No abstract

The frequency response of a pressure transducer is influenced by the natural resonance of the sensor structure, the spatial resolution of the sensor due to its diaphragm size, the sensor packaging, signal conditioning and mounting at the measurement location. The resonance of the sensor and aerodynamically-driven resonances related to the sensor packaging and/or mounting, specifically, can distort dynamic pressure measurements within the range of greatest interest (10Hz–20kHz), typically resulting in erroneous amplification. Historically, correcting for such errors within the frequency response of a pressure transducer or measurement system has been challenging, because such errors are hard to quantify with unknown resonant frequencies and damping factors (quality factors). However, with the ability to fully characterize resonant frequencies that lie within 10Hz–50kHz using a previously demonstrated dynamic pressure characterization methodology, it is possible to apply electrical filtering to substantially extend the flat (0±2dB) frequency response of a transducer before any digital signal conversion. In this work, we present a real-time frequency response compensation scheme that uses electrical filtering to correct for aerodynamically driven packaging or mounting related resonances while at the same time preventing signal distortion caused by the sensor resonances. The compensation extends the useable, flat amplitude bandwidth of the transducer while also correcting the phase response to maintain constant time delay over the extended bandwidth. This real-time frequency response correction scheme can be similarly used to compensate for chip resonances, which can limit the frequency response in applications such as shock and blast testing. A theoretical model of the frequency response correction methodology is presented. We additionally present temperature dependent experimental results that compare the frequency response with and without the correction scheme. These results demonstrate that the usable bandwidth of pressure transducers can be increased when real time, analog frequency response correction is applied. This work shows that if the frequency response of a transducer is well characterized, advanced signal conditioning can be implemented to substantially extend the flat bandwidth of the transducer without changes to the sensor, packaging or mounting.

Accurate static and dynamic pressure measurements in liquids, such as fuel, oil, and hydraulic fluid, are critical to the control and health monitoring of turbomachinery and aerospace systems. This work presents a theoretical and experimental study of the frequency response of pressure transducers and pressure measurement systems in liquid media. First, we theoretically predict the frequency response of pressure transducers based upon a lumped-parameter model. We then present a liquid-based dynamic pressure calibration test apparatus that validates this model by performing several critical measurements. This system first uses a vibrating liquid column to dynamically calibrate and experimentally determine the frequency response of a test pressure transducer, measurement system or geometry. Second, this calibration system experimentally extracts the bulk modulus of the fluid and the percent of entrained and/or dissolved air by volume. Bulk modulus is determined by measuring the speed of sound within the liquid and through static pressure loading while measuring the deflection of the liquid column. Bulk modulus and the entrained/dissolved gas content within the liquid greatly impact the observed frequency response of a pressure transducer or geometry. Gases, such as air, mixed or dissolved into a fluid can add substantial damping to the dynamic response of the fluid measurement system, which makes measurement of the bulk modulus and entrained and/or dissolved air critical for accurate measurement of the frequency response of a system when operating with a liquid media. All experimental results are compared to theoretical predictions.

Silicon micro-machined piezoresistive based pressure transducers are often used to make high frequency dynamic pressure measurements. The spectral or frequency response of these microelectromechanical systems (MEMS) is a function of the natural resonance of the sensor structure, sensor size, sensor packaging, signal conditioning and transducer mounting in the desired measurement location. The advancement of MEMS micro-fabrication, which has reduced sensor size dramatically, and the high elastic modulus of silicon have allowed the natural resonance of these devices to range from 100kHz to several MHz [1]. As a result, packaging and mounting at the point of measurement are the major factors that determine the flat (0dB) frequency response envelope of the transducer, which is typically quantified by a transfer function. The transfer function quantifies the difference both in magnitude and phase between an input signal and a measured signal in the frequency domain. The dynamic response of pressure transducers has historically been estimated via a unit step input in pressure created through a shock tube test that excites the high natural resonance of the chip. Unfortunately, these tests are less effective at accurately quantifying the frequency response of the transducer in the domain of greatest interest (DC-20kHz), specifically the bandwidth over which the response is flat (0dB). In this work, we present a test methodology using a speaker-driven dynamic pressure calibration setup for experimentally determining the transfer function of a pressure transducer from 1–50kHz. The test setup is validated using capacitive-based microphones with claimed flat spectral characteristics well beyond 50kHz. Using this test setup, we present experimental spectral response results for low-pressure miniature MEMS piezoresistive pressure transducers over the frequency range of 1–50kHz and qualitatively compare these results to traditional shock tube tests. The transducers characterized have been manufactured with several different standard sizes and front-end configurations.

No abstract