Notes from the Test Bench
By Bruce Hofer, Chairman & Co-Founder, Audio Precision
After forty years watching the lifecycle of product development and the ups and down of the economy in general, I firmly believe the test and measurement industry is a leading indicator for business. We see R&D orders first, and then production test orders heat up. A little while later, products are on the shelves—meanwhile, engineers are already working on the next big thing.
With that in mind, I’m happy to report that 2011 is looking good for the audio industry. In 2010, we saw a strong uptick in both R&D and production applications, which tells me that confidence is up and new opportunities are presenting themselves to our customers all over the world. No big surprise, but it does serve as a reminder of just how globally dispersed, yet totally interdependent, the audio industry is. Of course, no matter where you are, it’s more than likely that you’ll be working with Audio Precision gear, so at least some small part of your bench will always feel like home.
Output: Measuring Sound Pressure Level with APx
The level of acoustic noise or sound in the environment is typically measured with a handheld sound level meter. In the following, we explain some of the theory behind such measurements, and how to make them instead using an Audio Precision APx500 Series analyzer. The article below is an excerpt from Technote 113: Measuring Sound Pressure Level with APx, which may be downloaded at AP.com. The new APx Sound Level Meter Utility is necessary to make the measurements and may be downloaded as well.
Graphs shown above:
a) Pulsating alarm SPL measured at alarm speaker terminals.
b) Pulsating alarm SPL as picked up by the measurement microphone.
c) Pulsating alarm SPL from "b" above, after being A‑weighted and exponentially averaged by the APx Sound Level Meter Utility.
Using an APx500 Series analyzer to measure the level of acoustic noise or sound in the environment, instead of a handheld sound level meter, allows you to perform spectrum analysis and visualize the character of the sound signal over time. The latter is especially valuable when measuring non‑steady‑state sounds, such as a pulsating alarm.
To make the measurements described below will require an APx500 Series audio analyzer, a measurement microphone (such as the MMK‑2 available from Audio Precision), and the APx Sound Level Meter Utility.
Sound Pressure Measurement
A sound level meter (SLM) is an electronic instrument used to measure sound pressure levels, and contains a measurement microphone, preamplifier, level detectors, one or more frequency weighting filters, and a display. Typically, an SLM is a self-contained hand-held instrument. However, an SLM could also be comprised of separate components in one or more enclosures that work together as a system. The international standard governing sound level meters is IEC 61672, "Electroacoustics – Sound Level Meters," with Part 1 covering specifications. The US equivalent of IEC 61672-1 is ANSI S1.4, "Specification for Sound Level Meters."
Sound Pressure Level
Sound levels are typically expressed in decibels (dB) relative to a reference sound pressure level of 2.0 x 10-5 Pascal (or 20 µPa, where µ is the SI prefix for "micro" [x 10-6]). The reference sound level of 20 µPa, which is agreed to by international standards, corresponds approximately to the threshold of hearing in healthy human subjects, when the sound is at frequencies in the middle of the audio range.
Sound levels are usually measured with a frequency weighting filter known as A‑weighting. This filter provides significant attenuation at low and high frequencies, with a slight gain in the mid-frequency portion of the audible spectrum, as shown in Figure 1.
Figure 1 A weighting filter frequency response.
A-weighting is used in an attempt to provide sound level metrics which correlate with perceived loudness in humans, who are most sensitive to sounds in the mid-frequency range. When measured with A‑weighting, sound pressure levels in decibels are usually denoted as dB(A) or simply dBA.
Historically, A‑weighting originated as a result of pioneering work by Fletcher and Munson in 1933 on how variations in the level and frequency of sound affect perceived loudness. Based on experiments with pure tones presented to subjects through headphones, they produced a set of equal-loudness curves (Figure 2). Three years later, these curves were used in the first American standard for sound level meters developed by the Acoustical Society of America. The curve labeled 40 phon in Figure 2 formed the basis of the A‑weighting curve, which was later adopted by the International Standards Organization and is still in use today. The phon is a unit of perceived loudness level for pure tones.
Figure 2 Equal-loudness contours from Fletcher– Munson (blue) and the later ISO 226:2003 revision (red).
During the 50+ years since its widespread adoption, various researchers have pointed out problems associated with using A‑weighted sound pressure level measurements as the basis of studies in noise control, prevention of noise-induced hearing loss, and noise nuisance. Nevertheless, A‑weighted sound pressure level is almost universally used in noise regulations around the world. As a result, A‑weighting is a mandatory requirement for sound level meters in IEC 60261-1, whereas other frequency weightings are optional.
An important distinction for this discussion is the one between steady and non-steady sounds. Steady sounds are those which do not vary in level or spectral content with time. Examples include pure tones, or a white noise signal at a constant level. Non-steady sounds are those that vary in level or frequency content, or both—for example, a cymbal crashing, music, speech, or a pulsing alarm signal.
To measure non-steady sound levels, sound level meters use one or more different types of averaging. A conventional sound level meter uses what is referred to as exponential time weighting. This is essentially a moving average process, where samples of the RMS level are weighted exponentially, such that the most recent sample has a greater influence on the indicated level than previous samples. Exponential averaging is equivalent to a simple RC filter. Standard sound level meters include exponential time weighting with two RC time constants: Fast (F) time weighting has a time constant of 125 milliseconds, and Slow (S) time weighting has a time constant of 1 second.
