Direct Field Acoustic Noise testing - DFAN
DFAN, or Direct Field Acoustic Noise testing, is a method for acoustically testing spacecraft or other items without requiring a special, dedicated test facility with a reverberant room. The DFAN method is less expensive, and helps reduce that chances of damaging a spacecraft while moving it for testing.
When satellites are launched into space, while riding at the top of a rocket they are exposed to an extremely violent noise and vibration environment. The relatively delicate satellite with its structural and electronic components are somewhat acoustically protected by a fairing, as shown in Figure #1. This is limited protection, so it is prudent to ensure they can survive both the noise and vibration through a test campaign on the ground prior to launch.
Figure 1: Cut open view of a rocket fairing showing a satellite.
In this article we will cover a method for acoustically testing a spacecraft. Other community articles have addressed the vibration aspect of environmental testing, and for more on this topic, please see the article, “What is Vibration Control Testing” .
For the Acoustic Testing scenario, the excitation sources are noise generators, (for example modulators-horns and/or loudspeakers). Today’s standard practice is to use dedicated facilities like reverberant rooms to excite a structure acoustically. These are rooms in which the acoustic field is uniform and diffuse and the noise in the room is shaped to mimic the spectrum, (amplitude vs frequency) which the spacecraft would experience during launch inside of the shroud.
On one hand these facilities provide a very reliable means for testing, but on the other hand there are very few of these, and they are very expensive to build. For these reasons, over the past 20 years the industry has been looking at alternative ways of completing acoustic tests using a different excitation method.
An alternative to the reverberant room approach is to use speakers which surround an object and subject it directly with noise, hence the name Direct Field Acoustic Noise, (DFAN). In this article we will restrict the focus to this alternative method with the use of (commercially available) loudspeakers as a means to excite the Test Article (TA).
To summarize, there are 2 fundamental methods for acoustic testing a spacecraft.
These methods are discussed further in the next sections.
Reverberant Field Acoustic Noise (RFAN)
The first method, RFAN, has been the standard used from the beginning of spacecraft testing, but there are some benefits to consider for the DFAN method. These tests are generally for spacecraft and spacecraft component testing.
Firstly, the major drawback of the reverberant chamber is the cost. Running at several million dollars, these rooms are only available for companies willing to make such a large investment. An example of a reverberant chamber with a spacecraft is shown in Figure #2.
Figure 2: An example of a spacecraft being readied for an acoustic test in a reverberant chamber. The chamber shown is the NASA Glenn Research Center, Reverberant Acoustic Test Facility (RATF), which is capable of 163dB OASPL!
For companies without these facilities, they must move the spacecraft to an available chamber, (in some cases even to a competitor’s location) at significant cost, increased scheduling, and with the risk of damage.
Direct Field Acoustic Noise (DFAN)
It is worth it to emphasize that the benefits of DFAN are due to the lower cost and convenience rather than a better testing method when compared to a large reverberant test facility.
To setup a DFAN test requires any conventional room large enough to accommodate the space requirements, and also needs to withstand the high Sound Pressure Levels necessary to comply with the test requirements. This means an outdoor venue is also possible, as long as the neighbors aren’t disturbed with a simulated rocket launch!
The speakers and audio equipment can be trucked to different sites so it can be very portable, allowing convenience in spacecraft scheduling. An example of a setup for a DFAN test with a spacecraft is shown in Figure #3.
Figure 3: DFAN Test setup with a spacecraft mockup. There are a large number of speakers required to achieve 146dB OASPL in a DFAN test arrangement.
The figure shows an arrangement which requires a large volume to accommodate the size of the TA. There are also restrictions on how close the speakers are to the TA. The number of speakers can be on the order of 100 or more for spacecraft testing, which is dependent on the size of the acoustic volume to accommodate the test article, and the required acoustic level.
History and How DFAN Works
The idea of using speakers has been around since the late 1960’s, but it was not until the late 1990’s when the first experiments to validate the method commenced. Since then, and after much experimentation and control evolution, successful tests have been achieved, and it is now considered an acceptable method as an alternative to RFAN when necessary.
As part of the validation process, and for this method to be accepted and a test to be deemed successful, it must produce the 2 main results obtained from the RFAN method, and the system must be capable to provide this for a test duration of about 1 minute:
The reference spectrum to be controlled depends on the rocket that will lift the spacecraft into space. This will determine the OA level and the levels which are defined in 1/3 Octave bands. The center frequencies will range from about 25 Hz up to 10 kHz. A typical reference spectrum in Narrow-band format is shown in Figure #4. Note the focus of energy is in the band from 80 Hz to 300 Hz.
