• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily 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

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,151
Likes
3,829
Location
.de, DE, DEU
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:

The first thing you always have to ask yourself when measuring a loudspeaker is where is the nearest reflecting surface (even "small" surfaces count: chair, table, box, observer standing around, grill, protruding ceiling beam, cat, dog, mutated spiders, hollow man - not sure, ...) and how much longer is the time of flight of this first reflection compared to the direct path from the loudspeaker to the microphone - where the reflecting surface is located (top, bottom, left, right) doesn't matter.
Surface size does matter, but even small surfaces create a footprint in the frequency response.

The difference in time of flight of the first (closest) reflection, minus the direct time of flight LS to Mic, gives the gate for the quasi anechoic measurement.


If we look at the measurement situation on your balcony from above (view from the top), the following results:
1649599678065.png

The first incoming reflection has a 2m longer path than the direct sound (LS to Mic).
This results in a gate of 5.8ms and a reflection-free measurement down to 170Hz (but there with very strong smoothing of 1/1oct at 170Hz).

Update: And as said above, any increase in baffle size will affect the measurement, so the floor height of your balcony would also affect the measurement a bit.
 

eddy555

Member
Joined
May 29, 2021
Messages
17
Likes
10
when summing port and nearfield, do you have to manually adjust for the delay difference between the two before summing? I use loop back and when I look there is a 3 fold distance difference between the woofer and port. I mean time of flight.
 
Last edited:
OP
N

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
1,989
Likes
7,738
Location
NYC
when summing port and nearfield, do you have to manually adjust for the delay difference between the two before summing? I use loop back and when I look there is a 3 fold distance difference between the woofer and port. I mean time of flight.

I haven't used loopback for these measurements (again, just living with the Umik-1 here) so the only delay I really account for when necessary is the difference in distance from the woofer and port to the farfield microphone location. It never changes the overall low bass contour much, but sometimes reveals a peak or dip in the lower kids/upper bass.

What really matters here I suppose is the overall phase shift in the summation.

Is there a significant difference if you try to sum the woofer and port measurements with loopback on vs with loopback off? Curious
 

eddy555

Member
Joined
May 29, 2021
Messages
17
Likes
10
hi,

there's a difference above 300hz and up. As I understand above the tunnig freq port and woofer are in phase?
here no adjustment vs some small adjustment in the delay with the alignment tool. The graphs are with fullspace correction curve applied.
no adjustment makes it sum more positively above 300hz. It's an 8 inch woofer.
 

Attachments

  • nrfld.png
    nrfld.png
    108.4 KB · Views: 17

gabo4au

Member
Joined
Apr 18, 2022
Messages
34
Likes
20
In case anyone else runs into these problems, I thought I would drop a note here.

When I used the spreadsheets for diffraction using Libreoffice calc as I don't have excel, when I hit "save baffle diffraction response to frd file" I get nothing. So I couldn't figure out easily how to save the results.

Which was a good prompt to go try Vituixcad's diffraction tool as I know it's a better thing to learn anyway. Vcad is such a great tool, but for us simple minded guys, it is sometimes overwhelming to use! I don't do this every day, sometimes it's months between doing it. Every time I load up Vcad after it's been a while, I struggle with it. It's very complex and has so many great tools included.

So if you're a beginner at using the Vcad diffraction tool... Just be aware that after you put in all your parameters, baffle width/height/corners/etc., driver diameter/count/number/spacing. Distance/angle/ from your measurement, etc.

Then you have to hit "NEW" to get it to plot and show you the results! Took me forever to figure out how to get it to "calculate" the response so I could then save it to import into REW. There is no calculate button and no real time updating as you put in parameters like the spreadsheet. The "new" button effectively "IS" the calculate button!

Yes, I'm a bit slow :) And for us slow people, "new" usually means "clear out all the calculations and start over!" haha I got it now!
 

gabo4au

Member
Joined
Apr 18, 2022
Messages
34
Likes
20
So I tried my hand at this today. For now, just the on-axis measurements to make sure I'm doing all this right. In my shop I have an ancient pair or Proac Tablette 2's! The Tablettes are quite legendary, here is the history of these little speakers.

