• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. 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

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
When summing two identical woofers, you just add 6dB to its SPL. If you're using the method of aligning the low-end response tails, it doesn't really matter. I will note that if you haven't, you should also measure the midrange driver too. I don't know what the crossovers are for the speaker

So I'm not sure of the top of my head why this happened so dramatically @Bghead8che , but its basically the phase of the two measurements are not time aligned correctly. This is good for anyone following this thread, so for reference, this is the way the port and woofer are summing in @Bghead8che's measurements:
View attachment 193424

Whenever you see various dips like that in succession ("comb filtering") when summing measurements it's a sure fire sign that there's a timing issue.

If you go to the 'Impulse tab' and look at the two measurements, you'll see that the references timing is very different. Close to 0ms for the woofer, but -27ms for the port:

Woofer:View attachment 193419

Port:View attachment 193422

Whenever I see comb filtering like that, you can try using REW's "alignment tool" to sum the graphs and have control over timing as well.
(All SPL> Controls> Alignment tool). Here you can control both the gain (dB) of the response and the timing, and see what would happen if you changed the timing.

View attachment 193425


That showed that modifying the timing of the port by adding ~26-27 ms gets us a respones more like what we'd expect. A good way to do this is to tap on the little slider and hold the left right up or down arrow on your keyboard and see the response change live.

View attachment 193427

So I'm not sure why that difference in timing happened. You can see from the alignment tool that delaying the port by 26.75ms is equivalent to moving the port 30 feet/ 9m away lol. But that gets us closer to what we want. You can then tap on "aligned sum" to save the response.

Alternatively, you could make a copy of your responses, go to the impulse tabs, click on controls, and set t=0 at impulse start. That can cause issues during far field measurements but should be fine for these nearfield ones.

EDIT: On closer look, the measurements have a note saying you used an acoustic timing reference. Was that on purpose?

View attachment 193430

For some reason that seems to have screwed things up.

Also these measurements are using the 90 degree calibration file. Not a big deal for these frequencies, but FYI it should be the zero degree one.

You nailed it! I had a timing measurement on for both measurements which I had used previously for in room measurements. I am going to turn that off and change my calibration file. I'll measure the midrange as well though it may be difficult because it has a large phase plug. In your guide you say to just measure to the side of the phase plug.

<<When summing two identical woofers, you just add 6dB to its SPL. If you're using the method of aligning the low-end response tails, it doesn't really matter. I will note that if you haven't, you should also measure the midrange driver too. I don't know what the crossovers are for the speaker>>

I'm not sure if I understand you here. Are you saying if you align the low-end tails you do not need to add 6db? How exactly do you "add 6 db?" Do you add a 6 db offset?

In the guide you wrote the following in the notes:

<<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. >>

I'm a little confused by this. Are you saying instead of measuring the close to the woofer and flush with the port you should measure them 1 meter away? Sorry, this wasn't clear to me.

I am going to remeasure the midrange, both woofers, and the port. Hopefully you can then help me put them all together correctly. Then we can compare the final response to the official posted response and see how close we come!
 
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
You nailed it! I had a timing measurement on for both measurements which I had used previously for in room measurements. I am going to turn that off and change my calibration file. I'll measure the midrange as well though it may be difficult because it has a large phase plug. In your guide you say to just measure to the side of the phase plug.

<<When summing two identical woofers, you just add 6dB to its SPL. If you're using the method of aligning the low-end response tails, it doesn't really matter. I will note that if you haven't, you should also measure the midrange driver too. I don't know what the crossovers are for the speaker>>

I'm not sure if I understand you here. Are you saying if you align the low-end tails you do not need to add 6db? How exactly do you "add 6 db?" Do you add a 6 db offset?

In the guide you wrote the following in the notes:

<<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. >>

I'm a little confused by this. Are you saying instead of measuring the close to the woofer and flush with the port you should measure them 1 meter away? Sorry, this wasn't clear to me.

