• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

dcibel said:
Port measurement is loaded to "low frequency part" of merge tool. See measurement guide and manual section for merge tool.
Click to expand...
I think I'm getting there (slowly,first time I do this! ) ,thanks!

Port.PNG

Thanks to the author,amazing program.
 
@Sokel I presume your 3 way (if non-theoretical) is using a non-waveguided tweeter on a flat baffle with power response "bunching" around 2 - 5KHz? This may increase apparent in room brightness. You may want to explore waveguides. I've encountered some good improvements in directivity using a waveguide as shown (top set = SB26ADC no waveguide, bottom = SB26ADC in a Visaton WG148R). It's made for a much smoother power response. The waveguide helps ameliorate the peak dip one often sees with conventional baffle geometry on a tweeter.

PS: Ignore the upper frequency ripple in the top set of measurements. This is an artefact of my mic boom/clip setup (fixed with the bottom set).
1705637053410.png
 
Dave Bullet said:
@Sokel I presume your 3 way (if non-theoretical) is using a non-waveguided tweeter on a flat baffle with power response "bunching" around 2 - 5KHz? This may increase apparent in room brightness. You may want to explore waveguides. I've encountered some good improvements in directivity using a waveguide as shown (top set = SB26ADC no waveguide, bottom = SB26ADC in a Visaton WG148R). It's made for a much smoother power response. The waveguide helps ameliorate the peak dip one often sees with conventional baffle geometry on a tweeter.

PS: Ignore the upper frequency ripple in the top set of measurements. This is an artefact of my mic boom/clip setup (fixed with the bottom set).
View attachment 342996
Click to expand...
Thanks for the advice but I can't change much,it's a ready speaker which I only tried to simulate as I like to practice and see.
It's not flat baffle but no waveguided either.

It's in room looks like this. in MLP.
 
Sokel said:
Thanks for the advice but I can't change much,it's a ready speaker which I only tried to simulate as I like to practice and see.
It's not flat baffle but no waveguided either.

It's in room looks like this. in MLP.
Click to expand...

yes, looks quite good to me. Need to adjust to 50 dB scale and will look much better.:)
 
Rick Sykora said:
yes, looks quite good to me. Need to adjust to 50 dB scale and will look much better.:)
Click to expand...
I made it so it can fit all three levels the 30Hz room mode I have,there's a fresh one with EQ and without:

50 scale.PNG

(I left a little bit of the 30s,as a guilty pleasure)
 
Sokel said:
I made it so it can fit all three levels the 30Hz room mode I have,there's a fresh one with EQ and without:

View attachment 343078

(I left a little bit of the 30s,as a guilty pleasure)
Click to expand...

Sorry, but the scale on the measurement was fine. It is the VCAD sim that was twice the scale that is typical.
 
The Vcad is 50 to 90db,do I have to scale it less?
I just used the defaults I think,totally newbie to it!

I see what I can do.
 
What mic are
Sokel said:
Trying to simulate some speakers I have here and finding no other way to add the port,I measured it nearfield (very),I did the whole proses,etc and added it as a driver.
Is it right or wrong?Is there another way?

View attachment 326263 View attachment 326264
Click to expand...

The quasi anechoic measurement procedure I use is
1. Measure port nearfield
2. Measure woofer nearfield
3. Scale port down to match woofer
4. Merge
5. Apply baffle diffraction to merged near field response
6. Merge baffle diffraction nearfield with gated Fairfield

I can break this down further if useful. Also some variation of multiple woofers are in circuit

Having said all that the procedure documented in this very thread is the same
 
I really appreciate this tutorial - I have decided to give it a try but I got stuck when exporting all the IRs from REW

When I click here:
1708003409442.png


I do not get the same Export window that I shall; like this one:
image3.png


But this one instead:

1708003474609.png


So I just hit export but when I try to import these in VituixCAD I get the below error message:

1708003515619.png


Can anybody please advise?

FYI @ernestcarl
Thank you
 
ppataki said:
I really appreciate this tutorial - I have decided to give it a try but I got stuck when exporting all the IRs from REW

When I click here:
View attachment 349734

I do not get the same Export window that I shall; like this one:
image3.png


But this one instead:

View attachment 349736

So I just hit export but when I try to import these in VituixCAD I get the below error message:

View attachment 349737

Can anybody please advise?