1979 – named changed to ProAc and first Tablette released to international acclaim
1987 – Super Tablette introduced
1989 – Response series first released
1991 – Tablette 2 introduced, Response series updated to ‘S’ models
1993 – Tablette 3 introduced
1997 – Response updated to ‘SC’ and floorstanding 2.5/3.5 models released
1998 – Tablette 50/50 Signature introduced
1999 – Tablette 2000/2000 Signature introduced
2000 – Future series released, the first models to use ribbon tweeter
2001 – Response ‘D’ series introduced to great acclaim, new Tablette Reference released
2007 – ribbon tweeters introduced to the Response ‘D’ series with introduction in D38R
2008 – Carbon Pro series introduced, ProAc’s largest and most expensive designs yet
2010 – Tablette Anniversary released to celebrate 30 years as ProAc
2011 – K6 released, the first speakers to use the newly available Kevlar cone material
2015 – K8 released bringing Response series to its culmination
2016 – Tablette 10 released
2017 - DT8 & Tablette 10 Signature is released

I think I purchased my 2's around 1993 or '94 on a close out as by then they had the Tablette 3's in stock. Even though these speakers all share the same name, they are all a bit different. I believe the latest incarnation is a sealed enclosure with completely different drivers, but hey the box is a similar size for what that's worth :)

Here are a couple of pics, front and back.

krqXuWhRtOXCEun7d2WoNn9lTYVap_-dsx9XWi0r1b0AA8SZt2UFeEX7siqV_jQxlpt5C6WALNwPvEmmwive53oQY7ZL1caIkqOrU_sui3hKJTfc8ClRw6Vqsx_WBoeLMx3bduQA2u-EkXGy13KhkZTfTikZvDJA6HzNx_7Nc7zXafGbzr3LLJZLCP9t7fPfYUQ6TDxlmWAh7vILp39YKoypIbisKGDabrPcaVRpoY-OGFed-H6cH_UuauvVJxwWJRscWi6p5aT0BezsG7kTShuHJjH_IcDro58d2JZRihV49pzuINqvmMAi0tcdhd65saKpiMvxJEWy3cGhdVV04W4fx28pkCQEIAJbM5rAfeHnG8sJZX8K58NEGVyCXXcZouuBMGv3V5NuOudOHGkD66NjEJgJbqz8vIVGT4kHo6SQOzEGAtu-HWdtC6tJlaPYHF-ByJXyUNcTSsdNV0Q23CTkUH24MVTE4jUvecqNq5h3vxk5P9JQoERleZS67RLbNBDlSZ2y-sBiL_hOnrLixGDMVUIYKTDpcKqM6tW19uU7O37hcjKW1_wTffw0UkM9zC5kNpINpWjFBOH8ALZcVH3OiV094D7F02aOIiHr-UDgERTsNGOJoJnl86XbMH59_KOkNkw16kHsiUMUyyyYk3vEHvONs6Ks6hPb4gMGnXIMyJuDxcbt2gs-8Pm2R8jMOzdh10kRkGSVyL3K5LK_blJDJKC4LuC_SojLArBkdWo3N4-YQOLEuLnj6bl9CO-98YPp58_WnJVoCUkEVErrimGpY8sCR_2SuFRG=w463-h822-no
-So2V5H_40fuFQ_rn9iB77o-fUTLjMeJJrJxyiJqu-x3j04Bkcj9fD8-bbKdi5jeK47FPDa3qNvzGkfWWQkWVOrgoJurmQbFJExGCwgax1nsGZYvEaZRExfBRWRXiTtjiaqYSlKor1LlQXPyux_vK7ElCXx1gWXf6Q00SbpDRZRdSQvr5T1PlNLk45refe0_nS05Y15Or788-TKbzB_TICq4W2MEBUxTHiC2_y9Wzl1LZV9635tZw8XK3T9PgvO9nPoQU4fh5A8CojQ1jNzlfA-YeVXXW6jakR2AqWPh8QDXNXVYa1NdEMlz7Qt2I7McB3QuSxf9WOvMgQD41XiWV32QhaX8liY-Gp9cxh29xT-MZ2Aby4L1oDDA8TpcvoYaqiXkZcDJyiJUzass6VOqHYmhlUgo-jpTJFmcXBYoL7L638_Om3om7qiF1u11CtHhgtag-RSwJ5AjwLkG-IExJhDUHdEPPh2IWwE198Z8s5XRwbKyKZ7WKd7tQ_UPOSK7IyDnlBaSUnoDGh72okxey3PNat7FQG-5ayG-IADKWPrwd3hBiOioC399FYp4lxTG5g_ubAg9NB7thXMO7cu4paN3XrpZRsvk3LFFVGo2Ci9jBH7eoLwMggLkvKLmbDRTQDuBm0q4X_khGux80UkNy4niEI8E9FEptKKDd2xJ-0C2eYXCP1UbfNB2_Q0Ei0ZnerWT8Qi98RyNR6_eRTjJis2IO9o3Ai1gmpdA7JDCO_x6sgfzQzYc-LBeQGfrk-hGwISVbU6cLuHnPC8ixkBU6Zdl5xTbIZmANOL6=w463-h822-no



