Advanced Digital Filtering
Pressure sensors do not operate in silence. Fans, blowers, pumps, and flow turbulence generate high-frequency noise that folds into the measurement band. Once aliased, this noise is indistinguishable from real pressure variation and cannot be removed by host-side software. Advanced Digital Filtering addresses this within the sensor by combining a fixed high-speed acquisition rate, integrated analog anti-aliasing, and programmable precision digital filters to deliver a clean, low-noise output across the full range of measurement bandwidths.

What is Advanced Digital Filtering?
A real pressure-sensing environment is not quiet. Fans, blowers, pumps, HVAC equipment, flow-induced turbulence, and electrical noise from power supplies introduce high-frequency pressure components that can be orders of magnitude larger than the low-frequency signals the application is trying to measure. For example, a differential pressure sensor monitoring duct static pressure for a VAV control system may need only 10 Hz of signal bandwidth, while the fan in that duct generates blade-passage frequencies in the hundreds of hertz that are continuously present at the sensor’s mechanical input.
Conventional pressure sensor architectures simply sample this noisy input at whatever rate the host microcontroller requests, typically 10 to 100 Hz for low-bandwidth applications. Without an anti-aliasing filter placed ahead of the ADC, any noise content above half the sampling frequency folds into the measurement band and becomes indistinguishable from real pressure variation. A 300 Hz blade-passage component sampled at 100 Hz appears in the output as an artifact in the 0 to 50 Hz range, exactly where the duct pressure signal of interest resides. Timing jitter in the host’s sampling interval compounds the problem, spreading aliased energy unpredictably across the baseband. Once aliasing has occurred, no downstream filter can remove it; the artifact is mathematically identical to a legitimate low-frequency signal.
The NimbleSense architecture addresses aliasing at the acquisition layer. The sensor samples the analog pressure signal at internal rates from 3 kHz to 12 kHz, comparable to audio acquisition rates, regardless of the host-requested output data rate. At these rates, noise sources in the hundreds of hertz remain well below the Nyquist limit, their spectral identity is preserved at the correct frequencies, and they cannot fold into the baseband. The acquisition rate is deterministic and fixed, eliminating timing-jitter-induced spectral broadening entirely.
In addition to oversampling, the NimbleSense architecture integrates an analog anti-aliasing filter ahead of the ADC. This filter attenuates high-frequency energy before digitization, providing a second line of defense. Together, the high-speed acquisition rate and the integrated analog filter keep high-frequency noise at its true frequency in the signal chain, rather than allowing it to appear as a false low-frequency artifact.
The resulting signal is then processed through programmable, precision, multi-order digital FIR filters within the sensor. The filter bandwidth is selectable to match the application’s requirements, from a few hertz for slow HVAC control loops to several hundred hertz for fast closed-loop control. Because anti-aliasing is handled at the acquisition stage, narrowing the digital filter bandwidth lowers the noise floor without reintroducing aliasing. Engineers can tune the output bandwidth for a specific application via a firmware register write, with no hardware changes and no accuracy penalty. The result is a controlled, predictable noise floor across any selected bandwidth, with more than 10× noise reduction compared with conventional sensors and up to 1000× improvement in the most noise-constrained configurations. Below is a real-world example implemented for one of our customers:


Availability
Advanced digital filtering is available in all NimbleSense pressure sensors.
Advanced Digital Filtering FAQ
Why can’t engineers just apply a digital filter in the host MCU to remove the noise?
Host-side digital filtering can only operate on data accurately captured at the time of sampling. If high-frequency noise has already aliased into the measurement band before or during the ADC conversion, it is mathematically indistinguishable from real pressure variation in the sampled data and cannot be removed by any downstream filter. Effective anti-aliasing must occur before the sample is taken. Advanced Digital Filtering in NimbleSense addresses the problem at the acquisition stage, ahead of the ADC, rather than relying on post-hoc correction in host firmware.
What noise sources does Advanced Digital Filtering address?
Any high-frequency mechanical or electrical noise present in the sensor’s pressure input or analog signal path. In practice, this includes fan and blower blade-passage frequencies, pump ripple, flow-induced turbulence, mains-frequency electrical coupling at 50 or 60 Hz and their harmonics, and vibration from rotating equipment. Sources whose spectral content falls within the application bandwidth and below the selected filter cutoff will appear in the output as intended. Sources above the filter cutoff are attenuated before decimation.
How does selecting the output bandwidth work, and does changing it affect accuracy?
The filter cutoff frequency is selectable via a register write over I2C or SPI, ranging from a few hertz to several hundred hertz, depending on the product configuration. Because anti-aliasing is handled at the acquisition stage, changing the output bandwidth does not introduce aliasing artifacts. Offset, span, linearity, and thermal compensation are governed by the factory calibration and are unaffected by the selected filter bandwidth.
How are the independent calibration sets stored, and does this affect sensor size or power consumption?
The calibration sets are stored within the sensor. This is handled entirely by the NimbleSense architecture and requires no external memory components, no additional PCB area beyond the standard sensor footprint, and no increase in power consumption compared with a conventional single-range sensor.
Does Advanced Digital Filtering introduce output latency?
The multi-order FIR filter introduces a fixed, linear-phase group delay that depends on the filter order and the configured output data rate. This delay is deterministic and can be precisely characterized from the datasheet. For applications where output latency is critical, such as fast closed-loop pressure control, the Closed-Loop Control building block available on CP, HV, ND, and VN Series products is designed to minimize pipeline latency. Advanced Digital Filtering and Closed-Loop Control can operate simultaneously on supported products.
Is the filter bandwidth configurable in the field after deployment?
Yes. The filter bandwidth is set entirely through the sensor’s register interface and requires no hardware access or physical modification of the board. A register write updates the active filter cutoff within one measurement cycle. This allows the same deployed hardware to be reconfigured for a different measurement bandwidth as application requirements evolve, without board redesign or factory recalibration.






