Measure speaker frequency response – Different techniques
How to measure speaker frequency response ?
If you want to measure speaker frequency response, you will need some basic equipment, and sometimes, not so basic environment. The simplest measuring kit will include a microphone, an audio interface, and a computer with specialized software. The microphone is not just any mic. It is a condenser mic, purposely made to measure speaker frequency response. They need to have a ruler flat response, otherwise the result is compromised by the errors introduced by the microphone.
This doesn’t mean you have to buy very expensive microphones. Very affordable mics, with decent specs can be calibrated. The calibration file is uploaded into the measuring software and it compensates for the frequencies where the mic does not have a completely flat response. The audio interface needs to have an input for the microphone (usually XLR) , output for the speaker and USB connection for the computer. The software used to measure speaker frequency response will cost you anywhere from 0 $ to 3000+ $. If you are on a really tight budget, you can get a decent rig, for 150 – 200 $, that will give very accurate response charts.
Ways to measure frequency response
There are 4 ways to measure speaker frequency response, and they depend on the radiation domain of the setup :
- Anechoic or free-field (also called full-space or 4π).
- Half-space or 2π.
Anechoic or free-field measurements
These types of measurements are made without any objects, that can reflect the waves generated by the speaker, in all directions around the speaker. This means that the speaker has only air (no obstacles) around it, 360° in all directions (including downward). The allowed distance between the speaker and the first obstacle that might reflect sound, depends on the frequency bandwidth you want to measure. And the problems with this, is mainly the lower frequencies, because they have a very long wave lengths.
So if you want to make an anechoic measurement for full bandwidth 10 Hz to 40 kHz, you will need to have no obstacles in any directions for at least 34 meters. This can be done by measuring the speaker in a special anechoic chamber that absorbs these waves. Anechoic chambers are not easy to come by and cost a fortune to make (if you want no reflections down to 10 Hz). Another way to do it, is to suspend the speaker with a crane, up in the air, so there are no boundaries in any directions.
The first mentioned methods seem to be impossible for the average Joe, but don’t despair yet, because there is a compromise solution : Gated frequency response measurement. This means the signal picked up by the microphone is gated, so that after a certain time frame, all other received signals are ignored. The problem with gated measurement is that the shorter the gate time interval, the higher the cutoff frequency. This means that to get accurate data down to 20 Hz, you will still need no boundaries for at least 8.5 meters, which is pretty pointless indoors. However, you can get accurate results down to 200 Hz in a typical room, with gated measurement.
How to design loudspeakers - video courses
If you have available a slightly larger room and can measure speaker frequency response down to at least 100 Hz. You can make another measurement, called near field, just for the low frequencies. This can be done indoors. After that, you can splice the 2 frequency response together, using specialized software. Doing so, can yield accurate results for full bandwidth measurements. Remember that near-field measurement needs to be done up to 100 Hz the most, because of the change in radiation resistance. To perform free-field measurement, the microphone is placed at 1 meter from the speaker (industry standard)
This type of measurement has the same boundaries limitations as the anechoic measurement, except for the floor. To measure speaker frequency response using this method, you will have to go outdoors. You place the speaker on a baffle and dig a hole in the ground. Make sure that the baffle is perfectly flush with the ground. By doing so, you have an unlimited vertical space, and the only obstacles you need to care about are those on the horizontal plane. Depending on how wide the frequency response you are measuring, make sure you don’t have any obstructions on the ground for the appropriate distance. This type of measurement is called half-space or 2π, because the speaker radiates in a 180° hemisphere (only half of sphere). To perform this type of measurement, the microphone placement is at 1 meter away from the speaker.
Ground plane measurement
This type of measurement is done by placing the speaker onto a hard reflective surface and the microphone on the ground. Usually, this is done in a parking lot, with an asphalt or cement floor. This ensures the reflective surface and no nearby boundaries for at least 9 meters, for an accurate low frequency measurement. By doing this, the measurement includes the sonic mirror image of the speaker. For this reason, the microphone placement is on the ground, at the apex of the direct signal and its reflected sound image.
If you are measuring floor standing speakers, you should tilt them forward as much as possible, for a more accurate measurement. To perform this type of measurement, place the microphone 2 meters away from the speaker. This is because the sound picked up by the microphone contains the sound of the speaker + the sound of the mirror image, which adds up to +6 db gain. For every doubling of distance from the speaker, the sound looses 6 db. By placing the microphone at 2 meters away (not 1 meter) it compensates for this extra gain.
This is a technique used to measure low frequency response. Of course, all of the above mentioned measuring techniques can be used to measure low frequency response, but near-field is a quick and painless way to measure just the bass section. The microphone needs to placed as close as possible to the speaker, somewhere around 0.5 cm. Because the microphone is so close, we ignore any room reflections and baffle diffraction. For this reason, this type of measurement is closely related to the half-space measurement, because there is no baffle step (learn more about baffle step here).