I am going to remeasure the midrange, both woofers, and the port. Hopefully you can then help me put them all together correctly. Then we can compare the final response to the official posted response and see how close we come!

Yeah, adding 6dB is only if you're doing the manual calculation.

As for the 1 meter thing, I apologize, I can see how that be unclear. What I basically mean is that in an ideal world you would be measuring your nearfield measurements after your 1 meter far-field measurements without touching the volume control. This is so that you're maintaining the woofers frequency response at that specific volume level.

So for example, let's say you performed your 1m measurements at 90dB. Then as you move he microphone close to the woofer, it's going to read a much higher SPL, even though the volume hasn't actually changed. The problem with this is that the volume might now simply be too high for the microphone, causing clipping which probably isn't good for your microphone either.

So then you might want to turn down the volume in order to save your poor microphone. The problem with this is that it's possible the speaker has a different frequency response at different SPL levels. It's rarely an issue with passive speakers, but definitely something to keep in mind with active ones, which may have all sorts of DSP trickery going on.

So basically, if it's not too loud for your microphone, don't change the volume between making your farfield and nearfield measurements. If you have to it's fine, just be aware that the frequency response can change with SPL level with DSP enabled speakers in particular.
 

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
I measured all drivers separately. There are 75 db measurments for:

On Axis - 1 meter
Midrange - .20 inches
Woofer #1 - .20 inchers
Woofer #2 - .20 inches
Rear Port - flush

Napilopez, would you mind combining the measurements for a final on axis measurement? As the guide covers a single woofer only, perhaps you could post some screenshots so myself, and others, can learn how to combine measurements for a tower speaker. The baffle adjustment was posted in a previous post. I'm very curious to see if the curve matches the official Ascend Acoustics curve posted online. Thanks again for the awesome guide and your help!

I attached the 5 measurements. Due to file size, I put 3 of the measurements in the first file, and 2 in the second.
 

Attachments

  • Ascend 5 Raw Measurments Part 1.zip
    3.9 MB · Views: 109
  • Ascend 5 Raw Measurments Part 2.zip
    2.6 MB · Views: 68

as labs

Member
Joined
Mar 10, 2022
Messages
33
Likes
19
Location
hudson valley, ny
maybe this has already been covered, but what is the reason for using no timing reference? or rather, why not use the loopback timing reference?
 
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
maybe this has already been covered, but what is the reason for using no timing reference? or rather, why not use the loopback timing reference?

Loopback doesn't work with the Umik-1, which is what this guide was (primarily) designed for. I haven't tested the acoustic timing reference enough to know how reliable it is. It's not particularly necessary for the measurements this guide is intended for, though i recognize some people are using it for speaker design too now.
 
Last edited:

as labs

Member
Joined
Mar 10, 2022
Messages
33
Likes
19
Location
hudson valley, ny
Loopback doesn't work with the Umik-1, which is what this guide was (primarily) designed for. I haven't tested the acoustic timing reference enough to know how reliable it is. It's not particularly necessary for the measurements this guide is intended for, though i recognize some people are using it for speaker design too now.
gotcha, makes sense! umik may have it's own abilities for arrival time reference, i haven't used it

in the case of a "normal" measurement microphone through a two-channel audio interface, loopback is nice in the case that you want to compare arrival times of two sources or measurement conditions.

I'm curious how the spinorama accounts for arrival time variations, is it reliant on pinpoint accurate set up of the spinner (i.e., driver in precise center of rotation)?
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia

I'm curious how the spinorama accounts for arrival time variations, is it reliant on pinpoint accurate set up of the spinner (i.e., driver in precise center of rotation)?

Unless it solves for the phase center as part of the process in a least squares fashion?
Seeing the mic wand move in azimuth and elevation as well as in-n-out would lead me to believe that it might be doing that.
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
I'm curious how the spinorama accounts for arrival time variations, is it reliant on pinpoint accurate set up of the spinner (i.e., driver in precise center of rotation)?
Yes a reasonably precise centre of rotation needs to be kept. The standard sets out what is required.

eB9b7Fi.png

Holmz seems to be describing what the NFS does which is totally different. In that a specific reference point is defined and then the robotics takes care of positioning the mic relative to it. That is about taking the raw data points to allow the Sound Field Separation to work. Everything else is then back calculated.
 
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
I measured all drivers separately. There are 75 db measurments for:

On Axis - 1 meter
Midrange - .20 inches
Woofer #1 - .20 inchers
Woofer #2 - .20 inches
Rear Port - flush

Napilopez, would you mind combining the measurements for a final on axis measurement? As the guide covers a single woofer only, perhaps you could post some screenshots so myself, and others, can learn how to combine measurements for a tower speaker. The baffle adjustment was posted in a previous post. I'm very curious to see if the curve matches the official Ascend Acoustics curve posted online. Thanks again for the awesome guide and your help!

I attached the 5 measurements. Due to file size, I put 3 of the measurements in the first file, and 2 in the second.

I'm sorry for the delayed reply, have been super busy. I hope the following helps!

I finally gave this a try today and struggled a bit at first making this sum because my results seemed quite different from Ascend's; the data you posted earlier was of the non-RAAL version. Honestly, was banging my head against a wall for a bit and thought I should retire from giving anyone speaker measurement advice =]

However, I remembered that some of the ascend acoustic speakers that have been measured by Amir have differed significantly from the measurements on ascends website, so you've gotta stick by your own results sometimes. As we'll see, that paid off here.

This is not to criticize ascend, as they were among the few manufacturers providing measurements at all -- especially back then -- and they now have a Klippel NFS, which is amazing. But I think between when those original measurements were made and now, our standards and expectations for matching measurements have changed significantly.

Here's how the Tower measures based on the measurements on their website.

Snag_14d5c37b.png


That doesn't include the bass (it's quasi-anechoic without a bass splice, only accurate down to 250Hz). For the bass, we can look at the non-ribbon-versions measurements, as it seems this part of the speaker didn't change in its response.

FULL.gif


So if we took the bass of the non-RAAL measurement, and the rest of the response above 300 Hz from the RAAL one, we'd get this.
Sierra official website splice.png


Welp, I was ready to give up as I was writing this post since my results based on your data looked very different:
sierra tower on-axis.png


But while writing this post I noticed Dave from Ascend has been remeasuring his speakers on the NFS. Those measurements are not available on the main site but they are on the Ascend forum. And here's how the RAAL tower measures on the NFS:
Snag_14e0c4be.png


Not quite a perfect match, but definitely looks a lot closer. Not too bad considering I didn't make the measurements myself and am guesstimating some details, plus unit variation or whatever.

Here's a direct comparison:

Sierra tower comp.png

The differences might be largely explained by microphone position; some speakers are very sensitive to measurement height, for instance, and that probably explains most of it. Measurement Distance can matter a bit more for towers too.

So here was my process plus some extra thoughts on discrepancies.

Because the midrange seems have quite a low crossover (significant output to 100Hz), it's likely contributing significantly to the bass. Ideally, I'd be summing everything manually.

I'd need to know the pistonic radiating surface of each woofer, so I'd want to measure the diameter of each woofer starting from roughly the peak of the surround. I see you noted in the data that the pistonic area was 4 inches, which sounds about right.

If we know the make of the woofers, you might also be able to find the pistonic area (Sd). For example, I'd assume the low-end woofer and mid-range woofers would vary slightly due to the different design and phase plug on the latter. But we're just aiming for a reasonable approximation here.

I'd also need to know the port diameter/area, which isn't usually a specification that's listed. You'd need to measure the diameter from the position in which measured the port, so if you measured it flush with the port's mouth/the baffle measure the diameter at that point. I'd also make sure all of these measurements were made without changing the SPL of speaker between measurements, to ensure the scaling works properly.

Still, I decided to give it a go based on the information you provided. Here was my process:
  1. Made a copy of each of the responses so I don't lose the original SPL values while moving data around and testing things (All SPL> Controls> Measurement actions> Response copy). The following actions were all taken on the response copies.
  2. Simple sum of the two low-end woofers without changing the SPLs, as they are the same size.
  3. Scaled the port response down to match the low end tail of the summed woofer response at about 10-15 Hz (make sure to click "add offset to data," if using the measurements action menu rather than the alignment tool)
  4. Summed the sum of the two woofers with the adjusted port response.
  5. Optional: I also introduced a bit of delay to the port measurement using the alignment tool due to the difference in arrival time at the microphone, which I estimated to be about 1.5ms. This is not something I cover in the guide because it almost always makes an extremely small difference (just creates the small dip at 190ish hz in this case, and that's more of a difference then usual). But it was basically simple trigonometry and arithmetic to estimate the difference in distance from the mic to the woofers and the mic to the port.
  6. Summed the midrange woofer response to the previous sum of the two woofers and port without adjusting SPL, as it is roughly the same size as the low-end woofers we are scaling the SPL response to.
  7. Multiplied that final sum by the baffle step correction from the bagby sheet.
That gives me this result for the total low-end response:
sierra bass sum.png


I then lined that up with the gated farfield measurement and merged at 300 Hz:

sierra NF FF.png


The differences in the upper midrange with the NFS measurement may be explained by measurement height. You can see from Amir's measurements that the Ascend Sierra Horizon Center (which uses the same midrange and ribbon) show some deviations if you are above or below the tweeter:

index.php

And again, it's also possible that ideally the midwoofer should be scaled a little differently as it's not identical to the low-end woofers.

But all in all, a solidly close result to the NFS measurement. Certainly, our results were closer to the NFS one than Ascend's own original measurements =]

I hope this helps! Give it a go using the steps outlined above (you can ignore the delaying the port response)

Could be interesting to compare my eyeballed methods with the mathematical method once you measure the diameter of the port.
 
Last edited:

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
How cool is that?! Having very little knowledge of acoustics, and with your guide, I was able to product a fairly accurate curve. Per your request, I have posted exact measurements of the woofer(s), midrange, and port (see attached).

For those of us that are new to this, can you please explain exactly how to measure the port and woofer area? I'm not sure exactly what part of the port and woofer is measured. Perhaps you could draw a line overlay on the pics below, so we know what area of the port and woofer to measure.

I am very curious to see if using the exact measurements vs eye-balling the response tails makes a difference in this case. If you don't mind, please post a layman version of the math required to do the calculations.

Thanks again for your help! This has been such a great community resource.
 

Attachments

  • IMG_6949[1].JPG
    IMG_6949[1].JPG
    277.8 KB · Views: 57
  • IMG_6946[1].JPG
    IMG_6946[1].JPG
    318.1 KB · Views: 59
  • IMG_6945[1].JPG
    IMG_6945[1].JPG
    163.5 KB · Views: 56
Last edited:
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
Bear with me as I tend to be a little verbose for the sake of understanding!

For woofers, the general guideline is to measure starting from the peak of the surround. some people say to use 3/4 of the way into the surround, but I like using the peak or halfway through the surround because it's just easier than trying to guess where 3/4s is.

An even more accurate way of doing the calculation would be if someone could identify the parts used, that way we can get the exact pistonic area from the manufacturer (a value dubbed 'Sd').

As for the port, I've found you need to measure it from the position at which you've placed the microphone. For example, if you place the microphone at the throat of the port -- right before the flare begins -- then you need to measure the diameter of the throat. But if you measure at the 'mouth' of the port -- where the flare ends -- then you need to provide the maximum diameter of the flare. Measuring at the throat will be louder, but as the diameter/area is smaller, it's measurement will be reduced further.

Since you said you measured the port 'flush,' I assume you mean flush with the back of the speaker, or the mouth of the port. So that's the diameter I will be referencing.

As all the sound sources are circular here, I'll use the simplified scaling formula using the diameters: 20 Log ([port diameter]/[woofer effective diameter]). Keep in mind that we're using Log base 10, not the natural log. Should probably clarify that in the guide. Also, we can really generalize that formula to 20 Log ([smaller sound source/largest sound source)]

Based on your photos -- and keep in mind it's hard for me to judge from 2d images -- it seems to me that the midrange woofer might be a tiny bit bigger than the low end woofers? Then again, I'm not sure how much the phase plug reduces the overall area. Another reason why it helps to have the exact Sd, but we work with what we have =]

For the sake of experimentation, I'm trying to be as precise as I can based off your tape measurement, but in practice, I'm being excessive. it's not really necessary to scale the midwoofer and woofers differently as they're both roughly the same size drivers; any differences will be in a fraction of a dB and you don't know what the real pistonic area is anyway. But let's have some fun.

These are the diameters I seem to have gotten from your images:

Mid-woofer: 4.25"
Low-woofers: 4.125"
Port flare edge: 3.375"

As the mid-woofer seems to be the largest by a hair, I'm going to scale the other sources to the mid-woofer's SPL. Keep in mind that because you measured both woofers separately, you don't have to worry about increasing your measurement by 6dB. That only applies if you're being lazy and only want to measure one woofer.

We will simply be adding the two woofer measurements manually, which will basically result in the same measurement +6dB. The woofers seem to measure a teensy bit differently for whatever reason, so it's good that you measured them separately to get a slightly more accurate response, but it's splitting hairs.

For scaling the low woofers, we have 20Log (4.125/2.45) = -0.259
So in REW, I will adjust the response of each woofer by -0.3 dB (REW doesn't let you be more precise than a 10th of a dB)

For the port, we have 20Log (3.375/4.25) = -2
So in REW, I will adjust the response of each woofer by -2 dB

(Pro tip: make sure to write down how much you scaled down each measurement in REW's measurement notes to the left, in case you need to undo it for whatever reason).

I leave the midwoofer alone.

Once scaled, everything looks like this:

Ascend tower nearfield scaled.png


Note that on a speaker with a single woofer, we'd expect the low-end tail of the port and woofer to roughly align. But because we have two low-end woofers, we have to add those two together first to see that expected result...


Ascend tower low woofer sum.png


As you can see, simply aligning the tail of the port with the response of the total low-end woofers is an easy way to avoid doing math.

Now I can continue summing the remaining measurements together. Next, I add the port to the low woofer sum.

(Again, this is optional, but I added a 1.5ms delay to the port using the alignment tool to account for the significant location difference between the port and the woofers (about 20 inches by my estimate).

ascend tower port woofer sum.png


Now I just add the midrange woofer's response to the earlier sum and get:

ascend low sum no correction.png


Then we multiply that result by the baffle step correction file...

ascend baffle corrected low end no delay.png


...Which is nearly identical to the "eyeballed" response, within ±0.5dB. Basically just a matter of where exactly I lined up the port tail. to the woofers tail

ascend mathy comparison.png

Although I recognize it takes some experience to have that intuition. So it doesn't hurt to try both eyeball thingsanddoing things the mathy way.

if we splice that with the on-axis at 300hzand compare it to ascend's NFS response we have...

ascend mathy comparison.png


Soooo yeah, not too shabby.

In case you're wondering, I do think might be possible to get more accurate results by doing baffle correction using the bagby spreadsheet and positions of individual drivers for each individual measurement... but the time's I've tried it, it's just been much more trouble than its worth for nearly identical results. So I just prefer adding the baffle step correction to the total sum.

We can't reach perfection anyway as all this nearfield stuff is just a reliable approximation anyway. If I can get results largely within a dB or two of an NFS result, I'm pretty happy.
 
Last edited:

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
My DIY speakers are now completed, but before adventuring in measuring 1m (H) x 40cm (D) vs 40cm (W) @ 40kg speakers, i decided to try this process with my XTZ 99.25 MK1 speakers.

I currently don't have a way to elevate the speaker, so i've used a 60cm Atacama stands.
Gear:

Goals:
  • Be familiar with all tools, all configurations, etc
  • Did the same process Indoor, and Outdoor.
  • Only did Horizontal measures due to instability of the stand.

Nearfield Bass Measurements & Jeff Bagby's Diffraction & Boundary Simulator

Speaker dimensions: 244 x 379 x 352 mm (W x H x D)


ikiegZq.png



lptMOk7.png



Measure and Measure after correction

  • Measure was only taken indoor, as i forgot to do the same measure outdoor.

OWEOPcY.jpg





Indoor Measures:
  • An attic with very low ceilings, measures taken 0.5m from Speaker.
Iuf8PC0.jpg



Indoor REW Measures:

265H0VJ.jpg



Merge NF Fixed Bass Response with Indoor Far Field measures

vcGFrQp.png



Indoor SPL, GD & Phase

QTUoJH5.png



Indoor Power & DI and Dir (Hor)

uANuzxI.png



Outdoor details:
  • 7m x 7m yard, measures taken 1m from Speaker.
6zgj0NI.jpg



Outdoor REW Measures:

bopjBZd.jpg


Merge NF Fixed Bass Response with Outdoor Far Field measures


uX9Fgee.png



Outdoor SPL, GD & Phase

0NvbJQL.png



Indoor Power & DI and Dir (Hor)

mZFOG2J.png


All Measure Files available HERE (XTZ - All files.zip):
  • Measurements Files
    • XTZ.mdat => Indoor Measures
    • XTZOUTDOR.mdat => Outdoor measures
    • XTZ-BaffleDiffraction.frd => BaffleDiffraction calculated file
  • VituixCAD File
    • XTZ_Indoor.vxp
    • XTZ_Outdoor.vxp

Now, lots of questions :)

  1. Looking at both Indoor and Outdoor measures, the speaker elevation height seems the most important factor, and 60cm is obviously not enough,
  2. I was expecting the Outdoor measures to be much reliable than Indoor, but it seems the Indoor (at 50cm) have higher reliability than Outdoor (at 1m).
  3. I'm finding strange that the Frequency response from 200-20000Hz is higher than 20-200Hz! Is this due to an error on my nearfield measure? I followed the exact VituixCAD instructions
    1. Or could it have been an error on the Merger?
  4. What's happening with the phase on the Outdoor measures?
  5. Is there any step that i've messed up?
 
Last edited:
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
I will take a closer look at the data when I can, but a few things stand out to me.

First, yes, speaker height is extremely important important. The floor is a wall, and having it close to the speaker affects the sound as much as having any other nearby wall. Measuring outdoors makes no positive difference if the speaker is at the same height, because the nearest wall is still the same distance. In fact, because you measured from a further distance, the measurement becomes even less reliable.

Taking a look at the impulse response, there are some bumps and such within the 5ms window so the data doesn't seem as clean as it should be.

That said, the outdoor measurement can still be useful as a reference. if you remove the gating and smooth the measurement a bunch, like 1/3 octave, you should get a rough depiction of the speaker's anechoic contour. For example, with1/3 smoothing the outdoor measurement shows:
1649112568795.png


That leads me to believe that the dipped bass response is how the speaker actually measures rather than an issue with your measurement.

The other thingthat stands out to me is that the spinorama curves don't roughly align in the bass as expected. For a speaker this size, it should show a roughly omnidirectional response in the bass, which is not shown here. Sometimes VCAD has trouble guesstimating the bass directivity based on the quasi anechoic farfield measurement, especially if the gated data doesn't go very low, in which case you can use the "force to gradient" option, although all that really does is make the graph prettier and more accurate rather than inferring anything from the data itself.


Based on your outdoor photo, it looks like the speaker is far forward on the turntable. The baffle should be directly above the axis of rotation.

I realize this can be difficult to achieve with a heavy speaker and stand but it's really crucial for proper directivity measurements. The CEA 2034A document suggests the reference axis should not be more than 2 inches away from the axis of rotation, but as we are measuring from 1 meter, it's even more important that it's closer to that axis of rotation. If the position you used in the image is how you performed the measurement, then the directivity measurements are invalid.

I hope this doesn't come off as critical. The most difficult part of these measurements is really just positioning the speaker and stand, so trying to get those issues out of the way.

You could try simply placing the whole rig on a sturdy table, by the way. That's what I do. I'm using a folding table similar to this:

 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
I will take a closer look at the data when I can, but a few things stand out to me.

Thank you, but given the problems that you already identified, please don't spend too much time on this.
I'll have to address them first, before taking a new set of measures.

First, yes, speaker height is extremely important important.

Measuring outdoors makes no positive difference if the speaker is at the same height, because the nearest wall is still the same distance. In fact, because you measured from a further distance, the measurement becomes even less reliable.

You totally nailed, i need to address this first.

The other thing that stands out to me is that the spinorama curves don't roughly align in the bass as expected ... Based on your outdoor photo, it looks like the speaker is far forward on the turntable. The baffle should be directly above the axis of rotation
Correct, i did it due to stability issues, so that's the second thing i will need to address.

I hope this doesn't come off as critical. The most difficult part of these measurements is really just positioning the speaker and stand, so trying to get those issues out of the way.

Not at all, i just want to thank you for your kindness and patience to address all this questions.
So, thank you :)

I'll return to this topic for round 2 once i'm able to take care of the identified problems :)
 

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
<< For scaling the low woofers, we have 20Log (4.125/2.45) = -0.259
So in REW, I will adjust the response of each woofer by -0.3 dB (REW doesn't let you be more precise than a 10th of a dB)>>

Where did the 2.45 number above come from? Also, I want to make sure I am doing the calculation correct:

1. Divide 4.125 by 2.45.
2. Hit the log button.
3. Hit multiply X 20

I am getting 4.52 so I must be doing something wrong. Thank you!

Also, would you mind posting the final REW files for the before and after math so I can check my results? Thanks!
 
Last edited:

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,597
Likes
2,235
<< For scaling the low woofers, we have 20Log (4.125/2.45) = -0.259
So in REW, I will adjust the response of each woofer by -0.3 dB (REW doesn't let you be more precise than a 10th of a dB)>>

Where did the 2.45 number above come from? Also, I want to make sure I am doing the calculation correct:

1. Divide 4.125 by 2.45.
2. Hit the log button.
3. Hit multiply X 20

I am getting 4.52 so I must be doing something wrong. Thank you!

Also, would you mind posting the final REW files for the before and after math so I can check my results? Thanks!

Typo - should be 4.25 (the diameter in inches of the mid-woofer).

Edit - your arithmatic is fine.
 
Last edited:
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
Ah, sorry about the typo! Not only can I not spell words, I cannot spell numbers correctly either =]

Will do the REW files when I can
 

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
I do have a few questions about the procedure you posted above for measuring a tower.

1. <<I'd also make sure all of these measurements were made without changing the SPL of speaker between measurements, to ensure the scaling works properly.>>

I try to take my on-axis measurement at 85db. My understanding is you then measure the port because it will be the loudest. My question is at what volume should I measure the port and subsequent drivers at? I was getting peaks of 120db. Is there a recommended guideline for the individual drivers and ports?

2. You added some delay (-1.5s) to the midrange to compensate for the port being located roughly 20 inches below the woofer. You mentioned it involved some math. If the math is complex, I would rather skip it. Is there a way to eye-ball or estimate the delay? Also, I wasn't clear on exactly how you added the delay. I choose the alignment tool but I wasn't sure how to populate both the top and bottom fields and what button to click to apply the changes.

3. I have another speaker on-hand for us to try out, a Arendal 1723 tower (Tower link). The speaker is unique in that it has four, eight-inch drivers that all cover basically the same frequency range as well as three ports. Before summing the port response with the summed woofers response should I sum all 3 ports together first? They measure nearly identical with a +/- .5 DB variation. Also, for my on axis one meter measurement would you recommend aiming at the tweeter's center?

Thanks again!
 
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,110
Likes
8,433
Location
NYC
I do have a few questions about the procedure you posted above for measuring a tower.

1. <<I'd also make sure all of these measurements were made without changing the SPL of speaker between measurements, to ensure the scaling works properly.>>

I try to take my on-axis measurement at 85db. My understanding is you then measure the port because it will be the loudest. My question is at what volume should I measure the port and subsequent drivers at? I was getting peaks of 120db. Is there a recommended guideline for the individual drivers and ports?

2. You added some delay (-1.5s) to the midrange to compensate for the port being located roughly 20 inches below the woofer. You mentioned it involved some math. If the math is complex, I would rather skip it. Is there a way to eye-ball or estimate the delay? Also, I wasn't clear on exactly how you added the delay. I choose the alignment tool but I wasn't sure how to populate both the top and bottom fields and what button to click to apply the changes.

3. I have another speaker on-hand for us to try out, a Arendal 1723 tower (Tower link). The speaker is unique in that it has four, eight-inch drivers that all cover basically the same frequency range as well as three ports. Before summing the port response with the summed woofers response should I sum all 3 ports together first? They measure nearly identical with a +/- .5 DB variation. Also, for my on axis one meter measurement would you recommend aiming at the tweeter's center?

Thanks again!

1) Your understanding is correct. I often measure at a measly 75-80 dB so there's enough 'headroom' to not touch the volume and then measure the port and subsequent drivers. However sometimes it's just too loud so you may have to reduce the overall SPL when you start measuring the ports nearfield. This is usually not a big deal on passive speakers as the FR is unlikely to change significantly, but on active speakers it's something to be wary of.

2) I want to preface this by saying that I think it's rare for people to factor in a delay between the port and the woofer when doing quasi-anechoic measurements. Other than I think @ctrl , I don't think I've seen anyone mention doing it, but it seemed intuitive to me.

In this case the difference is just the presence of that dip at 190hz, and that's more than usual. It's a bit more noticeable than usual because the port is so far from the other drivers on this speaker.

Here's the summed response with (blue) and without (red) the delay:

1649129691238.png


No math required if you just manually measure the distance from the port to the mic and the distance from the woofer to the mic and subtract for the difference. Otherwise it's just some Pythagorean theorem and arithmetic.

I guesstimated the difference in distance from the woofer to the mic and port to the mic to be roughly 20 inches. You'll notice that if you move the delay slider in the alignment tool, it'll tell give you a distance associated with that delay. 20 inches is roughly 1.5ms. You could measure the real distance/delay, but that seemed to have been enough of an approximation.

In an ideal world you could also calculate the delays for each of the woofers too before summing but it's really overkill.

3) It seems like the arendal tower has two crossovers, one at 120hz and one at 1500hz. So i assume the 3 bottom woofers are covering the low end, and the top one is covering the midrange?

I would measure each woofer and port and just sum them in REW. You could add 6dB for each extra port but imo might as well just sum them as we did with the dual woofers of the ascend tower.

Since each of the woofers and ports are the same size, I'd think I'd follow this order

1)Measure each woofer and port
2)Identify the woofers that cover the low-end crossover (120Hz and below)
3)Sum those low-end woofers together
4)Sum all the ports together
5)Reduce summed port SPL to match low end woofer sum tail
6)Sum scaled port response with summed low end woofer response
7)Sum midrange woofer(s)
8)Apply baffle step correction.

I think that should get you at a good result, but you can confirm things the mathy way if it doesn't match arendal's measurement.
 
Top Bottom