AP - The Recognized Standard In Audio Test

By Bruce Hofer, Chairman & Co-Founder, Audio Precision

AP is buzzing right now as we head into the final stages of production for the APx515’s August 27 ship date. But while the new hardware has taken much of our focus, there are also a couple of significant new features for the entire APx Series in the upcoming APx500 version 2.6 software release.

Derived Results in APx adds the ability to perform math functions on measurement data. We’ve built on the basic computes available in the 2700 and ATS-2, with improvements such as making calculations non-destructive, allowing multiple groups of functions, and of course allowing much easier setup and reporting. I’ve found the ability to compute functions within APx very useful for my own testing, and I’m confident that you’ll see the time savings very quickly too. Our main article below goes into more details.

Also in this issue of Audio.TST is an excerpt from AP’s latest Technote, “Measuring the Sound Pressure Level Emitted by Portable Audio Player Headphones.” Special thanks to G.R.A.S Sound and Vibration for the loan of their KEMAR Manikin.

Finally, I’m very much looking forward to seeing many of you at the Technical University of Denmark next month, as I return for the fourth time to guest-lecture as part of Dr. Michael Andersen’s industrial and PhD course on switch-mode audio power amplifiers. This year, I’m particularly excited to look at the students’ designs with the benefit of our new BW52 ultra-high bandwidth option for the APx525. We should see some very interesting activity in the 300 kHz to 1 MHz range that would previously have gone unmeasured.

Bruce

Derived Results, coming at the end of August in APx500 version 2.6, give the ability to perform all kinds of mathematical functions on collected data. Similar to the Compute functions in other AP software, derived results have the advantages of non-destructive data manipulation and the ability to have derived results of other derived results.

For example, to the Primary Result (the raw measurement data) shown in the first graph below, we add the Derived Result of Smoothing (set to 1/3 octave), and then we add the Derived Result Invert onto the smoothed result. The Primary Result and all the intermediate steps remain available.

Derived Results are organized into five groups, as follows:

Smooth: Smoothing removes fluctuations from the raw data due to testing conditions, in order to make the results more readable. It’s frequently used for acoustical frequency response measurements, to remove small variations due to reflections that make it hard to see the overall response curve. It can also be useful for some noise and distortion measurements, where variations in the signal can prevent getting a steady reading. Smoothing choices range from 1/24 octave to a full octave in bandwidth—1/3 octave is the norm for acoustical measurements. Smoothing can also be applied to time domain results, in which case it is specified in fractions of a second.

Min/Max/Statistics (Geometric mean, mean, standard deviation): These functions can be useful for characterizing performance. Min and Max are especially useful when finding the peak or dip of a filter circuit, or when designing or testing speakers. Standard deviation can be useful for establishing practical limits. By exporting measurement data, and then re-importing multiple data sets into the same graph, the statistical functions can be performed across multiple devices instead of (or as well as) across multiple channels of the same device.

Normalize/Invert: Normalize is often used with frequency response to set the zero reference point at a certain frequency (usually 1 kHz). While similar to the existing Relative Level primary result, the normalize function can be added on top of another function, like Smoothing. The Invert function can be useful for creating an equalized generator signal or an analyzer post-eq curve, for working with equalized circuits and filters.

Compare: Compare is a great tool. It finds the difference from a reference value, or between two channels, or between two data sets. The results are displayed as a new trace (for x/y graphs) or as a bar graph (for single point measurements). It’s useful for comparing overall performance across channels, as well as for checking volume and tone control channel tracking and encode-decode tracking accuracy. By comparing against data stored in an Excel or Audio Precision .adx or .atsx file, it can also be used as a post-EQ function.

Offset: Offset shifts an x/y or bar graph by a selected amount. One application is to remove the fixed insertion loss of a component or circuit stage, in order to measure the output but display the results in terms of the component or circuit's input voltage.

As you can see, the Derived Results functions can be invaluable in converting raw data to a form that is easily and quickly understood and interpreted.

by Joe Begin, Director of Technical Support

This month we release a new Technote on how to measure the sound pressure level developed by portable audio players and their associated headphones, according to British Standard / European Norm 50332. The Technote also shows you how to use the new APx Portable Audio Player / Headphone Test Utility to facilitate making the necessary measurements with an APx500 analyzer. Special thanks go to G.R.A.S. Sound and Vibration for loaning us the KEMAR Manikin Type 45BA used in the Technote. A condensed version follows below. You can download the full Technote at ap.com/display/file/494.

The BS EN 50332 standard is intended to protect users of portable audio devices from exposure to excessive sound levels, and has been adopted by the 31 European CENELEC (European Committee for Electrotechnical Standardization) member countries. It specifies that portable audio players with headphones (or earphones), whether packaged together or bought separately, shall not deliver a maximum sound pressure level exceeding 100 dBA. It does not apply to acoustically open or acoustically closed headphones that are normally used with mains-operated home stereo receivers, nor does it apply to headphones used for medical purposes, or active noise cancelling headphones.

The standard has two parts:

Part 1: General method for “one package equipment,” covers portable audio players and headphones that are packaged together and sold as a unit.

