I've been building speakers for quite some time now, and to this day I've never come up with a reliable way to measure anything below ~300 Hz (anything below that is a hot mess, more a reflection of the room than the speaker itself). Things I've tried...
*Putting the mic in the port- this is what some recommend, but my findings are that the f3 is generally way too high using this method. So either the theoretical cabinet design software is wrong, or this method is not an accurate reflection of reality.
*Near field- take the NF measurement, then splice that into the actual measurement around 300 Hz. This seems to be the best method, though not entirely accurate if I'm using a woofer crossed at 300 Hz.
And I don't believe it's just me, a little later I'll post with 2 different measurements once I get screenshots of them, taken by 2 different magazines of the same PSB Imagine T2 speaker, and those 2 measurements are vastly different. I'll be honest, most of my speakers are done by ear below 300 Hz because I don't trust the measurements (aside from impedance). If you guys are willing to share, how do you get the most reliable measurement below 300 Hz?
*Putting the mic in the port- this is what some recommend, but my findings are that the f3 is generally way too high using this method. So either the theoretical cabinet design software is wrong, or this method is not an accurate reflection of reality.
*Near field- take the NF measurement, then splice that into the actual measurement around 300 Hz. This seems to be the best method, though not entirely accurate if I'm using a woofer crossed at 300 Hz.
And I don't believe it's just me, a little later I'll post with 2 different measurements once I get screenshots of them, taken by 2 different magazines of the same PSB Imagine T2 speaker, and those 2 measurements are vastly different. I'll be honest, most of my speakers are done by ear below 300 Hz because I don't trust the measurements (aside from impedance). If you guys are willing to share, how do you get the most reliable measurement below 300 Hz?
I know this isn't the main point of your earlier question, but Stereophile's method averages multiple measurements to smooth out some of the location-specific issues in the midrange and treble. Toward the bottom of the page linked below he describes the averaging method.
He also talks about his "acoustic black hole" between the microphone and speaker. Doing something along these lines is typically required, as floor bounce tends to seriously mess up your measurements in the 200-500ish Hz range (problem frequency range depends on speaker and microphone heights). If you have a tall room, angling the speaker up some and using a higher microphone position can help.
https://www.stereophile.com/content/measuring-loudspeakers-part-three-page-2
He also talks about his "acoustic black hole" between the microphone and speaker. Doing something along these lines is typically required, as floor bounce tends to seriously mess up your measurements in the 200-500ish Hz range (problem frequency range depends on speaker and microphone heights). If you have a tall room, angling the speaker up some and using a higher microphone position can help.
https://www.stereophile.com/content/measuring-loudspeakers-part-three-page-2
Last edited:
JA does not include the effects of baffle and box diffraction on his nearfield measurement which causes this rise in low frequencies. It is what it is, but it is not an accurate free field representation.
This Application note from ARTA might help
https://artalabs.hr/AppNotes/AN4-FreeField-Rev03eng.pdf
If you are wanting to measure just the speaker below 300Hz, and not the room piled on, best way (only decent way) I've found, is to measure out on a driveway or in a parking lot.
Ground plane measurement. Speaker on ground, mic on ground on hard surface. 2m minimum measurement distance, 4m is better for sure.
Try to have buildings, cars, etc, as far away as possible.
If you are measuring more than sub/low frequencies, like a full range speaker, tilt the speaker to aim perpendicularly towards the mic resting on the hard ground surface. For full range, I usually put a sheet of smooth plywood or glass, on top of a rougher asphalt driveway, to set the mic on.
Ground plane measurement. Speaker on ground, mic on ground on hard surface. 2m minimum measurement distance, 4m is better for sure.
Try to have buildings, cars, etc, as far away as possible.
If you are measuring more than sub/low frequencies, like a full range speaker, tilt the speaker to aim perpendicularly towards the mic resting on the hard ground surface. For full range, I usually put a sheet of smooth plywood or glass, on top of a rougher asphalt driveway, to set the mic on.
Member
Joined 2003
For most speakers, measuring near field for low frequency response and splicing to far field for high frequency response is a straight forward process. It really doesn't matter if its just a raw driver on a baffle or a complete speaker with crossover.
This process will create an "anechoic" response that is free from room reflection interaction, boundary reinforcement, etc. Depending on the location of speaker within a room and proximity to boundaries, real in-room response may be quite a bit different if that's what you're trying to gather.
Stereophile's splicing process is flawed since the nearfield response is not diffraction compensated, so their results usually show 5-6dB of gain at low frequencies which is incorrect.
You can follow the measurement process for VituixCAD and measure full 360 degrees of "spinorama" data if you want. Just look at the help page, you'll find instructions for REW, ARTA, SoundEasy or Clio. Single axis or full 360 degrees, the splicing process is the same.
Reference material:
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_ARTA.pdf
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_SoundEasy.pdf
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_CLIO.pdf
This process will create an "anechoic" response that is free from room reflection interaction, boundary reinforcement, etc. Depending on the location of speaker within a room and proximity to boundaries, real in-room response may be quite a bit different if that's what you're trying to gather.
Stereophile's splicing process is flawed since the nearfield response is not diffraction compensated, so their results usually show 5-6dB of gain at low frequencies which is incorrect.
You can follow the measurement process for VituixCAD and measure full 360 degrees of "spinorama" data if you want. Just look at the help page, you'll find instructions for REW, ARTA, SoundEasy or Clio. Single axis or full 360 degrees, the splicing process is the same.
- Measure far field response, 1000mm distance or at least 3x the baffle width, which ever is longer.
