Notes from the Test Bench
By Bruce Hofer, Chairman & Co-Founder, Audio Precision
In my last column I talked about some of my experiences as a young engineer encountering corporate policies that discouraged direct contact with customers. This month I would like to recall another true story.
About 10 years ago a certain customer of ours in an Eastern European country was having all sorts of problems trying to measure the frequency response of some filters they had designed. After several unsuccessful rounds of communications with our technical support group, the customer was still very unhappy. Normally our technical support group is very good at resolving problems or answering some of the more difficult questions related to audio measurement. However, this time the situation required my help. Fortunately, I had been planning a trip to Europe, so I arranged to add a stopover in Vienna to meet up with our local distributor who then drove us about 3 hours to a hotel in the city and country of the customer who was having a problem.
The next morning after exchanging pleasantries and the usual small talk with the customer, I was led into their lab and shown the problem. Within 60 seconds of observing their test setup, I realized the nature of the problem and what the customer had been doing wrong. It turns out they were designing filters for a balanced audio circuit (not unlike what is encountered in telephony). However, they were attempting to measure their balanced filter using an unbalanced generator and analyzer configuration. Using all the diplomacy I could muster, I explained the situation on a white board and suggested all they had to do was to change the configuration settings from “unbalanced” to “balanced”. Problem solved! They were incredulous at how quick I was able to understand the problem, so much so that I received a very good dinner that evening.
So what happened here? During all of the back and forth communications before my visit, no one had realized their filter was balanced in nature. The customer hadn’t realized how important that one little piece of information was to resolving their problem.
Sometimes only direct observation of a customer’s problem is necessary for a satisfactory solution. Without the direct interaction of a design engineer and the customer, this problem may never have been solved; and our technical reputation would have suffered. Engineers—take off your white gloves and get out in the field periodically to see how your customers are really using your products!
Output: APx500 3.3 Preview: Acoustic Measurements
Audio Precision is introducing two new acoustic measurements for loudspeaker test. Both measurements address failures commonly tested in production, but R&D engineers evaluating drivers or testing prototypes of powered speakers, soundbars, tablets, smartphones, Bluetooth headsets etc. will also find them useful.
An APx with the rub and buzz & modulated noise options is ideal as a "one stop analyzer" that can test a device's electronics and transducers with a single automated project and a single report at the end.
Rub & Buzz
Rub & Buzz problems are a result of mechanical and/or structural defects in a speaker drivers and finished products due to:
Rub & Buzz is characterized by rapid changes in signal, often heard as clicks, glitches and other artifacts with short transient characteristics that occur as surfaces scrape and collide. Another way to describe these signals is to say that they exhibit a high crest factor, with the unwanted noises represented as a string of impulses superimposed upon the desired signal. These unwanted signals do not represent pure harmonics of the input signal, making them very difficult to detect with spectral analysis tools like FFT or THD+N, and especially annoying to human listeners.
APx Rub & Buzz
The APx Rub & Buzz measurement avoids limitations of spectral analysis alone by detecting high crest factor events in a signal, using a chirp stimulus that is certain to cover all possible frequencies of interest.
APx Rub & Buzz is optimized for use in production environments, delivering results that are fast, repeatable and objective.
How it works:
Results are displayed in a multiple axis chart displaying both crest factor and peak ratio against frequency. The user can set limits on one or both y-axis (crest factor or peak ratio), and set a Fail condition either when one limit is crossed or only when both are exceeded.
With small, powerful active loudspeakers and other integrated audio products becoming increasingly popular, the challenges of speaker/enclosure interaction are greater than ever. Active designs often target higher sound pressure levels and deeper bass extension, and have I/O jacks and controls mounted on the enclosure as possible areas of leakage. The challenge is how to measure issues associated with air leaks, which are outside the normal range of audio test.
APx Modulated Noise
As air moves through a leak, it creates uncorrelated, relatively broadband noise. This noise is in turn modulated by the signal driving the speaker. APx Modulated Noise analyzes air leaks by characterizing them as a type of white noise (turbulence) that is modulated by the stimulus, resulting in a reliable means of detecting enclosure imperfections during design and production phases.
How it works:
A note on environmental factors
Loudspeaker artifacts are typically very low in level, even though they can often be highly audible. Rub and buzz and modulated noise measurements need to be very sensitive to pick up these artifacts, thus a sensitive measurement microphone is required (G.R.A.S. products are all excellent here). However, their very sensitivity makes them susceptible to environmental noise. This should be dealt with by reducing the background noise as much as possible, for instance by using an acoustic chamber, or a sealed box into which the driver faces. The modulated noise measurement in particular is susceptible to environmental factors because it is attempting to measure a low-level noise signature in the presence of other noise. To account for false failures due to environmental noise, APx includes the ability to retry a measurement a specified number of times before declaring a failure.
APx Rub and Buzz and Modulated noise will be available concurrent with the release of APx500 v 3.3 at the end of April and require the new SW-SPK option for all models. Please contact your local AP sales partner for pricing or visit http://www.ap.com/products/apx/loudspeaker for more information.
Sound Advice: How can APx retrieve a serial number from a DUT?
In this use case, a manufacturer wishes to have the serial number of each unit tested during production appear on APx reports. This can be done several ways as part of an automated test sequence.
The simplest way to get a serial number for the DUT in the report is to use the Prompt for Device ID feature at the beginning of the sequence (see screen shot below). Most users would check this box and have the operator scan a barcode of the serial number at the beginning of the sequence. This way the serial number is shown in the report and also gets recorded in the CSV log file every time the sequence is run.
Alternatively, you can add a User Input step to any measurement. This information can also be inserted via barcode scanner and is entered in the report.
Automated String Comparison
Another way to get a string into a report is with the Pass/Fail measurement. In this case, before the Pass/Fail measurement a sequence step is run which causes a string (for example, the software version number) to be written to a text file. The Pass/Fail measurement can then be used to read the text file and compare it to an expected string. In the example below, we have set up the measurement to read the file Version.txt from my “My Documents” folder and compare it to the string “12345”. As you can see, the actual string and the expected string both appear in the report.
Strings returned from applications
The Run External Program step may accept a string that is written to a console window from a Console Application (e.g., a Windows CMD window). We created a console app that writes the string “Hello World”, and configured the Expected Response to be the same as part of a sequence step. The setup is shown below with a page from the report. As shown below, the string is sent to the report and is used to determine the Pass/Fail status.
APx supports both human and automated capture of text strings such as serial numbers. For manual input, either the Prompt for Device ID property or a User Input step in a sequence may be used. If the DUT can be queried by an application that returns the desired string (e.g., a serial number) to APx as part of the test sequence, then each DUT report can automatically contain this data by including this application as an External Program.
Test Results: AP News & Events
AP Seminars in Seoul, Singapore and Penang with Dr. Thomas Kite, VP of Engineering
All audio engineers are welcome to attend. Please contact the people below for more information.
© 2013 Audio Precision, Inc.