The time constant represents the time that it takes for the step response of the system to reach 37% of its final indicated value. For example, when measuring a steady signal with exponential time weighting, if the signal is suddenly switched off, the indicated level will decay exponentially and approach zero asymptotically. With F time weighting, the indicated level would decrease to 37% of the original level in 125 milliseconds. With S time weighting, the same decay would take 1 second. Sound pressure levels measured with exponential weighting are often designated LF or LS, depending on the time constant used, and LAF or LAS when measured with an A‑weighting filter.
Another type of averaging used in some sound level meters is integrating-averaging. In this case, the meter integrates the data over the measurement time to determine a quantity called the time-average sound level in IEC 61672. This is commonly known as the equivalent continuous sound level (Leq) and is designated as Leq(t). The t in parentheses represents the averaging time, which in practice could vary from a fraction of a second to hours or even days. Leq measurements have the advantage that short duration measurements can be combined to determine the Leq for a longer averaging time. For example, if you measure and record Leq(1s) continuously for a 24-hour period, you can combine these short term measurements to determine the Leq(24hr). When an A‑weighting filter is used to measure Leq, the results are specified as LAeq(t).
Figure 3 Microphone calibration.
Measuring sound pressure level in APx500
The units of sound pressure level, dB re 20 µPa, are sometimes referred to simply as dBSPL. In APx500, there are two independent reference levels (dBSPL1 and dBSPL2) that can be used to specify the sensitivity of measurement microphones, so that measured results can be displayed directly in dBSPL.
The Microphone Calibration window (Figure 3) can be accessed either from the Set dBSPL button in the Reference Levels measurement, or from the Mic Calibration button in the Acoustic Response measurement. You can either enter the sensitivity of the microphone in mV/Pa as printed on its calibration certificate (94 dBSPL is equal to 1 Pa), or use a microphone calibrator to set the dBSPL reference directly. The APx500 context sensitive help explains how to do this more fully.
Figure 4 APx Noise measurement configured to measure A‑weighted dBSPL.
For steady sounds, the APx500 Noise measurement (Figure 4) can be used to measure sound pressure level directly in dBSPL, once the microphone sensitivity has been entered as described above. To measure A‑weighted sound pressure levels, simply set the Filter control to A‑wt (2 – 20 kHz. If you wish, you can change the graph title so that it is obvious that measured noise levels are A‑weighted.
When troubleshooting noise problems, the ability to measure the noise spectrum can be a powerful analysis tool. Figure 5 shows an FFT noise spectrum measured with the APx500 Signal Analyzer measurement while the microphone was positioned close to the ventilation port of a PC with a noisy fan. APx500 offers additional tools in the form of Derived Measurement Results, which can be used for such things as fractional octave smoothing, finding minimum and maximum levels, and more.
Figure 5 APx FFT spectrum of an acoustic noise signal.
It should be noted that virtually all measurements in APx500 are derived from FFT analysis of measured waveforms. In the case of the Noise measurement, the system acquires a waveform for a specified acquisition time of up to 5 seconds, performs an FFT of the entire acquisition, applies the selected filters in the frequency domain, and sums the FFT bins to determine the RMS level. Hence, the result is actually Leq(t), where "t" is the length of the acquisition.
While FFT-based measurements are preferable for typical audio applications, where the audio analyzer's generator is used to stimulate a device under test, this approach can be problematic for measurement of non-steady sound levels. This is because for non-stationary or transient signals, results of the FFT analysis depend strongly on the portion of the varying signal that is acquired and analyzed. As a result, there is no way to accurately reproduce with FFT analysis a time domain measurement like LAF of a time-varying signal.
This article continues in Technote 113, with instructions on using the APx Sound Level Meter Utility, which enables APx500 Series instruments to make accurate sound pressure level measurements of non‑steady‑state sounds.
Sound Advice: How to Get Settled Readings with the APx API
When using the APx API, it is important to get settled readings. Readings that are not settled may be unsteady and lack repeatable results. The following inquiry is typical of this situation:
When you open the APx500 software and select the RMS Level result of the Level and Gain measurement, you might notice that the levels indicated by the meters for each channel fluctuate continuously. These meters are constantly being updated according to the voltage sensed at the analog inputs. However, the levels indicated by these meters are not “settled” readings. Therefore, if the signal is not steady or is contaminated by the presence of noise, the readings indicated by these meters will probably not indicate the desired result.
To get settled readings in APx500, you must use the Sequencer. For example, to run the Level and Gain measurement, you would check the box beside this measurement in the Project Navigator tree and check the boxes of the desired results (for example RMS Level and Gain). Then, you could either run the entire sequence by clicking the triangular Run Sequence button, or you could right-click on the Level and Gain measurement and select “Start Selected Measurement”. This will start the generator, acquire data using the specified settling algorithm, and insert the data into the report. It also accumulates the settled readings into the Sequence Results collection, which can be accessed from the API. The settling algorithm is set to default values, which can be changed on the Advanced Settings panel of the measurement.
If you are using APx.LevelAndGain.Level.GetValues() or a similar API method to get the levels from the Level and Gain measurement, you will only return instantaneous unsettled readings (like the fluctuating levels on the meter displays in the GUI). To get settled readings, you need to use the Sequencer to run the measurement and then retrieve the results from the Sequence Results Collection. The example code below in VB.NET illustrates how to do this:
VB.NET code. View text version.
Other Recent Knowledge Base Articles:
Test Results: AP News & Events
©2011 Audio Precision, Inc.