Because the speaker is placed in a cabinet, the baffle will have a certain size. Depending on the size of the baffle, some frequencies will receive a boost in output (above the baffle step). Below the baffle step, the frequencies have a large wave length and act like they are in full space domain, and do not benefit from baffle boost. Think of the baffle like a reflector behind the bulb of a flashlight, it reinforces the light waves. If you measure speaker frequency response in half-space measurement, the baffle is infinitely large, so all of the frequencies receive +6 dB boost, therefore there is no step. In a similar fashion, for the near-field measurement, there is no baffle so there is no step either.
Near-field and bass reflex
This type of measurement works very well with sealed enclosures, because you have to measure only the speaker. For vented enclosures, the process gets a little more complicated, because the total output of the enclosure is done by adding up the output of the speaker with the output of the port. To do an accurate near-field measurement of a vented enclosure, you need to measure the speaker and the port separately.
For port measuring, place the microphone in the center of the port, flush with the baffle. For the measurement to be more accurate, you need to somehow isolate the speaker from the vent when measuring them individually. Placing a small cushion between the two is a good practice. But, even by doing so, port measurement above 1.6 * fb (tuning frequency of the enclosure), are not accurate, because of cross-talk between the driver and port.
One other thing to take note is that the port has a less radiating surface than the driver. Less surface means less near-field sound pressure. For this reason, some adjustments are in order. For example, if the port is half the size of the driver, this means you have to scale the port output by -6 dB. This is always the case, when the port and woofer are far apart, or on different baffles. However, if the port and woofer are close together, you can adopt a different approach. Simply place the microphone between the two and take one measurement.
Choosing your gear
If you want to measure speaker frequency response with high accuracy, it will depend on the environment you make the measurements in. Of course, making your readings in an anechoic chamber will yield more accurate results than making measurements in your home, using gated frequency response technique. The fact of the matter is that the differences are so small that the compromise solution is good enough. Ambient noise, location where you place the microphone, also contribute highly to the end result. Considering that we already covered that up in the above section, and you did all you could in terms of making the measurements as accurate as possible, let’s move on to gear.
Another piece of the puzzle that makes a big difference is the microphone. There are a lot of options out there, when it comes to microphones, starting from affordable ones and ending with quite expensive ones. Why should you pay more for a microphone ? There are 2 main reasons : frequency response linearity and frequency response bandwidth. The frequency response needs to be ruler flat, so it does not introduce unwanted response alterations to the readings. Usually, a low-cost microphone will not have a perfectly flat response, and calibration is mandatory. The manufacturer supplies calibration files with the microphone, but not always. Sometimes, you will have to go to a third party company that can calibrate your microphone. You upload these calibration files in the measurement software, to compensate for the peaks and dips in the microphone’s response.
How to design loudspeakers - video courses
Good microphone choices, for low-mid-high budget, in that order :
- Dayton audio EMM-6 – Budget microphone, good quality, comes with calibration file. Around 50$
- Audix TM1 Plus – Very good mid-line microphone, comes with calibration file. There is a cheaper version (without “Plus”). It’s the same as the Plus version, but it doesn’t come with calibration file. The response is pretty good, and the choice is yours. Around 400$ (300$ for non-plus version)
- Earthworks M30 – High end microphone. Wide frequency response bandwidth 5 Hz – 30 kHz. Around 700$
The audio interface
A microphone is worthless without an audio interface. Sometimes the microphone has an USB connection and the interface is not obligatory. Those are really cheap setups and I recommend avoiding those. A good microphone will need phantom power, which the audio interface will provide. The microphone will need 12 – 48 V of power, and this amount of power is distributed by the audio interface, though the audio cables. Make sure that your interface features phantom power, because not all of them do. When it comes to prices, you don’t need to go overboard with this equipment, because it doesn’t improve the accuracy. Just make sure you got all the connections you want and the features you desire, and that should be the main focus.
I usually recommend 2 inputs and 2 outputs. One pair is for taking measurements, and for the other pair, I like to make loop. A straight cable going from an output to an input. I set that as my reference. This is particularly useful for impulse responses and other time domain measurements. Another thing to watch out for, is the sample rate. You can measure speaker frequency response as high as half of the sample rate of your interface. So for 44.1 kHz sample rate, you can measure till 22 kHz, even if your microphone can go till 30 kHz.
Good audio interfaces :
Focusrite Scarlett 2i2 – Good audio interface , 2 inputs / outputs, 96 kHz sample rate. Around 150$
M-Audio M-Track Plus MKII – Good audio interface , 2 inputs / outputs, 96 kHz sample rate. Around 150$
There are a lot of options out there, and some of them are even free. You can go overboard with software and spend a few G’s on premium software. If you want to buy a 3000$ piece of software, you probably know what you are doing, and most likely you are not reading this article.
Software options :
As you can see, you don’t have to spend a fortune on measuring equipment. With a small budget you can get a decent rig to measure speaker frequency response. Remember that you will pay a bit extra for a microphone stand, cables and other little bits, but still the budget will remain low. When you will start to get a hang of it, you will start to measure other stuff, like the audio system in your car, the room response etc.
Real-world example of full range measurement using ARTA :