FYI @ernestcarl
Thank you
Click to expand...
I've only found success in exporting every measurement singularly, and renaming them to match the needed syntax for Vcad.
 
ppataki said:
I really appreciate this tutorial - I have decided to give it a try but I got stuck when exporting all the IRs from REW

When I click here:
View attachment 349734

I do not get the same Export window that I shall; like this one:
image3.png


But this one instead:

View attachment 349736

So I just hit export but when I try to import these in VituixCAD I get the below error message:

View attachment 349737

Can anybody please advise?

FYI @ernestcarl
Thank you
Click to expand...

Did you try clicking "Export all measurements as text?" instead of Export all impulse responses?
 
napilopez said:
4) Nearfield Bass Measurements

Because gating the impulse response means we lose bass resolution, we need to make up for it by capturing the bass some other way. Some people use the ground-plane method, which is essentially the gold standard outside of a giant anechoic chamber or NFS, but it requires having like 30 feet of open space around the speaker, which most of us don't have.

Nearfield bass measurements can be done virtually anywhere and can get us pretty darn close to an anechoic or ground plane response. As we are measuring extremely close to the sound sources, it drowns out any reflections from the room. Here's the summary of what we're about to do.
  1. Measure the bass sound sources from really close up.
  2. Adjust SPL levels for the different radiating areas.
  3. Sum the measurements
  4. Modify the sum to be accurate at farfield distances (baffle step adjustment)
  5. (Optional) Splice to on-axis in REW. This can be done later in VituixCAD too when creating a full spinorama.
For this guide we are going to use the most common speaker type: a ported two-way speaker (I'm using the JBL-HDI-1600 for this example).

We start by capturing the individual frequency response of the bass sound sources. You should start with the smaller sound source, usually the port (more on this later). Then:
  1. Measure the diameter/area of the port. If it's a flanged or tapered port, you can use either the 'throat' or 'mouth' of the port, but I lean towards the former as it'll be louder and have less potential contamination.
  2. Set your microphone flush with the port's throat or mouth, whichever diameter you measured earlier.
  3. Run your sweep (adjust volume and retry if clipping occurs).
  4. Measure and write down your woofer's effective radiating diameter/area. You can set your ruler or measuring tape starting and ending at roughly halfway through the woofer's surround. Important: this is not the woofer's spec-sheet diameter. A 6.5" woofer typically only has an effective diameter of ~5".
  5. Set the microphone as close as possible to the center of the woofer, leaving just enough of a gap to clear the woofer's excursion. Technically, the microphone should be about 0.05x the woofer's effective diameter, about 0.25" for a 6.5" woofer with a 5" effective diameter. But you can usually just eyeball it.
  6. Without touching the volume control, run your sweep. The woofer should be measured at the same source volume as the port.
Just to give you an idea, this picture shows how close to the woofer the microphone was when I measured the JBL L82.
View attachment 123440

Now that we have our two measurements, we need to reduce the port's SPL to match the woofer's.

The 'proper' way, as noted in Jeff Bagby's whitepaper, is to follow this formula if both port and woofer and circular: 20 Log ([port diameter]/[woofer effective diameter]). If one of them is not circular, you can use this generalized formula by calculating the area: 10 Log ([port area]/woofer effective area]).

The 'easy' way, which works just as well 95% of the time, is to simply eyeball it. Just adjust the port's frequency response until it aligns with the woofer's low-frequency tail, usually the region below 20Hz. You don't even need to measure the area of the port and woofer this way, although I still recommend going through all the steps just in case. This method is also useful when the port or woofer isn't circular, when calculating the precise area becomes more tricky.

View attachment 123441

To adjust the port's SPL, go Controls>Measurement actions and then select the port's measurement. Change the SPL offset until the port's low-end tail matches the woofer's. Note the SPL offset value, but I recommend you don't need to add the offset to the data just yet — this is just to figure out how much we need to adjust the port. We will add the two responses correctly next. If you want, you can write down the required SPL offset in the measurement notes on the left of the main REW window.