For my hardware setup, I have an older Edirol UA-25EX Interface that I use for this. I've done the interface calibration and it works pretty well. For a mic I have a Earthworks M23R mic, with the calibration file, although this mic only varies about 0.5db so it could be used without it. I also use one channel as a loopback interface wired directly from output to input. I only measured down to 40hz, I know there isn't anything that low and I didn't want to damage them.

Here's my first measurement, I took it at 18 inches instead of 1 meter. I don't exactly know why I did that! haha But it's a pretty small speaker so I think that's fine.

ProacT2-18in.jpg



Then I applied the IR Window at 6ms which seemed like a good place. My shop is pretty big and has a 9ft ceiling. Here are the results.

ProacT2-18in+window6ms.jpg


Next up my nearfield woofer measurement at about 1/4 inch.

Proact Woof at Dustcap.jpg


Next I measured the port, which is on the back of this speaker and is 2" round.

Proac T2 Port.jpg



Next I summed the woofer and port. Since my graphs do not have a nice "alignment point" like the example, I used the calculation method that was mentioned, which was 20 Log (port dia/piston dia). The port dia is 2" and the piston is 3.375", so 20 Log(2/3.375) which gives me -4.54db. When I put that in and added the graphs, here's what I got.

Proact Woof+port.jpg


Next I used Vituixcad's diffraction tool to arrive at the following compensation.

1651013956349.png



Then I imported that into REW and Summed it with the Nearfield and Port summed graph.

Proact Woof+port+Diffraction.jpg


Finally I can align the Windowed version with the Woofer+Port+Diffraction.

ProacT2-Full.jpg


And the final On Axis response.

ProacT2-Final-OnAxis.jpg





While I think I followed the instructions correctly, to my eye/ear, there seems to be something a bit off. The overall response seems about right with the exception of the bass extension. This method shows the -3db point being at around 64hz with the -10db point being around 45hz.

I can tell you, they sound great, but there is NO WAY these things go that low! To be honest, the original non-windowed measurement seems more correct, at around 80hz it shows about 4db lower output than the final merged graph.

So the question is, what did I do wrong? The piece I'm wondering about is the summation of the nearfield + the port. I reduced the port by using the formula which produced -4.5db. But that port is on the back of the small speaker, seems logically to me it might be producing less out in front of the speaker. If you took the final graph and hinged down the line from about 300db taking it down about 3db at around 100hz I think it might be pretty accurate.

So went back and just tried to lower the Woofer+port+diffraction graph by another 3db, then summed it again with the Windowed version. Oddly enough, it didn't really do much of anything. So I don't know what I did wrong.

But this exercise gives me a lot more respect for all you guys who do this so well! It's a tricky thing at least when you don't have any experience.


Edit: Just for fun, here's the original response at 18 inches, with an IR window of 25ms, which is 40hz, overlayed with my final merged graph.

Proac T2 25mswindow+merged.jpg
 
Last edited:

eddy555

Member
Joined
May 29, 2021
Messages
17
Likes
10
is this correct that if we have to calculate attenuation factor from the area (driver / port) then we should do log(difference)*10 and not by 20 ?
 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
137
Likes
119
Location
Portugal
Last round of measures, to compare Ground Plane vs far field.

The best i could was to put the stand in a ladder, which puts the speaker at 170cm from floor.
Mic was pointing directly to the tweeter center, at 2m.
I did my best to be as accurate as possible, but the stand was not stable enough placed in the ladder, so i wasn't able to place it perfectly, i was alone and afraid the speaker would fall off.
You can see, for example, that the mic was not at a perfect 0º (between mic/floor), not a perfect 90º (between speaker/mic), and not a straight line between mic and center of the tweeter.

