Acoustic Quality Control

Filtering out background noises and listening through an acoustic lens allows SenSound to listen for defects at near line speed.

"The easiest way to think of this technology is as an acoustic lens," says SenSound LLC (Detroit, MI; president, CEO and co-founder Sergio Mazza.

"The easiest way to think of this technology is as an acoustic lens," says SenSound LLC (Detroit, MI; president, CEO and co-founder Sergio Mazza. A spin-off of Wayne State University's School of Engineering, SenSound's SenQC diagnostic software has the capability of finding stuck valves in engines or determining if a drill is becoming dull. Users get a three-dimensional picture of the sound that shows which products are good, and which aren't. Mazza's biggest challenge, it seems, has less to do with the software and more to do with the Catch-22 of getting OEMs and suppliers to buy into this radical approach to quality control: "They'd rather wait for someone else to prove it works, especially when their contracts specify test methods-even if those methods have proven worthless."

The setup is simple. There is an array of five microphones in an X-shaped brace that is pointed at the object to be measured. There is no need for an enclosure or off-line testing. Products coming down an assembly line can be measured because the software can ignore the background noise.

It's understandable given that the presence of loud background noises and the small deviation between "good" and "bad" products reduces the confidence level in any acoustic measurements. But SenSound's software can effectively eliminate background noise, making it possible to "sanitize" readings and identify products that exceed pre-set parameters. "At its most basic," says Mazza, "we produce an acoustic signature of the plant that allows us to eliminate this noise from our measurements, teach the software what a good product sounds like by measuring representative samples in a quiet environment, and-if there are specific defects we need to detect-instruct the software to recognize them via their particular acoustic signature." Thus, even though the microphones are picking up every sound in their area, the results from the adjusted measurements, claims Mazza, are not only consistent enough to support a statistical quality control system, they correlate well with measurements taken in an anechoic chamber. "The only time we have a problem is when the products that are considered good are so inconsistent that the measurements are all over the place," he says.

Noise, however, is not the only thing the SenSound software can measure in its search for defects. Often it is the absence of noise that is the true measure of whether or not a product is defective. For example, a valve clatters as it closes against the valve seat. A stuck valve does not. "You'd never be able to tell that if there is any background noise in your measurement system," says Mazza. To test the system's capabilities, SenSound and an unnamed OEM mapped engine defects by comparing conventional sound and vibration tests against SenQC. According to Mazza, the conventional test identified 10 of the 15 known defects while the SenSound system identified all 15 reliably-without placing engines in a testing enclosure, making it possible to conduct the test during cold testing on the assembly line when the engine is spun by an electric motor. Moreover, if the electric drive system is fitted with the system, it can detect changes in performance. "Machine condition monitoring is an area with great postential," says Mazza. By listening to the machines used to make the product, for example, it's possible to determine when a drill bit loses its sharpness, and correlate this with field data to replace it before defects are introduced. "Why tear down the offending piece and track the problem back after the fact?," he asks. "Drive this information back into the process, and you will create a complete quality loop that includes every part of the process necessary to create that product."