The CEA-2010 process uses a ground plane measurement technique, placing the subwoofer under test on the ground and placing the test microphone on the ground at a certain distance from the subwoofer.
A series of 6.5-cycle tone bursts centered at 63, 50, 40, 31.5, 25 and 20 Hz is played through the subwoofer. These tone bursts (waveform shown below) are intended to simulate the signals that a subwoofer would receive when playing typical music or movies. They are also short enough in duration that they are unlikely to damage a subwoofer due to thermal overload (i.e., burning out the voice coil or the amplifier output transistors) You can find these tones in 256 kbps MP3 form on the Tech page of this website. Here’s the waveform of the 63 Hz tone burst. Note that the peak level of the tone burst is about -1 dBFS.
The technician raises the playback level for the tone bursts until the distortion exceeds any of the thresholds set by CEA-2010. For example, the threshold for the 2nd harmonic (126 Hz for a 63 Hz tone burst) is -10 dB below the peak level of the fundamental tone. So if the peak level of the fundamental is 120 dB SPL, the level of the 2nd harmonic cannot exceed 110 dB SPL. Here’s the complete list of CEA-2010 distortion thresholds:
2nd harmonic: -10 dB
3rd harmonic: -15 dB
4th and 5th harmonics: -20 dB
6th, 7th, 8th harmonics: -30 dB
9th and higher harmonics: -40 dB
You can gauge the level of the fundamental and harmonics by viewing them on a real-time audio spectrum analyzer or by using the freeware CEA-2010 program written by Don Keele. Here’s how a CEA-2010 test looks on TrueRTA:
In this example, the product has exceeded the CEA-2010 threshold at the 2nd harmonic. The peak of the fundamental is at 97 dB, while the 2nd harmonic is at 89 dB, just -8 dB below the peak. In this case, you would turn the level down until none of the harmonics exceed the CEA-2010 threshold. Then you note the peak level of the fundamental. That’s your CEA-2010 result. Repeat the process for all six test frequencies. (Some technicians measure at lower and/or higher frequencies, too, but in my opinion the six test frequencies from 20 to 63 Hz give a complete picture of a subwoofer’s performance.) Now that you understand the basic procedure, let’s get into the details.
I wrote this manual in response to requests from manufacturers who want to do CEA-2010 measurements but are confused by the variety of methods and reporting practices used. I hope this manual helps to clear things up and to expand the use of bass output measurements, without which the performance of subwoofers and large speakers cannot adequately be evaluated and classified.
As of this revision (October 2015), I have been doing CEA-2010 measurements for about five years, and have measured more than 200 products using this technique, including traditional subwoofers, passive loudspeakers, subwoofers included in 2.1 soundbar systems, and even 2.0 soundbars and portable/compact audio systems such as AirPlay and Bluetooth speakers.
In my earlier days of doing this measurement, I often found it difficult to correlate the results of my CEA-2010 measurements with other technicians’ measurements. Since then, many of the reviewers and manufacturers doing CEA-2010 -- including Don Keele, the audio engineer on whose work much of CEA-2010 is based -- have collaborated to try to ensure their measurements are in the same ballpark. While acoustical measurements always have some amount of inaccuracy, especially outside an anechoic chamber, I think most of the leading technicians now doing this measurement would achieve results within about +/-1 dB with the same sample of the same sub. In fact, I'm proud to say Audioholics now uses my results for their subwoofer ranking system for subwoofers they haven't had a chance to measure.
However, as more manufacturers and reviewers adopt this method (my fingers are crossed!), I expect some will get results that do not square with mine. I have found many reasons why CEA-2010 results often disagree. These include:
1) Some people do not seem to have read the standard. However, I can sympathize because the price for the latest version of the standard, CEA-2010B, is now up to $92 on Techstreet, which seems pricey for an e-mailed 11-page PDF.
2) The standard is incomplete in places, and in the absence of specific instructions, technicians employ what they consider to be reasonable and appropriate practices. But of course, these practices may vary from person to person.
3) The reporting standards specified in CEA-2010 conflict with legacy reporting standards.
4) Unless they’re done in an anechoic chamber, acoustical measurements are inherently imprecise, especially when done outdoors (as CEA-2010 measurements almost always are), and especially when the devices under test are being pushed to and past their limits and are may be subject to performance inconsistencies due to thermal fluctuations and other issues. Sample-to-sample variations may occur as well. Thus, minor inconsistencies among measurements done by different technicians should be expected even under perfect conditions.