Bh3KMll.png


Next, i performed the Ground plane measure, and here i placed everything perfectly, having used a meter to ensure the 2m distance, and the speaker tilted exactly 8.5º, according to the formula here, using a laser meter. So the GP measure was as perfect as possible.

0dtEh8E.png


Its worth noting that there was a lot of ambient noise, don't know how much it could have impacted both measures.
For far field, i was only able to take 1 measure (due to the risk of speaker falling).
For ground plane, i took 4 measures to check they were all as similar as possible.

RWwBgV7.jpg


I'm a bit perplexed on how well both measures track above 1.5 kHz.

The glitch at 1.2 kHz seems to be a measuring error, unfortunately i was only able to take one measure, so i can't confirm.
But given how well both measures track, it seems that its probably an error.

Finally, both measures and a nearfield measure corrected with Baffle Diffraction:

9hXNGlR.jpg


It should be noted that this XTZ speaker has a Port, but i'm using the foam blockers to close the port.
Nevertheless, i assume it still behaves partly as a ported speaker, so i don't know how accurate the nearfield measure is.

If we discard the 1.2 kHz glitch as an error, the far field (at 1.7m), vs GP track almost perfectly from 1.1 kHz to 20kHz.
From 600 Hz to 1.1 kHz tracks well, with a slight increase on the far field measure.
The biggest discrepancy seems to be below 600Hz.
 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
137
Likes
119
Location
Portugal
@ctrl , @fluid Could you please have a look on the above, and provide your opinion on if, given the above, do GP measures have the required resolution for crossover development?

My own measures, as flawed as they may be, seem to indicate they do, and this seems in line with Erin's research.

Next week i want to start measuring my 1m/60kg speakers, and GP measurements is the only safest way i have to be able to perform full off-axis measurements.

rF3ufrP.jpg


The speaker uses the following drivers:
- Beyma 15MC700Nd
- Beyma TPL-150H
- AE TD12M

Crossover points will be around:
- 250Hz-300Hz - between the 15" and the 12"
- 1400-1600Hz - between the 12" and the AMT

Measures seems to track very well above 1.1 kHz, so that one, which is the most critical, should have very accurate measures.

But, for the crossover on 250Hz-300Hz, i'm a bit lost on what should i do:
1 - Should i just trust the GP measures?
2 - Should i do a nearfield measure, adjust for Baffle Diffraction?

I'm sorry for tagging both directly, but this build started on 2017, and now i just want to make a decision, complete the speakers and have them running, no matter what! But before starting, your informed opinion can prevent me from doing errors, or at least be able to get the most accurate measures i can do, with all the known caveats.

Thank you!
 
Last edited:

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,432
Likes
1,839
I'm a bit perplexed on how well both measures track above 1.5 kHz.

Why perplexed? I would expect them to agree very closely.

Crossover points will be around:
- 250Hz-300Hz - between the 15" and the 12"
- 1400-1600Hz - between the 12" and the AMT

Are you not concerned with beaming/waist-ing between the 12" and the AMT, given that xover range?
That seems like a recipe for major directivity mismatching.
 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
137
Likes
119
Location
Portugal
Why perplexed? I would expect them to agree very closely.
Most of the information on GP measures focus on low end, Erin research was probably the one i saw using it for the high end spectrum.

Are you not concerned with beaming/waist-ing between the 12" and the AMT, given that xover range?
That seems like a recipe for major directivity mismatching.

According to LineSource on diyaudio, "At the crossover frequency, the TPL-150H 80-degree horizontal polar pattern should match the "early beaming" 80-degree polar of the TD12M, which occurs around 1500Hz."

On diyaudio, there are a lot of information matching the TPL-150H to 12" drivers, specifically the AE TD12M and the Beyma 12p80nd, with very good results. This thread has a lot of information on that.

Javs on avsforums is using the TPL-150H with the Beyma 12p80nd with great results, crossing at around 1250Hz.

Mike Garrett on AVS is also using the TPL-150H with the AE TD12M (and AE TD18+), and usually mentions it as one of the best speakers he ever heard, independent of the price.

The TD12M should be able to go a bit higher than the Beyma 12p80nd, John from AE also recommended trying a crossover between 1400/1600.
I'll be taking full 0-180º measures, so final crossover points will be optimized using VituixCAD.
 
Last edited:

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,432
Likes
1,839
Most of the information on GP measures focus on low end, Erin research was probably the one i saw using it for the high end spectrum.

