A sensor product rarely fails because the model was wrong. It fails because the pressure sensor drifts with temperature, because the two sources were never time-aligned, because the analog front end sits next to an antenna, or because nobody defined what "the signal is missing" means. BesTom designs that chain: sensor selection, calibration and compensation, front-end layout, duty-cycled power, data-quality gates, and on-device fusion inference on Rockchip MCU-class and Linux-class platforms.
Four domains, and the same discipline in each: know the error budget before choosing the part.
PPG, ECG, body temperature, blood oxygen. Contact quality and motion artefact dominate signal quality far more than sensor grade. See wearable vital-sign monitoring.
Temperature, pressure, flow, gas, humidity. Cross-sensitivity is the core problem — a flow or pressure reading that ignores media temperature is a reading you cannot act on.
IMU, mmWave presence sensing, ultrasonic ranging, magnetic. Selection turns on whether you need absolute position or only rate of change.
Camera plus IMU for stabilisation, dead reckoning or activity recognition. Time alignment between the two is the whole problem — see Smart Vision.
Wireless mesh and battery nodes reporting into one host. Clock drift across nodes and packet loss are the design constraints, not radio range.
An MCU-class core keeps sensing and decides when the Linux-class SoC wakes. Average current is set here, not by the application processor.
The same product can put feature extraction on a microcontroller and fusion on a Linux SoC. Getting that split right is what makes the power budget achievable.
| Role in the product | Platform class | Why it sits there |
|---|---|---|
| Always-on acquisition & feature extraction | MCU-class — e.g. RK2108D (Cortex-M4F + HiFi3 DSP), as used in the Bestom M08D audio module | Sampling, filtering and wake decisions must run at microamp-to-milliamp average. DSP handles filtering and feature extraction without waking the SoC. |
| Sensor plus camera in one node | RV1106B / RV1106 (Cortex-A7 + MCU, SiP DDR) | The MCU side covers continuous sensing while the A7 side handles light vision — one part instead of two, and no external memory rail to power. |
| Real-time closed-loop control | RK3506 (3× Cortex-A7 + M0) | The M0 co-processor drives real-time I/O (CAN / UART / SPI) while Linux runs the UI and gateway logic. Sensor actuation loops must not wait on Linux scheduling. |
| Audio sensing & acoustic analysis | RK3308 / RK2118 audio platforms | Microphone arrays, PDM and SAI interfaces with an audio-NPU where acoustic features carry the signal. See our RK3308 acoustic AI cases. |
| Multi-modal fusion & inference | RK3576 / RK3588 (6 TOPS NPU) | When several sensor streams must be fused with a learned model, or fused with vision and a local LLM. See Edge AI Host. |
| Wireless sensor mesh | Wi-Fi / BLE mesh companion SoC alongside the host | Node-level radio and clock behaviour are designed separately from the application SoC so mesh firmware can be validated on its own. |
None of these appear in a demo. All of them appear in the field.
| # | Failure mode | How we design against it |
|---|---|---|
| 1 | Cross-sensitivity ignored — a pressure or flow reading is reported as truth while the medium temperature moves underneath it | Characterise the cross-sensitivity matrix on the bench, then compensate in firmware. Ship the calibration procedure, not just the coefficient. |
| 2 | Sources never time-aligned — streams are fused as if they were simultaneous when they are tens of milliseconds apart | Define the timing budget up front: hardware trigger or timestamped capture, and a stated acceptable skew per stream. Validate skew by measurement. |
| 3 | Analog front end next to a radio — everything works on the bench and the noise floor rises the moment the link is active | Small-signal routing, ground partitioning and shielding reviewed together with antenna placement and transmit duty cycle, not as separate tasks. |
| 4 | Long-term drift — accurate in week one, out of tolerance by month six, and no recalibration path exists | Decide the drift budget and the recalibration strategy before part selection. If the product cannot be recalibrated in the field, the sensor grade must be bought with money, not firmware. |
| 5 | No definition of bad data — missing, saturated or contact-lost samples are fed to the model as if they were valid | A data-quality gate before inference: validity flags, range checks, confidence and a defined behaviour for each failure class. What the product does when it does not know matters more than its best case. |
| 6 | Power budget built on active current — the datasheet number looks fine and the battery still dies in weeks | Budget the average, not the active. Duty cycle, wake settling time and false-trigger rate decide the real number; measure it over a full cycle, not a snapshot. |
| Stage | Deliverable | Acceptance criterion |
|---|---|---|
| Feasibility | Sensor selection report with error budget, cross-sensitivity risk list and platform split | Measured data from at least one candidate part in your operating range |
| Hardware | Schematic, PCB, analog front-end and RF coexistence review | Noise floor measured with the radio transmitting, not only idle |
| Firmware | Acquisition layer, filtering, calibration and compensation routines | Compensated output within the agreed tolerance across the full temperature and range envelope |
| Data quality | Validity flags, anomaly handling and degraded-mode behaviour spec | Each failure class is injected deliberately and the product responds as specified |
| Fusion / inference | Feature pipeline, model quantization and deployment | Accuracy held within the agreed threshold after quantization, with the false-positive rate reported |
| Power | Duty-cycle design and measured consumption profile | Average current over a full representative cycle, including wake settling |
| Mass production | Calibration fixture, one-point/two-point test SOP, drift sampling plan | Line-level calibration repeatability recorded and signed off |
Design house since 2008. We are a design service, not a storefront — finished-product volume runs through MeToclaw, the product brand of the same group.
Wearable vital-sign monitoring, acoustic sensing on RK3308, and multi-sensor industrial measurement — documented under client cases where the client allows it.
The always-on layer and the application layer are designed together, so the power split is decided by engineering rather than negotiated between two suppliers.
The M08D module (RK2108D + RK962) exists as a building block when the product needs microphones, DSP and voice in the same node.
Bench characterisation, firmware compensation and production-line calibration are designed as one chain with the same error budget.
Rapid prototyping plus contracted manufacturing, with DFM, thermal and EMC work in the same loop as the sensing design.
If the error budget says a sensor cannot meet the requirement at the target price, we say so before you commit to tooling.
Physiological sensing where contact quality and motion artefact set the signal quality.
Camera design from optical selection to mass production, including vision-inertial fusion.
Microphone arrays and acoustic feature analysis in delivered products.
When fused sensor data needs a local model large enough to reason about it.
Full platform comparison across the Rockchip AIoT range.
Send the measured quantity, required accuracy, operating range and lifecycle — we reply with an error budget and a part recommendation.