In this manual, I will do my best to reconcile the recommendations of the CEA-2010A document with the practices currently in use. My goal here is not to advocate any specific measurement or reporting practices, merely to come up with methods that conform as best as possible to the published CEA-2010 standard, to common bass output measurement practice and to good basic audio testing practices. Since I wrote the original version of this manual, CEA-2010B has been released. The changes in CEA-2010B involve more notes for testing of passive subwoofers; adding 80 to 160 Hz to the test frequencies; and a modification of some of the distortion thresholds. The former two are mostly irrelevant for home subwoofers, and I doubt most people will adopt new thresholds, since no one I know even adopted the changes specified in CEA-2010A.
If you have any questions or suggestions regarding this manual, please contact me at the e-mail address found on this website. Except as noted, all material presented here ©2015 Brent Butterworth. It is free for all to access and use, but none of it may be replicated for any purpose without my permission.
The hardware required for CEA-2010 measurements is a computer, a USB stereo audio interface, a measurement microphone and a microphone calibrator.
The USB audio interface isn’t critical as long as it’s a reasonably good one. It must have a microphone input with 48-volt phantom power output, separately adjustable left and right recording level, and preferably a headphone output. Most professional-style USB interfaces will work. I use the M-Audio Mobile Pre, which is no longer available. Other USB interfaces that should work for CEA-2010 (although I have not tried them) are the Focusrite Scarlett 2i2, the PreSonus AudioBox USB and the Sound Devices USBPre. I know the USBPre has been used successfully for CEA-2010 by at least one manufacturer.
The microphone should be an electret condenser measurement microphone. I cannot find a specification for the microphone in the CEA-2010 document, but any electret condenser measurement mic should work. To my knowledge, all of these microphones have flat response below 1 kHz (adequate for CEA-2010 because the 9th harmonic of 63 Hz, the highest test freqeuncy, is 567 Hz). As some subwoofers I have measured are capable of delivering 126 db at 1 meter, it seems to me that a microphone rated to 130 dB would be the best choice. Everyone I know who does CEA-2010 tests uses the Earthworks M30 microphone, and that’s why I use it, but there’s no reason a different measurement mic couldn’t work.
The calibrator can be any standard microphone calibrator that accepts the size of measurement microphone you use. The M30 is a 1/4-inch mic, but Earthworks sells an adapter for use with 1/2-inch calibrators. (Unfortunately the adapter, which is just a little plastic cylinder, cost $30 last time I checked.) My preference is the Reed SC-05 but any calibrator that fits your microphone should work.
The distance from the microphone to the subwoofer under test has been the subject of some controversy. The original CEA-2010 document specified a distance of 1 meter or greater, but most of the people doing subwoofer output measurements were measuring at 2 meters, to incorporate a better blend of the contributions of all radiating devices (drivers, ports, passive radiators). The new CEA-2010A standard was changed to accommodate this concern, but the new document specifies a minimum test distance of 3 meters. In my opinion, 3 meters seems unnecessarily far because it is unlikely to significantly improve the incorporation of all radiating elements, while it reduces the signal-to-noise ratio of the measurement. I don’t know of anyone doing CEA-2010 who is using 3 meters as the measuring distance.
Another problem is that with devices that have relatively low bass output, such as wireless speakers and the subwoofers that come with 2.1 soundbars, a measurement distance longer than 1 meter reduces the signal-to-noise ratio to the point where it is often impossible to get CEA-2010 results at frequencies below 31.5 Hz. Also, for such small devices, a 1 meter measurement distance seems adequate to incorporate the contributions of all radiating elements.
My recommendations for test distances are as follows:
Powered subwoofers with drivers 8 inches or larger: 2 meters
Powered subwoofers with drivers 7 inches or smaller, wireless speakers, 2.0-channel soundbars: 1 meter
I hope this manual has clearly explained CEA-2010 and answered all of the questions you may have about it. If you have any further questions or suggestions for how this manual might be improved, please contact me at the e-mail address found at the top of this page.
Using an RTA to do CEA-2010 (as seen in the TrueRTA screen shot above) can work, but it’s cumbersome. To do this, you play the tone bursts from your computer (the output of the USB interface is connected to the subwoofer), and watch the results on the RTA. Put the RTA in peak hold mode and slowly raise the output level until one of the distortion thresholds is exceeded. As you can probably tell from the screen shot above, using an RTA to do CEA-2010 measurements is kind of a pain. You have to watch the levels very carefully, continually watching all of the distortion harmonics and always remembering what all the thresholds are.
The software that most technicians use for CEA-2010 measurements is a freeware program developed by Don Keele. The program is free, but it runs on the $595 Wavemetrics Igor Pro graphing/analysis software package. It’s a lot of money to spend, but using this program makes CEA-2010 measurements much easier and more accurate. Here’s the screen as I have it configured. You can also open up more windows that monitor, for example, the outgoing and incoming waveform.
