How to Measure a Loudspeaker

How do you measure a loudspeaker? In this short video, we demonstrate a basic production line loudspeaker audio test using SoundCheck – the same measurements that are made on thousands of loudspeaker production lines worldwide.  It’s important to understand the measurements we make on this simple device, as they form the basis of the electroacoustic measurement suite for a whole host of devices ranging from headphones to smart devices and more.

In this demonstration we use a standard SoundCheck test sequence (included with the software) to measure a loudspeaker’s frequency response, distortion, sensitivity, polarity and impedance. We discuss the measurement results and the information they provide, and explain how the test sequence works. Like all SoundCheck test sequences, it can easily be customized and expanded to add additional analyses.

Watch this Video Demo of Measuring a Loudspeaker

Learn More about How to Test Loudspeakers

Here are some of our other online resources for more information on how to measure loudspeakers:

You can also check out our wide range of  pre-written test sequences for loudspeaker testing. These can be used as-is or modified for your specific requirements, saving you considerable test development time.

Video Script: How to Measure a Loudspeaker

In this short video, I’ll demonstrate testing a loudspeaker driver using SoundCheck with one of our standard speaker test sequences which is suitable for production or QC testing. This example sequence is one of many that come pre-installed with SoundCheck  to get you up and running quickly.

When we’re testing for production or manufacturing, we need to make sure the manufactured speaker closely matches the specifications of the original design, and it’s important that the test is fast, accurate, and cost-effective.  We’ll limit the test to the most important metrics –  frequency response, distortion, sensitivity, polarity and impedance. First I’ll show you the test, then I’ll tell you a little more about the measurements we made.

My test setup is very simple. I have the SoundCheck test software installed on my computer, and all I need with this is a measurement microphone, plus our AmpConnect 621 all-in-one test interface. This is a high resolution audio interface which also includes microphone power, an amplifier, and impedance measurement, all in one unit with a single USB connection to the computer. This replaces the need for separate hardware and is true plug and play, making it quick and simple to set up and use, and of course very cost-effective.

The AmpConnect 621 comes fully calibrated from Listen. Because it’s true plug and play, I simply connect USB to my computer, power on and SoundCheck will auto-populate hardware inputs and outputs including calibrations, sampling rate, bit depth and more. I’ve already calibrated my measurement microphone, so let’s run the sequence.

When I hit ‘start’ you’ll hear the test signal play. We’re using our unique smooth stepped sine wave, the Stweep as this is fast, and all the analyses are done from a single test signal. The microphone captures the signal, the amplifier impedance circuit measures the current, and the software analyzes the responses.

There you go, we just ran a complete loudspeaker test. Did you notice how quick and easy that was?  Let’s take a look at the results.

This graph shows the Frequency response and impedance. Frequency response is the best overall test to make sure the loudspeaker is working properly, and here we’ve plotted the level vs. frequency on an XY graph. Ideally, the loudspeaker should have a flat response but in general, we are looking for a smooth response, especially in the midrange from 800 to 5kHz, without too many big dips or peaks. 

We also have the impedance plotted on the same graph. This measures the resonant frequency of the loudspeaker, and should be a single, smooth and symmetrical bump. This tells us how the speaker driver will perform at low frequencies when mounted in a loudspeaker enclosure. 

You can plot these separately if you prefer, I just have them together to save screen real estate. You can see that the left axis is marked in dB SPL for the frequency response curve, and the right indicates the impedance magnitude in Ohms. 

Now let’s take a look at distortion. This sequence measures three types of distortion, THD, Rub & Buzz and transient distortion. THD is total harmonic distortion and, like frequency response, is a good overall indicator that the product has been correctly manufactured.

Rub & Buzz distortion is a periodic distortion caused by rubbing parts in the speaker, for example, an incorrectly centered voice coil or poorly trimmed lead wires. This is measured using our enhanced perceptual algorithm which examines higher order harmonics, typically 10th and above, using a perceptual algorithm that mimics the response of the human ear. This tells you whether you have any audible rub & buzz present in your device. 

The third type of distortion that we measure is loose particles or transient distortion. Transient distortion is caused when debris gets trapped behind the dust cap during manufacture. Because these trapped particles vibrate randomly rather than periodically, we measure them in the time domain with our enhanced Loose Particles algorithm. What’s neat about this compared to the way that other people measure distortion is that we separate out loose particles from Rub & Buzz distortion – most measurement systems combine the two. This really helps with troubleshooting production line issues, as these distortion types  are caused by very different problems.

Here’s our sensitivity measurement. This calculates the average level across the frequency range, and this is a good indicator of whether the loudspeaker magnet is  properly magnetized and helps match speakers for headphones and multi-channel systems.

This is the polarity measurement. This is a simple check to ensure that the loudspeaker has been wired properly and the leads are not reversed. This is important because if leads are not correctly wired, you may encounter low frequency cancellation with other drivers or speakers in an enclosure.

So there’s all the measurements in our basic loudspeaker test sequence. And as you saw, it only took about a second!

Now I’m going to show you a little more detail about the test sequence I used. In SoundCheck, we use a sequence for any test that we want to run over and over again, as opposed to quick, on-the-fly measurements. Let’s open it up. It’s really simple to program a sequence, we have a massive library of steps, which are the building blocks for the sequence. And you simply drag and drop these and adjust the values. You can also create your own custom steps, but that’s a whole other topic in itself.

OK, so here is my sequence…

It executes this series of steps from the top down, and our memory list manages the data that is analyzed. You’ll see the step prefix: for example, this is a message step… this is a stimulus step… etc. In short, what we’re doing here is defining the test stimulus, playing it, recording the response from the reference mic and impedance circuit, and analyzing the response to calculate and display frequency response, impedance, sensitivity, distortion, etc.  

Let’s take a closer look at the stimulus step. In this test, I used a compound stimulus for speed. It has 2 parts to it – the frequency sweeps from 20kHz to 300Hz in 12th octave steps and then transitions to 3rd octave steps at low frequencies from  250Hz to 50Hz. Because loudspeakers tend to have a smoother response at low frequencies, we can still get an accurate measurement with lower resolution. Optimized test frequency resolution and sweeping from high to low frequencies to minimize speaker ringing, saves us valuable test time. 

Another part of the sequence I want to show you is limits. I ran this sequence with arbitrary limits, I didn’t set them for my device, so you probably noticed it failed some of them. In a production environment, we typically measure a golden unit or run statistics on a bunch of good units, then set pass/fail limits based on desired tolerances. These limits are easy to set, and we have many options  from straight lines to offset curves. 

And lastly, here we have the  display step. This defines how we want the data to be displayed, and you can see here it’s instructing the software to plot the data on the graphs and tables I just showed you.

If I wanted to automatically save data from the sequence, this would be easy to do. I’d just need to add an Auto Save step to tell it what to save and where to, and that will happen after every measurement. SoundCheck can save data to a wide range of formats including Excel, and MATLAB to name a few, and can even save metadata along with the results.

So that’s how you can implement a complete loudspeaker test right out of the box with SoundCheck – all you would need to do to make this sequence work for you is set appropriate test levels and limits. Of course there are also countless ways you could expand and adapt this for your needs beyond the production line. For example, you could add a bluetooth interface if you need to test a bluetooth speaker, you can run open loop tests for voice-activated speakers, or add a turntable if you need to make directional measurements and plot the polar responses. And there is a whole load of additional software functionality if you are doing lab measurements, for example, Thiele Small parameters, custom equations and more. 

If you’d like to learn more about loudspeaker testing and/or discuss your test application, please contact Listen or your local distributor.