- Measure near field response. For low frequency accuracy use sine sweep, not noise or MLS signal
- Measure port response. For low frequency accuracy use sine sweep, not noise or MLS signal.
- Simulate baffle diffraction using VituixCAD diffraction tool. For this tool, place the mic directly in front of the speaker just as you measured in the far field, and enter the mic distance at the far field measurement distance. Export the result.
- Use the VituixCAD merge tool.
- Load near field and port response to the "low frequency" section.
- Select "diffraction response" and load the saved diffraction response from step 4.
- Load far field response(s) to the high frequency section.
- Adjust port response scale if needed, to align with near field response at low frequency. These responses should converge on each other as they approach 0Hz.
- Adjust scaling of low frequency portion to align with far field response amplitude. If you've done this process correctly, there should be a nice overlap in 300-500Hz range to splice the responses together.
Reference material:
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_ARTA.pdf
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_SoundEasy.pdf
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_CLIO.pdf
Thank you, I'll give this a whirl! A few questions...
1. When I measure far field, I would do this on the tweeter axis?
2. When I measure NF, I assume this would be center of the woofer cone roughly 1" away?
3. When I measure the port response, basically put the mic about an inch into the port opening?
Member
Joined 2003
If measuring a complete speaker, yes. But, for a complete speaker, true far field response is more like 2m which will be problematic indoors, so just understand that the frequency response at 1m may not completely represent the frequency response at listening distance, especially for larger speakers.
If measuring individual drivers for design, measure at driver axis if your following VituixCAD methods for full spinorama of info. Just follow the guide in this case.
I’ll copy the text from the reference material provided: “Measure near field response of one woofer cone at 5 mm from center of dust cap. Measure at 5 mm from cone close to phase plug if the driver has phase plug. If two woofers have shared box, feed signal to both woofers and damp the other (which is not under test) gently with pillow to prevent midrange frequencies going to mic too much.
Measure near field response of reflex port(s) or passive radiator(s). Mic in the center of vent at baffle surface if vent is not rounded. If vent has rounding, penetrate few millimeters inside, where tube with constant diameter begins. Not too deep.”
Last edited:
From the documents if you are taking measurements to design a crossover
"Elevation of mic is at the center point of driver under test i.e. mic and driver have the same Y-coordinate in mm"
If you do it all from one position moving the Y parameter inside Vituix to simulate moving the driver becomes less accurate, particularly when the driver is further away from where the mic was when the measurement was taken. The exception is listed in the document.
Member
Joined 2003
No prob, I just edited my post above with some added detail, just want to post here to make sure you don’t miss it.
Member
Joined 2003
If you're just trying to verify box tuning against the computer model, nearfield measurement of driver and port is all you need. Port measurement amplitude will need to be adjusted in VituixCAD merger tool as I mentioned above. Below tuning freq as frequency approaches zero the port output and driver output should converge, this can be done visually but can be difficult with such low tuning to begin with. The computer model can be used as a guide.
Alternative is to calculate the adjustment mathematically, without getting out your calculator it can be done for you using VituixCAD merge tool. Load both nearfield driver measurement and port measurement to the "low frequency" portion, enter Sd for the driver, and cross-sectional area of the port. Select "no baffle loss" for near-field result to compare against the computer enclosure simulation.
Alternative is to calculate the adjustment mathematically, without getting out your calculator it can be done for you using VituixCAD merge tool. Load both nearfield driver measurement and port measurement to the "low frequency" portion, enter Sd for the driver, and cross-sectional area of the port. Select "no baffle loss" for near-field result to compare against the computer enclosure simulation.
Thanks again! I'm just trying to get consistent, repeatable measurements for 300 to 400 Hz and below to make designing crossovers easier for me. I get great measurements above those frequencies, but for 3 way crossovers it can be a little tricky for me. I have good ears and things sound great, but it would be a lot easier to rough in a crossover if I could "see" what's really going on down there, rather than just guess and listen.
Member
Joined 2003
Ok, no problem then, pick your favourite measurement software and follow the instructions for VituixCAD provided above. For that process, USB mic is not to be used, a real XLR mic and 2channel audio interface is required to capture timing data (phase). If USB mic is used, single axis design information is all you get and a simulation at measurement distance only, and a different measurement instruction than above.
Also, it is important to correct for the apparent doubling of the acoustic image by the groundplane, if you wish to compare or combine the GP measurement to quasi-anechoic measurements. Read Keele's paper on the subject. Also Andrew Jones does the correction by generating a close facsimile of the groundplane image doubling using diffraction modeling software (like TheEdge) and then applies the inverse of that to the GP measurement to get something that can be combined with the quasi-anechoic type measurements. If you do not make that kind of correction your GP measurements will be +6dB higher at low frequencies, trending down to 0dB at higher frequencies.
Dang it, that rules my microphone out, I'm using OmniMic
. I need to consider how much this matters to me, seems like this will involve a lot of change, and change is bad lol.
Member
Joined 2003
You can use Omnimic, just have to join the Xsim gang instead of the cool kids that use VituixCAD. From experience, I will say the change is well worth the investment for the 2 channel measurement process.
Of course, you can use VituixCAD still, just have to throw away the power and DI chart, directivity data, follow a different measurement process and the developer will get quite grumpy with you when you mention the use of a USB mic. Better results will be had by far with the 2 channel process
Of course, you can use VituixCAD still, just have to throw away the power and DI chart, directivity data, follow a different measurement process and the developer will get quite grumpy with you when you mention the use of a USB mic. Better results will be had by far with the 2 channel process