype. Circuit design, board layout, and embedded software are developed as parts of the same system.
We check power, connections, physical fit, and operation with the equipment it must work with. We also review component availability, lead times, and sourcing risks, identifying alternatives where possible and checking proposed substitutions against the design.
We run the physical tests agreed in step 2 and record the results. Normal operation and the agreed fault conditions are checked against the same acceptance limits.
The records identify the hardware revision, firmware version, test conditions, measurements, and pass/fail results, so the evidence can be traced to what you receive. The engineer remains responsible for the test plan and acceptance.
If we haven’t completed the agreed work—including passing final physical validation—by the agreed deadline, we’ll refund your payment in full.
You receive working hardware with integrated firmware, design files, source code, manufacturing files, and validation evidence.
The handoff includes editable schematics and board layouts, fabrication and assembly files, a bill of materials, and the delivered firmware and settings. Connection and first-use instructions explain how to put the hardware to work. We walk through the setup and the agreed procedure for restoring the delivered firmware on functioning hardware.
The test records connect the delivered hardware and firmware versions to their measured results. You also receive the agreed documentation and sourcing information, so you or another engineer can understand the design, arrange manufacturing, and develop the next revision.
In this hypothetical example, a customer’s equipment uses several separate electronic modules to read a sensor, control a motor or valve, and communicate with the existing software. The customer needs one custom board to replace those modules while preserving the rest of the equipment’s operation.
The customer’s sensor, motor or valve, and software already perform their intended functions. The problem is the collection of separate electronic modules and cables between them. An integrated replacement brings those functions together without replacing the whole machine.
The proposed board and its firmware read the sensor, send the required control signals, and communicate with the customer’s existing software. Its connections and behavior match the equipment’s requirements, replacing several separate modules with one integrated assembly.
The example test plan applies known sensor signals and measures the board’s response, then checks what happens when power or communication is interrupted. The tests below define correct operation and fault behavior against the requirements agreed with the customer.
In this example, the customer and engineer agree on the requirements below and how each will be checked before design begins.
Example acceptance plan. Carry out the tests at a temperature of 20–30 °C. For normal operation, the host computer must send the board a valid enable command and valid heartbeat messages. The enable command allows the board to operate; heartbeat messages tell it that the host is still communicating.
Measure sensor-input and actuator-command voltages relative to signal ground. Test with board supply voltages of 18, 24, and 30 V and output loads of at least 2 kΩ, except where the tests below specify a supply ramp or short circuit.
After the engineer connects the test fixture, the automated bench applies the specified inputs, captures measurements and waveforms, and checks each result against the limits below. Each test record includes the board revision, firmware version, test conditions, supporting evidence, and pass/fail result. The engineer remains responsible for the test plan and acceptance.
Required relationship: the actuator-command voltage must be twice the sensor-input voltage: Vout = 2 × Vin. Here, Vin is the sensor-input voltage and Vout is the signal sent to the actuator, not a power-supply rail.
Apply: set the board supply to 18, 24, and 30 V in turn. At each supply voltage, apply sensor-input voltages of 0, 1.25, 2.5, 3.75, and 5 V from a source with resistance of 100 Ω or less. Repeat with 2 kΩ and 10 kΩ loads on the actuator-command signal.
Pass: the measured actuator-command voltages are 0, 2.5, 5, 7.5, and 10 V respectively, each within ±50 mV of its target. The sensor voltage reported by firmware is within ±10 mV of the applied sensor-input voltage. At the endpoints, error is measured relative to 0 V and 10 V; a negative output is not required.
Evidence: automatically logged applied sensor-input voltage, measured actuator-command voltage, and firmware-reported sensor voltage for every supply/input/load combination, with pass/fail results checked against the agreed accuracy limits.
Apply: step the sensor-input voltage from 1 to 4 V and back, at each supply voltage with a 2 kΩ load on the actuator-command signal. Hold each input level for at least 100 ms and record the full 100 ms following each edge.