Part 2: Matching of sets with headphones if either or both are offered separately, provides standard procedures for measuring:

• The maximum voltage that can be output by an audio player under standard test conditions, and
• The sensitivity of a set of headphones or earphones—a measure of what sound pressure level they will produce for a given input voltage, under standard conditions.

A Head and Torso Simulator (HATS) is required for measuring the headphone sound levels. A HATS is a special manikin used for sound quality assessment, engineered to have head and torso dimensions representative of a typical adult. For this Technote, we used a KEMAR Manikin Type 45BA, provided courtesy of G.R.A.S. Sound and Vibration.

The HATS is equipped with a pair of removable pinnae (the outer visible section of the ear), molded from a soft rubber-like compound, and mechanical couplers called Occluded Ear Simulators at the locations of the inner ears, to simulate the mechanical impedance of the ear to incoming sound. Measurement microphones are situated at the location of the eardrum.

Measuring the sound pressure level at the eardrum is necessary to account for the interaction between the headphones and the ears. However, to correlate sound pressure levels measured at the eardrum with published data from hearing impairment studies and international standards, the raw data must be converted to free field values. This is accomplished using the free field frequency response of the HATS, which represents the difference, as a function of frequency, between the sound pressure level at the ear simulator microphones when the manikin is present and when it is not. The free-field frequency response of the HATS is sometimes referred to as the Head-related Transfer Function (HRTF). The graph below shows HRTFs for the KEMAR manikin with different styles of pinnae.

BS EN 50332 requires the use of a special test signal called “program simulation noise,” whose spectral content is representative of music and speech. Real music cannot be used for the test signal because music continuously fluctuates in both level and spectral content. Pure tones cannot be used either, because the results would be inaccurate due to the considerable variations in the frequency response of typical headphones.

The program simulation noise can be created by passing pink noise through a special filter network defined in IEC 60268-1. BS EN 50332 adds an additional requirement—that the crest factor of the test signal (the ratio between the instantaneous peak level of the signal and its RMS level) be between 1.80 and 2.2.

We used Matlab to create a digital filter having the same frequency response as the analog filter network described in IEC 60268-1. We then passed pink noise through the filter, adjusted its crest factor using a soft clipping algorithm, and set its level to –10 dBFS.

To make the measurements in Part 1, the program simulation noise is played on the portable player, with volume and tone controls set to maximum. The headphones are mounted on the HATS, and the sound pressure level measured by the two ear simulator microphones is averaged and plotted with 1/3 octave resolution. Then the HRTF (Head-Related Transfer Function) of the HATS is subtracted to derive the free-field response. The resulting curve is A-weighted, and then the overall sound pressure level is calculated in dBA. This process is repeated five times, remounting the headphones on the manikin each time to average out any variation in placement.

Part 2 of BS EN 50332 specifies how to test portable audio players and headphones that are not sold as a package. Unlike packaged sets, the sound pressure level that could be developed in this case cannot be specified in terms of a single metric. Instead, it requires two characteristics—the maximum output voltage (V_m) of the player, and a characteristic called the “Wide Band Characteristic Voltage” for the headphones.

The audio player maximum output voltage (V_m) defined by the standard is the un-weighted true RMS voltage at the load, measured using an averaging time of 30 seconds or more, using the program simulation noise signal. The player’s volume and tone controls are set to maximum, noise reduction (if present) is turned off, and the output is terminated with a resistive load of 32 Ω.

The Wide Band Characteristic Voltage (WBCV) is measured by driving the headphones with an amplifier (output impedance 2 Ω or less) instead of with the portable audio player. WBCV is defined as the un-weighted true RMS voltage measured at the headphones, when the sound pressure level at the HATS is 94 dBA. If the headphones are known to be linear, then the test does not have to be conducted at 94 dBA—instead, the WBCV can be calculated from the measured sound pressure level and RMS voltage using the equation in the standard.

BS EN 50332-2 specifies limits (see table below) for the player maximum output voltage and the headphone Wide Band Characteristic Voltage. Note that headphones having the minimum allowed WBCV of 75 mV driven by a player putting out the maximum allowed V_m of 150 mV would be expected to develop a sound pressure level a factor of two (or 6 dB) higher than the level of 94 dBA at which WBCV is measured. This corresponds with the 100 dBA limit of Part 1 for a matched set of player and headphones.

The last part of this Technote describes how to use the APx Portable Audio Player/Headphone Test Utility. The full Technote may be downloaded at ap.com.

Downloads:

• APx Portable Audio Player / Headphone Test Utility
• Technote 107: Measuring the Sound Pressure Level of Portable Audio Player Headphones

• Running AP Software on Windows 7 (32-bit and 64-bit)
• Driving Relays with the APx Aux Control Out

Audio Precision is very proud to be co-presenter of the 2010 Resolution Awards. Each of the companies represented has demonstrated a strong commitment to audio. We believe that publicly recognizing their efforts will raise the quality bar for the entire pro audio industry and will inspire others towards new innovation and even greater levels of performance. May the best designs win!

back to top >