CEA-2010 routines have also been created for existing audio analyzers such as Audiomatica Clio, but I haven’t tried them yet. If you already own a Clio, it seems like a good idea to at least try their CEA-2010 routine, which from what I can gauge from their description of it looks like it’s pretty well thought-out. (Unfortunately, Audiomatica didn’t release their CEA-2010 routine until after I’d already bought Igor Pro.) CEA-2010 capability has also been added to Room EQ Wizard, a free acoustical measurement application, but I haven’t had a chance to try it yet.
If you’re using an RTA to do CEA-2010, the connections are obvious. Connect the microphone to one of the inputs on the USB interface. Connect the output of the USB interface to the subwoofer. Set the input level for the microphone to minimum or near minimum, because the incoming signal will be high. Calibrate the microphone using whatever features your RTA software offers for mic calibration. (In TrueRTA this is the “SPL Calibration” function under the Audio I/O pull-down menu.
If you’re using the freeware program with Igor Pro, the connection/calibration process is somewhat unintuitive. Connect the microphone to input 1 of the USB interface and set the input level to minimum. Connect input 2 of the USB interface to output 2, so it forms a loop, and set the input level to maximum. (I use the rear-panel line input/output jacks on the M-Audio Mobile Pre for this connection; that’s the blue cable toward the top of the photo below.) You can see what the connections look like here. Note that I’ve taped “reminder labels” on the Mobile Pre so I’ll remember what the correct settings are.
Don Keele suggests using the headphone output of the USB interface to drive the subwoofer, which is a good idea because it allows the output level driving the subwoofer to be adjusted using the headphone volume control. However, because the maximum output of the headphone jack is typically much higher, this can overdrive the sub input. (On my M-Audio Mobile Pre, max output from the line outs is 1.8 volts RMS, and from the headphone output it’s 3.6 volts RMS.) I recommend testing the headphone output of your USB interface with a -1 dBFS sine wave, noting the headphone level setting required to get a 2 volts RMS output, and not exceeding this level during testing. A subwoofer should be able to reach maximum output with a 2 volts RMS signal. Unfortunately, maximum test signal level is not specified in CEA-2010. In a future revision of the standard, I hope that either a maximum test signal level is specified, or the maximum output level for the USB interface with a 0 dBFS signal is specified
I set the levels within the Windows Control Panel at 50 for input level, 98 for output level, and I make sure the system volume control (the one accessible at the lower right in Windows) is turned all the way
Every time you set up for a CEA-2010 test, you should calibrate the microphone. To do this in the freeware program, click the Calibrate Mic button in the Calibration window in the software. I calibrate at 114 dB SPL, because that is closer to typical subwoofer measurement levels than the usual 94 dB. (Technically this shouldn’t make a difference, though, and if your calibrator can output only one level, just calibrate the software to that.)
Turn on the calibrator, hit the Start Data Gathering button and let it run until the calibration level stabilizes (usually 10 or 15 seconds). Hit the Stop Data Gathering button, then the OK button to complete calibration. Note that the Counts Per Pascal readout shouldn’t change much from your last calibration. You can double-check your calibration by changing the Burst Frequency setting to 1,000 Hz in the Operation window, then hit Do Single Burst Test in the Operation Window. In the Acoustic Data window, the Peak SPL, dB result should be 117 dB if you calibrated at 114 dB RMS.
This is another area where practices among those doing CEA-2010 measurements have varied. The CEA-2010 document is specific in saying, “This SPL shall be referenced to a 1 m ground plane level,” and as described above this is a peak level measurement. However, common practice before CEA-2010 was to report the results at 2 meters using RMS values. Many, perhaps most, of the technicians currently doing CEA-2010 are still reporting at 2 meters RMS. This is not a problem for people who understand what these measurements mean, because the 2 meter RMS result can easily be converted to a 1 meter peak result by adding +9 dB. However, this has caused some confusion for some consumers who understandably look at the numbers without checking to see what the numbers are referenced to.
My practice with published reviews, such as this review of the Outlaw Ultra X10 published on Home Theater Review, is now to show both results — 1 meter peak and 2 meter RMS — in the hope that this will make consumers aware that there are two different CEA-2010 reporting standards in common use. For my consulting work, I always report 1 meter peak results.
Results for each test frequency should be reported, and the averages of results for each octave (20 to 31.5 Hz, 40 to 63 Hz) should also be presented. These averages must be calculated by converting the results in dB to pascals, averaging the results in pascals, then converting the pascal average back to dB. I have created an Excel sheet that does this calculation for you, which you can download here.
On smaller devices that do not have measurable output at 20 Hz, you can still calculate an average output for 20 to 31.5 Hz by taking the 25 Hz result, subtracting -18 dB, and using that figure for the 20 Hz result in the average calculation.You can see an example of how I report CEA-2010 results on this page. Note that I use an “L” after the dB result to denote measurements on which none of the CEA-2010 distortion thresholds were exceeded.
