How to measure distortion in audio systems
THD measurement using Room EQ Wizard
You can easily learn how to measure distortion in audio systems with the help of some basic affordable equipment and free software. You can get the basics about distortion, and specifically harmonic distortion, in this article. If you’re lazy enough not to read it first, I’ll make some references along the line. But, if you start having a bad time understanding things, go for a quick read and then return here.
Equipment and software used
To make distortion measurements we need few items : a measuring microphone, a sound card and a computer/laptop with measuring software installed. I used the following :
- Dayton audio EMM-6 – Budget microphone, good quality, comes with calibration file. Around 50$
- Focusrite Scarlett 2i2 – Good audio interface , 2 inputs / outputs, 96 kHz sample rate. Around 150$. Other sound cards can be used regardless if they are internal or external.
- Room EQ Wizard – Very good application, Free.
This is what I use. Of course, you can use something else, as long as it gets the job done. I’m going to use some specific terminology and measuring techniques which are used in measuring frequency response and are applicable here as well. In conclusion, if you have a hard time understanding what’s to come, you can have a look at How to measure frequency response and How to make a full range frequency response measurements using ARTA. Trust me, it helps. However, if you’re lazy, like most of us, no one is going to judge you if you skip ahead.
Equipment setup and initial configuration
Before we begin, the equipment needs to be set up in a certain way. I compiled a diagram for easy following :
In the image above, you can see a simple speaker, but we are going to measure the distortion levels of a 2-way bookshelf speaker that I just made. After you fire up Room EQ Wizard (REW), you need to do some basic adjustments. Go to preferences and select the appropriate sound card and inputs / outputs.
The Focusrite sound card has 2 inputs and 2 outputs. We need only one of each. Select the correct one. Like in the diagram, I chose the left channel, but the right one can be used also.
Secondly, you need to load the calibration file for the microphone.
Go to the 2nd tab and the load the calibration file that comes with the microphone. If you have the Dayton microphone, visit their website, enter the serial number of your particular mic, and then you can download the calibration file (which is a text file). Being such a cheap microphone, it doesn’t have ruler flat frequency response. To fix this issue, a calibration file is in order, to compensate for the response irregularities of the microphone. Everything is pretty much ready, so now I can show you how to measure distortion in audio systems.
Like with measuring frequency response, the absence of an anechoic chamber, poses some difficulties in obtaining accurate results. Most importantly, the room reflections will corrupt the harmonic distortion numbers, on the lower end of the frequency spectrum.
In conclusion, if you want to learn how to measure distortion in audio speakers, you first need to find an adequate room. And by adequate, I mean sufficiently large and quiet. Just like in the frequency response measurements, you need to place the speaker on a stand. Ideally the stand should be tall enough to be half-way from the floor to the ceiling. Microphone should be placed also on a stand, directly in front of the speaker.
Let’s say your room is 2.5 m high and you place the speaker on a 1.25 m stand. Make sure that there are no other boundaries around the speaker in all direction for 1.25 meters.
Here are some suggestions on the distance between the speaker and the microphone :
- If you are measuring tweeters, place the microphone at 10 cm away. Here you will basically ignore the room reflections, because the bass response will be unusable for a tweeter anyway. Also, 10 cm is basically far-field, compared to the size of the speaker.
- For larger speakers, place the microphone at 0.5 m away. Depending on the size of the room, your lower end of the distortion chart will be corrupted with room response. But mid and high frequencies will be accurate. You can judge tall order harmonics on this chart.
- To measure distortion for the lower part of the spectrum, do a near-field measurement (place the microphone as close as possible to the speaker). Depending on how big the speaker is, this will be accurate up to a certain frequency. But most of time, take this measurement for frequencies of 250 Hz and below.
Audio settings before measuring
To take a reading, click the “Measurement” button at the top left corner. You will be greeted by this window :
Here you will set the frequency response bandwidth. Since we are measuring a 2-way speaker with an appropriate crossover, I’m going to leave the 20 Hz – 20 kHz interval untouched. If you would measure a tweeter, for example, that doesn’t have a filter of any sort, cutting out the lower frequencies would be of utmost importance (so you don’t burn out the tweeter).
The length of the sequence is also important. The longer the sweep, the more efficient it is at picking up distortion. A 512k sweep is long enough.
The volume is also important. I suggest taking different measurements at different volumes. Here you need to set :
- Amplifier volume.