View attachment 123442

Whether you calculate the adjustment or eyeball it, next go to Controls>Alignment tool, select the port and woofer measurements, and change the port's gain to the correct SPL offset. You should see a preview of the summed response, and you can mess around with the gain and delay if you want to see how that affects things. Then tap on 'Aligned sum' to add them up.

View attachment 123443

The resulting response, seen above in white, will have an exaggerated bass. This is because we need to correct our nearfield measurements for baffle step. There are many ways to do this, but we will use Jeff Bagby's Diffraction & Boundary Simulator, as IMO it is the easiest.

Enter the speaker and woofer's specs into the simulator (sorry non-Americans, the simulator uses imperial units). Don't worry too much about being perfectly precise, but for the 'speaker piston diameter,' remember this is the effective diameter. Again, a 6.5-inch woofer typically has a 5-inch pistonic region.

You will see a handy preview of the predicted correction for the nearfield response.

View attachment 123444

Once complete, tap on 'Save Baffle Driffaction.' You can then import (Ctrl+I) or drag and drop the resulting file right into the REW window.

From here we just go to Controls > Trace Arithmetic and select our Aligned Sum and the correction we just made, and set the operation to A*B. Tap generate, and you'll get the corrected response.

View attachment 123445

The corrected response will have less exaggerated bass, usually down by 4-6 dB by 100Hz.

View attachment 123446

This new response can now be aligned with your on-axis response. If all is well, you should see decent agreement between the contour of your corrected nearfield summation and your on-axis measurement. Let's use the 'Measurement actions' tool again to bring down the summation to the SPL of the on-axis (always bring the corrected measurement down to match the on-axis, not the other way around).

Line them up as best you can between roughly 200-800 Hz. Some deviation is fine (remember, our on-axis has lower resolution in the bass), they should just generally match. This time, do tap on 'add offset to data.'

View attachment 123447

Now, we can merge the corrected nearfield response with the on-axis response. This will be done later in VituixCAD if you plan on measuring a full spinorama, but I often like to do a preview first in REW anyway.

Go back to Trace Arithmetic, select your on-axis response under field A, and your corrected nearfield response in B. Then select Merge B to A as your operation and choose the frequency roughly where the curves best line up. I recommend selecting the 'blend' function on, this will smooth out the transition within 1/3 of an octave. It just looks a little nicer.

View attachment 123448

And voilà! We now have an on-axis response that should be pretty close to an anechoic result.

View attachment 123449

As has been demonstrated a few times on this forum, getting perfectly consistent bass results among different measurement techniques, even with trusted methods like the Klippel NFS, anechoic chambers, and ground plane, can be difficult. So don't fret too much if your results don't line up 100%. Our goal is to get 'close enough,' as the bass response will be altered dramatically by the room anyway!