CEA-2010 testing is best done in a large open space, so reflecting objects cannot significantly affect the measured sound waves. Typically the measurements are done in a parking lot, because many parking lots have nearby AC outlets to power the subwoofer under test. There are, however, alternatives for those who do not have access to such a space.
The CEA-2010 document mandates that the minimum distance from the microphone and the subwoofer to the nearest reflecting object be 0.75 times the wavelength of the minimum measurement frequency. For 20 Hz, this distance is 11.8 meters or 38.7 feet. When you incorporate the mic-to-subwoofer distance, this means you need a cleared, obstruction-free area about 26 meters or 84 feet in circumference. The CEA-2010 document also mandates that the maximum ambient noise level be no higher than 50 dB RMS SPL at any frequency, although by averaging multiple test bursts, which is allowed in the CEA-2010 software, you can increase the signal-to-noise ratio — i.e., averaging four test bursts improves S/N ratio by +6 dB — and thus do your measurements in a noisier environment.
For those who do not have access to a large, relatively quiet space with an AC power source, CEA-2010 permits the use of a room correction factor to compensate for the effects of nearby reflecting objects. This allows CEA-2010 measurements to be done in a room or in a relatively small outdoor space. (I use my backyard.)
To create a room correction factor, you must first measure the ground-plane frequency response of a sealed-box subwoofer in a test environment that meets the standard cited above, or in a calibrated anechoic chamber. I built a sealed-box subwoofer for this purpose, using a 15-inch driver. The design of the subwoofer is not critical as long as it is a sealed-box design and has reasonable output at 20 Hz. (Mine is around -10 dB at that frequency.) Because I use a battery-powered amplifier to drive the subwoofer for these measurements, I was able to do them on a nearby soccer field that does not have available AC power. I recommend you run these measurements at 1, 2 and 3 meters.
Now run the same frequency response measurements in your testing environment, with the subwoofer and the microphone at the exact positions you intend to use for testing. If you display the open-space frequency response measurement on the same screen as the measurement taken in your test environment, you can gauge the room correction that is required at each frequency. Here is the result comparing the measurement in the soccer field (purple trace) and the measurement in my backyard (green trace).
Write down the difference between the free-space measurement and the test environment for frequencies in 1/3rd octave steps starting at 10 Hz and going up to 2 kHz. (The Enter Chamber Correction Data screen in the CEA-2010 software goes up to 10 kHz, but I have filled in the data only to 2 kHz, because higher frequencies are rarely if ever significant in CEA-2010.) If the test environment attenuated the audio from the device under test, write the result as a negative number (i.e., -2.3 dB at 40 Hz). Transfer these values to the Enter Chamber Correction Data screen and hit OK when finished.Note that if you intend to use different mic-to-subwoofer test distances (we’ll get to that next), you will need to create chamber correction data for each measurement distance, then save each routine as a separate Igor Pro project, such as “CEA-2010 2M” for a 2-meter distance. Then when you test with a 2-meter mic-to-subwoofer distance, open this project and your settings will be correct.
I can find no specification in the CEA-2010A document for the settings on the subwoofer or other device under test. In the absence of a standard, and in the interest of simplicity, I do CEA-2010 measurements with the device under test set to full volume. If the subwoofer has a built-in crossover, I set it to bypass or to the maximum possible crossover frequency. I deactivate all EQ modes and set the subwoofer or device under test for the flattest possible response.
Of course, subwoofers come in many configurations, with drivers, ports and passive radiators in different positions. CEA-2010 mandates that the device under test be oriented so that the measurement microphone is equidistant from the major radiating elements. In the case of a subwoofer with a driver on one side and a passive radiator or port on an adjacent side, the subwoofer should be angled 45 degrees. If the subwoofer’s radiating elements are on opposite sides, the subwoofer should be turned so its radiating elements point 90 degrees away from the microphone, as seen below. In the case of a subwoofer with radiating elements on three sides (usually a driver on one side and passive radiators on two sides), I point the driver directly at the microphone. The measurement distance is considered to be the distance between the microphone and the closest point on the device under test.
A complication arises with subwoofers that have active drivers on both ends. With these subs, the radiating elements (drivers) are further away from the microphone than they would be with a front-firing subwoofe. There are various arguments one could make for and against awarding extra dB to subs of this design (and not to subs with a front-firing driver and rear-firing port, which according to CEA-2010 requirements are also measured from the side and not the front). One could argue, as the website Data-Bass did (although not without reservations), that a compensation curve could be used that incorporates the differences in room gain among subs of differing designs. I've tested only one large sub with this configuration, and in this case, I calculated the theoretical reduction in dB that the additional driver-to-mic distance would cause, and incorporated it into my results.