- Sound card volume.
- Microphone volume.
- REW levels.
How to design loudspeakers - video courses
Fiddle with the sliders and then press the “Check Levels” button and see if you are in the green. Don’t be afraid to push the speaker. At high volumes is where most of the distortion happens. But distortion can come from almost every components in the audio chain :
- Amplifier can introduce distortion. Make sure you have a working amplifier and you don’t set the gains too high.
- The microphone can introduce distortion. This is especially for the nearfield measurement, since the mic is super close to the speaker. Make sure you dial the amp down, and maybe the microphone sensitivity also.
We are trying to measure the distortion of the speaker. When you actually hit the “Start measuring” button, you will see a live “Headroom” counter. If it goes into the red, the microphone might of started clipping. So dial the volume down a bit and retake the measurement. Also, make sure that everything is quiet. If you hear some random noise like a car horn, dog barking, while taking the measurement, cancel it and do it again.
How to measure distortion in audio speakers
After you pressed the “Start measuring” button, and the “Headroom” reading didn’t go red, the measurement should be complete. Click the 3rd button above the chart : “Distortion”. Here is what we are interested in :
To interpret the data we have to play around with the controls. Switch back and forth, to see the whole picture. Click the cog icon at the top right corner (“Controls”) and check the boxes :
- Plot normalised to fundamental. This will make the fundamental a flat like, and we can easily compare to it.
- Distortion figures in percent.
- At the bottom of the graph un-check everything except the fundamental and THD.
Now we look to see if there is any frequency above 1%. It is widely accepted that distortion under 1% is not noticeable. Some will debate that distortion under 2% is difficult to detect also. But let’s analyze the graph.
There is no frequency above 100 Hz which exceeds 1% THD. This is good news. We do see a large spike at around 80 Hz. Now we know that the lower frequencies don’t provide reliable data, and that spike is most likely due to room gain. Also, starting with 50 Hz and below, the distortion goes higher and higher. Again this may be a room factor or it might be connected to the fact that the enclosure is a bass reflex design. The port is tuned at 53 Hz, and anything below that will cause cancellation with the speaker. Since the port is not acting like a resonator anymore, it’s basically a hole in the box and the speaker excursion increases dramatically. As a result, the distortion should increase too.
Let’s click the cog once again a try another perspective. Un-check “Plot normalised to fundamental” and the distortion figures in db SPL. Then move down and un-check THD, and check all the harmonics, the fundamental and the noise floor.
Take reference the brown line, which is the noise floor. You can see that all the tall order harmonics (4th, 5th, 6th and 7th) are below the noise floor. This means that they basically don’t exist. Only the 2nd and 3rd harmonic is over the noise floor. Keep in mind that higher order and odd number harmonics are more aggressive to the ear. So, the 3rd harmonic is something to look at. Anyway, if you look at the numbers, it’s 40 dB below the fundamental, which is in no way detectable on normal music playback.
Let’s check the near-field measurement to check for distortion on the lower frequency notes :
If we analyze the chart we can see that, for the most part, all the harmonics above 3rd, are below the noise floor, which is great. However, we can see that the 2nd order harmonic is quite high when you compare it to the noise floor. Compared to the fundamental it’s 30 dB lower, which is OK, could of been better. On the good news side, we are talking about the 2nd harmonic, which is the least offensive to the ear. Because it has the lowest order and it’s an even number harmonic. So, I’m quite glad with these distortion figures.
Case study : Defective tweeter vs working tweeter
Several months ago I ordered a pair of tweeters. Unfortunately, I got a bad sample. It was in working order, but on certain music I could hear that something was wrong with it. So I decided to measure it. Here is the distortion chart for the defective tweeter :
And here is the chart for the working tweeter :
As you can see, the defective tweeter had 20 dB more worth of distortion. While it seemed fine, on certain music, you could hear the distortion on convenient musical sections. And the chart backs up this claim. With these graphs I managed to get it replaced. Although they did a Rub & Buzz test and said that the tweeter is working correctly, they were kind enough to replace the said defective tweeter.
Now you have some guidelines on how to measure distortion in audio speakers. You now have a place to start in identifying problems in your audio system. Like mentioned above, you can pin point defective speakers. Another popular use for distortion measurements is to find out the appropriate crossover points for speakers. If done correctly, there is a lot of information you can extract from these charts.
- Image source : link.