Some more notes you might find useful:
  • It is important that you measure the smallest sound source first because it will appear loudest to the microphone. If you measure the woofer before the port, you may find the port measurements clipping at your chosen volume.
  • Ideally, the port and woofer should be measured at the same 1-meter SPL as your on-axis, but you might find this is too loud when the microphone is so close to the sound sources. For passive speakers, the bass response will rarely vary dramatically at different SPLs, so just use the loudest SPL you can get away with.
  • But keep the above in mind if measuring a speaker with DSP, which may have a response that varies significantly at different SPLs. You could also try changing the gain on your Umik-1.
  • If a speaker has multiple identical woofers, ports, or radiators (and you're sure they cover the same frequencies) you can simply add 6dB for each additional unit rather than taking an additional measurement.
  • Some DSP speakers, like the D&D 8C and the Buchardt A500, have different timing between different drive units. For example, on the D&D 8C, the rear subwoofer units actually fire before the front woofer and tweeter, leading to summations that seem off. For this, you can mess around with the delay settings in the alignment tool until the measurement looks correct. Often times with such time alignment, the delay may simply be roughly equal to the difference in distance between the drivers.
  • If you have doubts about the accuracy of your summation, you can roughly confirm it by measuring the speaker from a few close distances, say 1 or 2 feet. At this point, your room will have an influence on the measurements, but you should see the general trends line up decently.
  • For 3+ way speakers, the 'eyeball it' method gets trickier, so you might just want to do the math.
  • On the other hand, sealed speakers are easy-peasy. All you need to do is measure the woofer and do baffle step correction.
  • The nearfield response can sometimes show low-frequency resonances, but they don't always show up quite the way they would in an anechoic or ground plane measurement. So keep that in mind that it still doesn't quite perfectly capture measurements in this region.
  • If the woofer has a phase plug, or other unusual design, just get as close to the center as you can. If the woofer has a permanent grille, I usually put the microphone flush with the grille, although you should see if the grille is removable without damaging the speaker first.
  • If you want to be as accurate as possible, you can try to apply baffle step correction to the woofers and ports individually and then add them up. In my experience, the difference is so tiny that it's usually not worth the extra hassle. On the other hand, it does allow you to see a more accurate preview when using the alignment tool, which is nice.
I know, I know, this is all very long, but it's a lot more tedious than it is difficult. Once you've done it a couple of times, manipulating the data doesn't take long at all; It's only the measurement setup that really takes that much time.
Click to expand...
Many thanks @napilopez for an excellent thread. I plan to share some of my self built constant directivity loudspeaker in a separate threat later but have a question for the forum whether members have experience measuring and processing quasi-anechoic passive cardioids (or dipoles) in the bass/midbass region. Standard near field measurements used for woofers/ports don’t work here as the cardioid effect is inherently a farfield effect eg woofer front output and the (attenuated, delayed, and inverted) acoustic resistant port output sum in the farfield to either provide a small boost (on-axis) or gradually cancel towards the side and back progressively with increasing angle. One idea I am playing around with is using the alignment tool in REW to combine & align (attenuate & delay) near-field measurements for the cone and port to create a simulated farfield response at a certain distance but the difficulty is off course as to how to correct for baffle-diffraction (which are different for cone & port hence I will apply baffle-correction-curve separately to the individual near-field curves before applying alignment). Simultaneously will be measuring with higher pole to get gating resolution to ~90Hz which is sufficient as the passive cardiod is crossed at around 250-400Hz to a cardioid-pair of active woofers hence goal is to have reliable data through the cross-over region to model drivers separately in VituixCad. Ideas & experiences very much welcomed!
 
Last edited:
Very good thread. Spinorama is a good way to understand the speaker, but it has its limitations as standardized today. Tilted setups, computer speakers, different heights, etc., sometimes require the idea of Spinorama, but need different angles for the real listening spot. So, building your own in Excel is a sure and traceable way for a modified Spinorama. The formulas are quite simple.

If not weighted, the first step for linear scale and square is: =(0.0002*(POWER(10,(Measured/20))))^2. This is used to take all measurements to a new sheet with squared linear value. The second step is =20*LOG10((SQRT((AVERAGE(SquareRange))))/0.0002), where you take several measurements and set them back to the average in dB scale.

In the case it is weighted, the formulas are quite similar. The first step: =Weight*(0.0002*(POWER(10,(Measured/20))))^2 and the second =20*LOG10((SQRT((SUM(WeightedSquareRange))))/0.0002).

Of course, these steps can be split differently as needed or preferred. As REW defines, it uses the same square root average method for visible measurements also. So, you can do all non-weighted ones in REW as well. Just to remember, that 2 times horizontal is needed for correct average.
 
mwmkravchenko said:
And you will want to do a polar measurement to get an understanding of the actual radiation pattern. It is both educational and useful for placement in your home. Plus, don't you just need to know?

Mark
Click to expand...
Yes I do. I'm just starting with the off axis measurments and I was almost sure that nearfield measurments will not work here. I was just curious if the ground plane would be better or in the open space :)
 
You must log in or register to reply here.

Similar threads

D
How quiet for quasi anechoic measurements?
Replies
11
Views
862
digahill
D
A
Speaker Review Anechoic Measurements- Grossly Inadequate and Why
2
Replies
26
Views
4K
boxerfan88
boxerfan88
G
Magnat Monitor Supreme 102 Bookshelf speaker measurements
Replies
1
Views
963
GyTomi38
G
edc
Questions about taking relevant speaker measurements / spinorama data
Replies
3
Views
853
edc
edc
Ageve
  • Poll
Infinity Primus 150 spinorama measurements (CTA-2034)
2 3 4
Replies
65
Views
13K
RobinOC
R
Back
Top Bottom