The AquaPing is an ultra-low power smart sensor that provides stand-off acoustic water leak detection, even for leaks behind walls. No contact with plumbing is needed. All signal processing and analysis occurs at the edge, so no audio is streamed to the cloud and eavesdropping is impossible.

Principle of Operation

When pressurized water escapes from a seam, crack, or loose fitting on plumbing, broadband high frequency acoustics are emitted. These signals can travel 30 ft or more in free-space and be detected by a remote sensor. This suggests a stand-off method for water leak detection that is substantially different compared to conventional contact moisture sensors and various types of smart flow meters installed on or inside of plumbing lines.

Acoustic leak detection combines the principles of glass breakage and smoke detectors. Glass breakage occurs on a timescale of a few seconds or less, requiring continuous data acquisition at rates approaching 1 kHz or more. Almost all plumbing leaks are persistent, which allows for sub-Hz sampling rates and exceptional battery life.

Glass breakage has the advantage of being generally louder than the ambient environment so that detector sensitivity is not an issue. Small water leaks, especially when separated from the sensor by acoustic barriers such as walls, are likely to be at or below the environmental background. Much higher sensitivity and noise mitigation are essential. This is addressed by mechanical, electronic, and digital filtering along with judicious use of statistical analysis.

The sensor is designed to be deployed as a remote, wireless network node. There are many dozens of wireless protocols, however, so it was decided to make the AquaPing agnostic to the host hardware and thus have the widest possible use cases. It will work with any controller having an I2C interface and 3V3 power bus.

Application & Usage Questions

There are dozens of water leak sensors on the market. Why do we need another one?

There are two approaches to water leak detection currently available: contact moisture sensors and smart flow meters. They work well and can provide protection in many applications. The AquaPing is adept at identifying the onset of very small leaks via acoustic sensing and could be useful in situations where installing existing products may not be easy or convenient. An automated, inexpensive, battery-powered, smart acoustic detector can complement and augment the capabilities of a building monitoring system. Anticipated use cases are in smart homes and buildings, security systems, irrigation, dental and medical offices, critical plumbing installations, insurance loss mitigation, and water conservation. Although design development has targeted water leaks, users are encouraged to experiment with applications that involve pressurized gases such as air, refrigerants, or steam.

This acoustic approach to water leak detection is not capable of identifying all possible water issues. For example, it will not respond to leaking roofs or melting ice. The nature of the pressurized orifice is also important. A smooth hole drilled into a pipe will produce a Laminar flow jet that emits a much weaker acoustic signal compared to a jagged crack or loose fitting at the same backing pressure.

What is the maximum stand-off distance?

Reliable identification of leaks at a distance is a challenging problem that depends on a variety of parameters, including the shape and size of the leak orifice, water pressure, acoustic path, and ambient background noise. Independent third-party verification of stand-off acoustic detection was done with a prototype device at UL (Underwriters Laboratories) in Chicago. They established that a 0.04 gpm simulated leak from an irrigation fogger could be detected at a stand-off distance of 33 feet, which was the maximum separation that could be attained in their plumbing lab. The detection distance will be reduced dramatically when an acoustic barrier (eg. door or wall) is in the path between leak and sensor.

How can the sensor reliably identify leaks behind walls?

This capability was developed through a series of experiments with simulated water leaks placed inside a drywall test structure. Best results are with the sensor directly on the wall and the microphone facing inward. An acoustic access hole to the interior is not necessary. Documentation of these tests can be found on our GitHub page. Experiments with walls of different material composition and thickness are described here.

Why is it necessary for the user/controller to set some of the sensing parameters?

Acoustic environments vary a lot. Data acquisition and analysis should account for different and changing conditions. Consider the kitchen in a restaurant, for example. When the restaurant is closed, the kitchen is nominally quiet and sensitivity can be set at maximum. During business hours, noise from operating activities will mask leak signals and dramatically reduce sensitivity. It may make sense to disable the device in this situation to prevent false triggers. Instructions on how to configure the sensor can be found in the User Manual.

Can this design be integrated with an always-on voice recognition system?

No. The sensor is constructed to measure acoustic energy in the frequency range outside of normal conversations. A key design goal was to make eavesdropping on conversations physically impossible, thus assuring privacy. Operating at high frequencies is also essential for highest sensitivity and reliability. An open-source design allows verification that the hardware cannot be used for eavesdropping.

Features & Specifications

• Texas Instruments MSP430FR5994 MCU performs FFT spectral analysis > 10x faster/efficient than ARM-Cortex M0+
• Direct Memory Access manages data without CPU intervention
• Operational voltage range: 1.8–3.6 V
• Stand-off acoustic detection of pressurized water leaks – even behind walls
• All analysis is processed locally. No audio streamed to cloud.
• Only analyzes frequencies outside of human conversations, making eavesdropping impossible
• Two-way communication with host controller via I2C.
• Sensor is highly configurable

Comparisons

Open Source

This project is fully open source and welcomes your input. You can find all hardware files, firmware, and docs in our GitHub repository. Also available is an example sketch and documentation for interfacing with an ESP8266 master controller using the Arduino IDE.

Support & Documentation

• 9-page user guide with design description and I2C interfacing instructions (pdf)
• Firmware repository for the MSP430FR5994 MCU
• KiCAD repository with schematic and layout files
• Example firmware for use with the an ESP8266 WiFi module and ThingSpeak IOT platform
• Two page report describing tests with simulated water leaks behind drywall (pdf)
• One page engineering brochure (pdf)

Manufacturing Process

The AquaPing is composed of 43 total electronic components, of which 29 are unique. Everything except the interfacing header are surface-mount and all can be sourced from Mouser Electronics, which is Crowd Supply’s parent company. The PCBs are panelized for efficient pick-and-place assembly. A small rectangular plastic structure is glued to the assembled PCB to enclose the MEMS microphone and form a Helmholtz resonator, which is visible in photos of the assembled PCB. This part can be inexpensively 3D printed using a variety of local sources. Updates and modifications to the firmware can be accomplished by the user with either: 1) a TI MSP430 Launchpad connected to the Test and Reset terminals using the Spy-bi-Wire protocol or 2) an Olimex MSP430-BSL to download pre-compiled binaries via the I2C interface.