Figure 4: Defining the target reference spectrum in narrow-band format including the alarm and abort levels is similar to a structural vibration control test.
In effect, the objective of a DFAN test tries to achieve the acoustic levels and the diffuse field effects of a reverberation chamber. Another criteria to strive for is to have as uniform an acoustic field as possible, meaning the same SPL anywhere in the acoustic volume.
To achieve these specifications can require a large number of speakers surrounding the test object. The output from the speakers is controlled with a MIMO Random Control system, and how this control is achieved will be explained further in this article. Even after achieving the general requirements of replicating the acoustic environment of a reverberation chamber, there are still concerns with regards to the corresponding structural responses compared to the RFAN method, and this aspect is still being closely watched with this DFAN method.
The most optimal configuration to carry out this type of acoustic test is to position the loudspeakers in a circular pattern to surround the test article. The speakers will provide the excitation of the acoustic volume, with several microphones spatially distributed in the volume for control and measurement. The number of speakers can be quite large. A typical layout is depicted in the schematic in Figure #5.
Figure 5: The schematic shows a typical arrangement of individual speaker stacks surrounding the TA. Note the non-symmetric arrangement of speaker stacks, which is intentional. The control and monitoring microphones are randomly placed in the acoustic control volume.
The number of microphones can vary too. Some typical numbers are for 12 control microphones and an equal number or more of monitoring microphones. One of the lessons learned from years of experiments is to avoid symmetry in the speaker and microphone arrangements to reduce the wave interferences. There are typically also some accelerometers that are attached to the TA to monitor the structural responses.
The microphones are randomly spaced in the acoustic volume, as shown in the figure. There are some guidelines for positioning the control mics, with a minimum distance from the speakers and from the TA, which will reduce the near-field effects of the test object and speakers. As specified in the NASA standards, the positioning of the mics are placed about 2-ft from the test object and 3-ft from the speakers. There are algorithms available which can determine the best location of the control microphones based on either simulations, or actual measurements as part of the test process after a selfcheck.
Multiple Input Multiple Output (MIMO) Control and Frequency Ranges
The MIMO Random Controller software used for the acoustic application functions the same as for the structural application, it is just a different form of excitation and response.
To cover the full range of frequencies, each single stack consists of a number of 3-way speakers with combined high, mid-range, and low frequency response from 60 to 18k Hz. There are also a number of sub-woofer speakers to cover the low frequency range from 28 to 90 Hz. This can vary a bit depending on the speaker and manufacturer, but a typical configuration of a single stack is shown in Figure #6.
Figure 6: An example for a single “stack” of speakers will cover the entire frequency range, and a single drive output signal from the controller is assigned to each individual stack.
The number of these stacks is determined by the level of noise that must be produced, and the size of the acoustic volume. The acoustic level is usually extremely high and around 146dB OASPL for spacecraft applications! (Currently this level is about the limit of sound pressure achievable with DFAN, so for higher levels an RFAN chamber is required.) For every single stack of speakers, there will be a single independent control drive signal from the control system.
To power these speakers requires a set of high-powered amplifiers, which also provide the cross-over filters just like in a home stereo system. An example of a cross-over filter set is depicted in Figure #7.
Figure 7: The cross-over network filters are similar to a home stereo system. The specifications of these can vary depending on the speaker and manufacturer.
The amount of equipment and electrical power required for this can be enormous. A photograph of a partial set of high-powered amplifiers is shown in Figure #8.
Figure 8: The amount of electrical power and equipment can be enormous. The photo shows a sample of the amplifiers needed to power the speakers.
The test duration is for 1 minute, so the power must be continuous for the duration. Protection of these systems is provided by monitoring parameters such as current and temperatures to make sure they are not exceeded.
Spacecraft Component Testing with DFAN
Another application for DFAN is in spacecraft component testing, (e.g. Antennas, Solar Panels, etc.). These components also must adhere to a similar set of required environmental acoustic tests. For a given spacecraft, the number of acoustic tests is greater due to the number of components that need to be tested, unless they are tested concurrently when attached to the spacecraft. An example of this test arrangement is shown in Figure #9.