Pass: the actuator-command signal changes from 2 to 8 V and back. Measured from each sensor-input edge, it enters the ±50 mV band around the new target within 10 ms and stays in that band through the end of the 100 ms observation window. It rises no more than 100 mV above the 8 V target on the upward step and falls no more than 100 mV below the 2 V target on the downward step.
Evidence: automatically captured sensor-input and actuator-command waveforms, with settling time and overshoot checked against the limits. Timing starts at the applied sensor-input edge and includes filtering and firmware delay.
Apply: hold the sensor-input voltage at 2.5 V. At each fixed board supply of 18, 24, and 30 V, switch the actuator-command load between 10 kΩ and 2 kΩ. Separately, hold the load at 2 kΩ and ramp the board supply from 18 to 30 V and back at 1 V/ms. Start with no latched faults.
Pass: the actuator-command signal stays between 4.90 and 5.10 V during either disturbance and returns to 4.95–5.05 V within 5 ms after the load switch completes or the supply ramp ends. The processor does not reset or report a new fault during either test.
Evidence: automatically captured actuator-command and board-supply waveforms aligned with the load-switch event or supply ramp, plus reset-counter and firmware event logs. The bench checks voltage limits and recovery time against those event timestamps.
Apply: apply a 5 V sensor input to request a 10 V actuator-command signal, then short that signal output to ground for 1 s at each supply voltage using a controlled fault fixture.
Pass: current through the short stays below 15 mA from 100 µs after short application until the short is removed. Firmware reports the fault within 20 ms of short application. After removal, the actuator-command voltage stays below 100 mV until a fresh valid enable is accepted; within 10 ms of that enable, it returns to the 10 V target within test 1’s ±50 mV limit. Test 1 must then pass again.
Evidence: automatically captured short-circuit current and actuator-command voltage, fault and enable timestamps, and before/after accuracy logs, checked against the limits. Disconnect the actuator command input for this fault test. Capture the first 100 µs of short-circuit current, but do not grade it against the 15 mA limit. Set a separate peak-current and pulse-energy limit from the selected output circuit’s ratings before approving hardware for this fault test.
Apply: send valid enable and heartbeat messages, then messages with invalid checksums, then silence. Only valid heartbeat messages refresh the 500 ms host timeout; enable messages and messages with invalid checksums do not. Maintain a 4 V sensor input and 2 kΩ output load at each supply voltage, so the enabled actuator-command signal starts at 8 V.
Pass: messages with invalid checksums neither change the control state nor extend the timeout. At 500 ms after the last valid heartbeat, declare host loss; by 510 ms after that heartbeat, the actuator-command voltage is below 100 mV and a host-loss fault is latched. Resumed heartbeats alone do not re-enable the output. Require a fresh valid enable after heartbeat communication resumes.
Evidence: automatically captured and decoded RS-485 messages, actuator-command waveforms, and firmware status. The bench checks timeout timing against the last valid heartbeat and records behavior during heartbeat restoration and explicit re-enable.
Apply: with a 4 V sensor input and a 2 kΩ actuator-command load, run 100 power cycles at each supply voltage. Separately, run 100 watchdog trials at each supply voltage: enable the 8 V actuator-command signal, then deliberately stall the firmware control task.
Pass: during startup, the actuator-command voltage remains below 100 mV until a valid enable. Within 100 ms of the control task being stalled, the independent watchdog triggers and the actuator-command voltage falls below 100 mV. Reboot records the reset cause and requires a fresh enable.
Evidence: automatically captured board-supply, reset, and actuator-command waveforms with boot logs and a timestamp marking the induced stall. The bench checks startup voltage, watchdog response time, reset cause, and re-enable behavior. Record each power cycle and each watchdog trial separately; every trial must pass.
In this example, the prototype quantities, files to be delivered, and supporting test evidence have been agreed with the customer before work begins. The handoff below reflects that agreement.
Tell us about your project and what you need it to do.
We use Google Analytics cookies to measure site usage. Details