AFAIU, issues arise at higher frequencies using GP primarily due to offset between mic and the reflection plane.
Minimize that offset and one can minimize diffraction issues.

Javs on avsforums is using the TPL-150H with the Beyma 12p80nd with great results, crossing at around 1250Hz

In keeping with The Matrix theme, I guess you're Neo_2009 on AVS?
I flipped through a few pages of the above thread and did not come across off-axis measurements.
Still skeptical on the directivity choices. I hope your build turns out well. It should definitely get loud! :)
 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
137
Likes
119
Location
Portugal
Yep, neo_2009 on AVS :)
The above thread is from Javs, and you're correct, there are no off-axis measurements.

I should be able to perform the full measures next week, i'll create a new thread here with all the details, from how i will set up the manual turnover, to all the measures taken, to the VituixCAD project, up until the final result :)

I'm now just trying to understand what to do regarding the low end measures, i hope ctrl and fluid can chime in !!!
 

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,432
Likes
1,839
Yep, neo_2009 on AVS :)
The above thread is from Javs, and you're correct, there are no off-axis measurements.

I should be able to perform the full measures next week, i'll create a new thread here with all the details, from how i will set up the manual turnover, to all the measures taken, to the VituixCAD project, up until the final result :)

I'm now just trying to understand what to do regarding the low end measures, i hope ctrl and fluid can chime in !!!

What width of gate/window were you using to truncate the measured impulse response?
That will determine resolution and useful range of < 1kHz measurements.

i.e. 5ms window==200Hz resolution. I would not trust anything below ~0.7kHz in that case.

Could try to get a full range, non-gated measurement using a loud sweep from a close range (maximizing SNR).
I'm sure ctrl & fluid will have other, better ideas.
 

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,432
Likes
1,839
I've followed VituixCAD instructions for REW, and used a XLR mic with a loopback connection.
Given that the speaker was placed at 2m from the mic, i used 5,814s.
(in Vituixcad instructions it shows the offset for 1m):

NKUT7tp.png

That’s an offset for time-of-flight (TOF), I believe.
I was asking about the duration of the impulse retained for Frequency Response calculations. This windowing technique is used to exclude environmental effects (reflections) from a measurement. The trade off is a reduction in measurement resolving power.
 

morpheusX

Active Member
Joined
Nov 7, 2019
Messages
137
Likes
119
Location
Portugal
I was asking about the duration of the impulse retained for Frequency Response calculations. This windowing technique is used to exclude environmental effects (reflections) from a measurement. The trade off is a reduction in measurement resolving power.

I think this is what you are referring:

FOx4hzY.png
 

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,432
Likes
1,839
I think this is what you are referring:

FOx4hzY.png

Yes, that's it.
Looks like my guess was fortunate: you do have a measurement resolution of 200 Hz.
I don't recall if it's an option in REW, but plotting the FR with dots instead of lines would clearly show this idea.
 

fluid

Senior Member
Joined
Apr 19, 2021
Messages
424
Likes
709
There is a lot going on in the above posts. The gate should be set just before the first reflection by looking at the impulse response. Change the IR window settings so the first obvious reflection is just removed to make the window as large as possible. By calculator it should be 5.65ms if the first reflection distance to the floor is 1.7m. With the ground plane measurement the distance would seem to be to the wall behind the speaker or maybe to the side but by checking and zooming the IR it should be clear where it is.

I'm not sure why you think the 1.2K is an error, a dip right after the baffle diffraction peak is quite normal in sharp edged cabinets. Look at the diffraction sim that you made and see what the predicted shape of the baffle diffraction is. The ground plane could amplify this as it appears as if the baffle is twice as tall due to the acoustic mirror image.

The fact that the ground plane and off ground measurement match shows that you did a good job with the ground plane and the tiles being reflective and hard are a good surface for this method. Normally with a ground plane you have more space before reflections as it is most predominantly used for low frequencies. You might need to do some experimentation to see where the best place is to get the longest gate. It may be that measuring at 1m helps, the down side to that is that you are unlikely to be in the far field of the larger DIY speaker at 1m but it may be the better compromise to get cleaner data to a lower frequency.

If you want to avoid using the platform then experiment some more with the ground plane placement. You can look to try and undo the extra baffle added by the ground plane by comparing the difference in Vituix and seeing if it is significant enough to worry about accounting for.
 
Top Bottom