Figure 9: A typical Direct Field Acoustic Excitation setup is shown. In this case the DUT is a spacecraft antenna dish. The acoustic level of this test is typically the same as a full spacecraft test at about 146dB OASPL.
The scale of this test is smaller and does not require the larger volume, and hence the number of speakers, as a full spacecraft test. However the same acoustic levels and conditions must be met for these tests.
Controlling the Acoustic Volume
In a previous article on MIMO (Multi-Input Multi-Output) Testing, we discussed the differences of controlling the multiple excitation sources, which were classified as either “Open-loop” or “Closed-loop” form.
The DFAN test is considered a “Closed-loop” form of MIMO Random testing, where the random voltage output signals from the control system to the amplifiers/speakers are adjusted based on the control microphone responses. The concepts that were described in the context of structural testing apply equally to the DFAN test. For that reason it is highly recommended to read the community article: “Multi Input Multi Output MIMO Testing” to understand more about the function of the controller.
As for the structural case, the control algorithm uses a fully populated spectral density matrix (SDM) that contains the acoustic PSD which serves as the main target spectral response. Also included in the control are the cross spectrums of the control channels, which are defined with phase and coherence requirements to update all drives simultaneously based on the responses of the independent control channels.
Narrow band spectrum control allows finer resolution in controlling the acoustic responses and hence uniformity. However the DFAN test results are typically provided in 1/3 Octave format to adhere to the traditional RFAN method of displaying data. Using narrow-band control also allows finer fidelity in limiting high Q responses that may occur in the acoustic volume or on the structure. A typical result shown in narrow-band format is shown in Figure #10. This shows the spectrum averaged across all control responses, along with the reference and corresponding alarm and abort limits.
Figure 10: The control response spectrum averaged over all control channels is shown in narrow-band format.
The same results which are averaged into 1/3 Octave band format is shown in Figure #11.
Figure 11: The narrow-band control response is typically shown in 1/3 Octave band format.
Besides controlling the spectral responses, the data acquisition hardware will measure the control mics, and any other type of response sensor, such as other monitoring microphones, accelerometers, and strain gages, combining acoustic response and structural response in a single database. The data is also typically recorded simultaneously in the time domain for post-processing activities.
Pre-test Analytical Modeling and Simulations
DFAN testing is another application which can be supplemented with simulation to achieve a more efficient test. There is a lot of variability in the test arrangement possibilities with the DFAN method. For instance, how to determine the number and type of speakers, the position of the speakers, the number and position of the control mics, and even the position of the test article, are all variables which could be optimized with some Pretest simulations.
A DFAN test, while overall is less expensive than an RFAN test, still has significant cost and time to be invested. Also, the structural responses can be risky, so analysis can have a profound effect on the success and quality of the test results. A structural response prediction of the TA can also ensure levels will not be exceeded, or even allow for the anticipation of response limiting, all of which can help in “de-risking” the entire test.
After setting up and running a DFAN test, finding out too late there is not enough acoustic energy due to having too few speakers or not enough electrical power will result in a disaster. On the other hand, having too many is a waste of resources. Using some level of simulation, and possibly including models of the different components which make up a DFAN scenario, such as the speakers, amplifiers, microphones, and even including the controller in a control loop simulation, is highly desirable and will help to verify and optimize a test setup.
Using simulation can be used to determine the optimal location of the control mics. This helps to improve the uniformity of the acoustic field, which is one of the main objectives of the DFAN method to replicate the RFAN method. There are simulation algorithms available which can pick and choose out of a large number of microphone locations, a reduced number of control microphone locations, leaving the rest as monitoring microphones. This also takes into account the phase relationship of the control mics, which will reduce the risk in having unwanted constructive and destructive wave interference and standing waves. All of this contributes to the acoustic level uniformity and optimization of the drive level requirements.
There is a lot of work to be done in this area, so the simulation tools for this DFAN application are still evolving, but there are a lot of possibilities in this area. A sample simulation model which has been used for speaker and microphone position studies is shown in Figure #12.
Figure 12: Using Simulation tools can help produce a more efficient test. The number of speakers and mics, and the optimal position of these can be determined through modeling.
Further Information Regarding DFAN Testing
NASA has recently developed a technical handbook which describes this type of test, NASA-HDBK-7010.
Related Siemens Testing Community Articles:
Related Internet Links to this article: