• 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
Attached are the on-axis and individual driver measurements.

The hi-end response deviates quite a bit from Ascend's official chart. However, I did find some independent measurements that are much closer to my measurements. I've attached some response images below for a baseline below. Unfortunately, I don't believe there are any Klippel measurements available.

1649139484786.png
1649139547466.png
1649139588396.png


I followed your suggestions above, however, I get some wonky bass response that is too loud and rolls off to fast. I'm not sure what I am doing wrong. Interestingly, the top two woofers extend to 1500hz and the bottom two extend to 120hz.

My steps:

1. Combine the 120hz drivers.
2. Sum all three ports together. Is the correct way to combine the three ports to go to trace arithmetic and sum ports #1 and port #2 using A + B? Then repeat the steps and use the Sum of ports 1 and 2 plus port 3, again using trace arithmetic and A + B? I summed the three ports using this method.
3. Then I reduced the summed port response by going to measurements actions, entering the negative offset, and then clicking on add offset to data. I'm curious what offset you would use. Do you align just the very tip? Or line them up around 15hz where they line up better? I aligned just the very end tip of the response curves.
4. I then summed the scaled the port response with summed low end woofer response by clicking on trace arithmetic A + B.
7. I then summed the two midrange 1500hz drivers by clicking on trace arithmetic A + B.
8. I think you left out a step on what to do after summing the midrange drivers. I simply added the summed midrange drivers with the summed port response + woofers using trace arithmetic A + B. I assume this is what I was supposed to do with the summed midranges.
9. I then entered filed out the baffle correction spreadsheet using the following settings:

1649140588467.png


I wanted to make sure I filled this out correctly. When you are measuring a tower speaker what should you put for the speaker location fields? I just used the lowest driver's location on the speaker (the bottom woofer). I then entered a 1" baffle radius as the sides (only) are curved in about an inch. Lastly, I filled in the piston's diameter by measuring 1/2 into the surround on one side to 1/2 the surround on the other. My understanding is the surround is the portion that sticks out in a circle relative to the rest of the driver. In my case I measured 6.63 inches.

10. I then imported the baffle response and clicked on trace arithmetic A * B (with A being Sum of the bass response and B being the baffle measurement).
11. I then spliced the files at 210hz as there seemed to be a natural match there between the on-axis and bass response.

My final response:

1649143389734.png


Like I said, my bass response seems to be way off and with a strange dip compared to the examples I provided. I'm guessing I added the two midrange responses incorrectly into the final response. I would imagine having two woofers, two midranges, and three ports also complicated things.

Thanks again for your help! I'm curious yet again how close our response will match to others. I'm also going to run the Ascend on-axis at a slightly higher position when I get time to see if we can match the response even closer than we have.

*** Update ****

I added the sealed woofer response files (Arendal 1723 Sealed (1) and (2) files). I sealed all three ports and measured the woofers. If we take the math approach the three ports are 2.56" wide and piston area of the four woofer is 6.63".
 

Attachments

  • Arendal(1).zip
    4.1 MB · Views: 79
  • Arendal(2).zip
    4.1 MB · Views: 81
  • Arendal(3).zip
    4 MB · Views: 67
  • Arendal 1723 Sealed(1).zip
    2.8 MB · Views: 46
  • Arendal 1723 Sealed(2).zip
    2.4 MB · Views: 47
Last edited:

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
<< 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 haven't posted my off axis response yet, however, I think I did my off-axis measurements wrong. I had the baffle of the tower speaker aligned with the front of turntable. If I understand correctly, the baffle needs to be lined up with the middle of turntable, not the front. With my speaker beings 16 inches deep I'm afraid the speaker might fall right off the turntable. I bought the turntable you recommended which is only 15.5 inches. If I positioned the loudspeaker halfway that would mean only 8 inches are on the turntable and 8 inches are hanging off. Ugh. I might need a bigger turntable.
 
Last edited:

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
So, round 2 :)

This time, and based on these topics, i decided to try if Ground plane measures would allow me to avoid the complexity of elevating my monster speaker boxes:
- Measuring Response: How close is "close enough" to anechoic?
- ErinsAudioCorner - (Ground Plane measures)

This isn't an 100% accurate setup, but only a test to check if using ground plane measurements would be possible for producing all measures needed for crossover development.

Based on the above topics (with the caveat of anything wrong due to some misunderstanding on my part), i tried the following combinations:
  1. Mic and speaker flat on the ground,
  2. Mic flat and speaker tilted
  3. Mic tilted and speaker flat
  4. Mic and speaker tilted.

On measures:
  • All measures were taken with mic and speaker with a 2m distance.
  • 5 measures were executed for each variation, with the following settings to detect variations from wind, outside noise, etc:
    • Length 1M, repetitions: 2
  • As per VituixCAD rew instructions, i use a time offset for 2m (in the Vituixcad instructions it shows the offset for 1m):
NKUT7tp.png



1) Mic and Speaker flat on the ground

rHk6Hh9.jpg


Measures

jSolufJ.jpg





2) Mic Flat and Speaker tilted


Speaker tilted detail:

nSz6b3z.jpg



Measures

8uXynKQ.jpg



3) and 4) Mic Tilted

All variations with the Mic tilted, produced unreliable measures.

A detail on how Mic was tilted:

QN1Yenz.jpg


A measure with both 3) and 4) variations:

OkF0LO7.jpg



Mic and Speaker flat on the ground (1), VS Mic Flat and Speaker tilted (2)

Assuming 1), and 2) produced reasonable accurate results, here's a comparison between both:

W9Ytxvj.jpg



Given how similar both measures are between 100Hz to 1000Hz, i compared them with yesterday's Nearfield measure with baffle compensation, and they are reasonably similar:
AJ96iOL.jpg




I've found an online measure for the XTZ 99.26, i adjusted the X and Y scales, and the measure is very similar to these ones:

Iw0f91u.png



Finally, a comparison between the ground plane measures, and the Indoor measure at 50cm (which showed greater accuracy than the outdoor ones at 1m), time aligned and level matched at 8kHz:

MUt8E0T.jpg



Conclusions / Considerations (could be wrong :))
  • Ground plane setup, i.e., the mic to speaker distance and angle, were done by eye, so i assume if using a laser to measure distance and angle, results would be much accurate,
  • Speaker Tilt vs no Tilt produces the same results up to 1000H,
  • Ground plane measures seem much more accurate than regular measures at an incorrect height (as i did yesterday at 60cm height).

Questions:
  • Have i failed anything on the Ground plane measures?
    • When tilting the Mic, i tilted it downwards. Should i have tilted it upwards, angled to the center of the tweeter?
  • Would it be worth to do new Ground plane measures, but using an accurate setup (laser measuring and angling, etc) ?

What's my goal ?

My goal is to measure my DIY speakers, that are composed by 3 isolated boxes for each driver, in order to use the files to design the active crossover:

cNXrgVr.png


To see how big they are:

rF3ufrP.jpg


HwG8YlG.png


I'm currently waiting for the delivery of the 15" driver (expected to be delivered by June), so meanwhile, only the first two speakers are concluded:
TkWHYoC.jpg


The TPL-150H box weights 10kgs, and the TD12M box weights 30kgs.


What to do next?
Basically my question is simple, how can i produce measures accurate and reliable enough for developing a valid crossover?
Can Ground plane measures provide the required accuracy? And if so, what approach to use? Tilt or no tilt on the speaker?
Moreover, it would be very simple and safe to do a lazy Susan turntable to use for off-axis measures for GP, as Erin did here:

EBWhcgx.png




Open to comments, corrections, ideas or opinions on what how to proceed :)

As mentioned, ultimate goal is to have an accurate and secure process for performing the measures for crossover development in VituixCAD.
The most important part for the XO development will be the selection of crossover points and time align drivers.

EDIT: Given that based on the TPL-150H and the TD12M characteristics, i'm expecting a XO point between 1200-1600Hz, so this would be the frequency interval where i would need maximum accuracy. Given that both methods (Speaker tilt and flat), for this specific region, produce results with less than 1db of difference, would it matter which method to choose (just for XO selection point) ?

Ground plane MDAT project available HERE.

Thank you!
 
Last edited:

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
I think you want them 5 feet up to get rid of the bounce within 10ms.
The only tilt of the speaker I would consider is 90 degrees, but then you still have the back lobe.
 

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
I think I'm still confused by how to measure the "speaker piston diameter" for the baffle compensation spreadsheet. In the pic below do I measure halfway from the blue line or the yellow line? Thanks!

Surround.jpg
 

Bghead8che

Member
Joined
Jul 1, 2019
Messages
29
Likes
6
Looks like 4” to me. Blue line.
It’s just a circle…
Sorry, I should have been clearer. I'm asking where to start the measurement for the piston area... half-way into the blue line or half-way into the yellow line? If we measure from the blue line the piston area is 4.5 inches. If we measure from the yellow line the piston area is 4.0 inches.
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Based on the size and weight of my diy speakers, elevating them into the required height for proper measures is neither safe nor practical.

Based on the research made by Erin, it seems that ground plane measurements should provide the required resolution for a proper crossover development.

Nevertheless, i would like to have a way to compare myself a proper open free field measure at a considerable height, versus a correct configured ground plane measurement in my backyard, to see how they compare, to understand how well they would track.

A picture of my backyard:

zdKvZZy.png


Basically, its a 7m x 7m yard covered in ceramic tiles, which should provide a reflective surface for proper ground plane measurements (the measures i posted in previous posts, even if setup by eye, seem to confirm this).

For a proper free field measure with the required height, it would be easy to use a ladder to place the XTZ 99.26 into the position marked as 1), which is at 4m from the ground.

Sorry for the direct tag, @hardisj @napilopez @fluid, based on your knowledge,
Would a speaker measured in this position provide a correct free field measure?

Thank you in advance!
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
Basically, its a 7m x 7m yard covered in ceramic tiles, which should provide a reflective surface for proper ground plane measurements (the measures i posted in previous posts, even if setup by eye, seem to confirm this).

For a proper free field measure with the required height, it would be easy to use a ladder to place the XTZ 99.26 into the position marked as 1), which is at 4m from the ground.

Position 1 is not usable for measurements because the wall below the DUT acts like a baffle enlargement and will therefore falsify the measurements in the low frequency range (in the low frequency range the frequency response would be raised).

Measurements of the tweeter would probably work on the wall ledge in position 1 (the low frequencies would be raised and vertically the edge diffraction changes at low frequencies, but this does not matter in this case), but only if the tweeter is as far away from the wall ledge as possible (e.g. big speaker HT at 1m height).


If the backyard floor is level and smooth, then a gated GP measurement should succeed. At 7mx7m and 2m measuring distance, a gate with up to 14-15ms should be possible - If you place the DUT and Mic so that all reflective surfaces are as far away as possible.

So a measurement down to 80-100Hz should be possible - but in this range the measurement is extremely smoothed. So you need a near field measurement of the woofer/BR port/... and must combine them with the gated GP woofer measurement.

From the Arta tutorial/handbook about GP:

1649534828533.png 1649534845648.png
 
Last edited:

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Position 1 is not usable for measurements because the wall below the DUT acts like a baffle enlargement and will therefore falsify the measurements in the low frequency range (in the low frequency range the frequency response would be raised).

Measurements of the tweeter would probably work on the wall ledge in position 1 (the low frequencies would be raised and vertically the edge diffraction changes at low frequencies, but this does not matter in this case), but only if the tweeter is as far away from the wall ledge as possible (e.g. big speaker HT at 1m height).


If the backyard floor is level and smooth, then a gated GP measurement should succeed. At 7mx7m and 2m measuring distance, a gate with up to 14-15ms should be possible - If you place the DUT and Mic so that all reflective surfaces are as far away as possible.

So a measurement down to 80-100Hz should be possible - but in this range the measurement is extremely smoothed. So you need a near field measurement of the woofer/BR port/... and must combine them with the gated GP woofer measurement.

From the Arta tutorial/handbook about GP:

View attachment 198789 View attachment 198790
What is considered farfield exactly? Because from the document you shared they can just reduce the measuring distance to 1m and do a regular non-gated GP measurement without any loss in resolution.
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
What is considered farfield exactly?
The distance of the DUT must be large enough to include all possible effects (e.g. baffle step, edge diffraction, ...) of the loudspeaker cabinet.
The rule of thumb suggested in the Arta handbook is:
1649538041079.png

But at a distance of 2m the effects should be minor for most loudspeakers and mainly affect the low frequency range, which is extremely smoothed anyway when using a gate.

I can recommend anyone interested in measurements to skim the Arta handbook, as it repeatedly contains generally valid instructions and explanations.


Because from the document you shared they can just reduce the measuring distance to 1m and do a regular non-gated GP measurement without any loss in resolution.
The duration of the gate would be slightly increased by reducing the distance from 2m to 1m, but the area of 7mx7m is too small to perform measurements without a gate. Decisive for the measurement resolution (and whether a gate is needed) is the distance to the first reflection .

To be able to measure down to 20Hz, the time until the arrival of the first reflection must be at least >1/20 seconds, i.e. 50ms, which means that the first reflecting surface should be at least around 10m away - but even then a gate is still needed and the result is strongly smoothed.

And 1m distance is again problematic for larger speakers, since not all cabinet effects are included (using a short measurement distance).
 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
267
Likes
262
Location
Portugal
Position 1 is not usable for measurements because the wall below the DUT acts like a baffle enlargement and will therefore falsify the measurements in the low frequency range (in the low frequency range the frequency response would be raised).

Would this happen if the speaker were measure using the balcony behind the rooms, placed in the wood rod, thus 4m above the ground?
The (1) marks the floor of the balcony, which is 1m deep, so there would be a wall 1m behind the speaker:


pIFi3nN.png
 
Top Bottom