VituixCAD v2

[kimmosto]

Possible bug?

See my previous post. 'Crossover of driver' checkbox is for temporary testing only. Status of the checkbox is not saved to enclosure project or user config for the next session. Status of the checkbox is not read from user config or enclosure project. More permanent link would be very fragile because main program and enclosure tool are different applications with own projects which can change in a second.
Multiple driver databases and enclosure project lists, and separate files for project and drivers is already very fuzzy and fragile system. I don't like to make this any more complex. My preference would be to remove selectable driver databases and project lists, but all that was paid feature so let's "enjoy" them at least this year.

[Reet]

To expand on Kimmo's response, if all you want to do is view the nearfield cabinet response with a filter response applied, simply export the enclosure response and load it into the crossover tool as a driver response, then filter as you wish from here.

[kimmosto]

Exporting half space or full space on-axis response and impedance response from Enclosure tool to the main program has "always" been available, but that does not help people who want to check is power and excursion capacity of the radiator adequate for bass extension target, or what kind of high-pass is needed to get target SPL without exceeding Xmax. This is optimizing both max SPL and bass extension. My approach is to select driver+box -combination with adequate SPL capacity. By looking maximum SPL trace in Enclosure tool if it's not obvious. Target in crossover design is sound quality, and adjustments to XO are done when cabinet and measurements are available. No need to play with XO before actual system is measured. This is visible also in user manual as recommended design procedure.

One possible option is to replace 'Crossover of driver' -link with e.g. five active filter stages in Enclosure tool. Something like PEQ block in XSim3D or filter in WinISD. Passive XO would not be supported.

I did not like to do crossover calculations and UI in two places just for X and SPL capacity checking which is the reason why this fragile XO link was added (on request).

[Nil L]

kimmosto,
if I do a impulse export from the impulse view window, and set the left window to 145ms, and the loaded responses from REW were with a left window of 125ms, do I add non-existent data to the export? Which left window should be used to save data as much as possible?

[kimmosto]

Local active filter in Enclosure would look like this. Max six optional stages with DSP selection. Passive filter will not be supported. Crossover of driver -link totally removed from Enclosure so one pita less for me and users.

Any comments, Justin et al.?

[kimmosto]

kimmosto,
if I do a impulse export from the impulse view window, and set the left window to 145ms, and the loaded responses from REW were with a left window of 125ms, do I add non-existent data to the export? Which left window should be used to save data as much as possible?

I don't understand what you are trying to do with exported IR in REW, but maximum data is exported with rectangular window, maximum length from 0 to taps.

[kimmosto]

Local active filter in Enclosure would look like this.

This feature is already programmed and tested, but will not be published before some feedback. User manual is not updated. Copy-paste filter stages from Enclosure to XO schematic is possible in theory, but not very easy due to quite complex class structure of XO part.

[Reet]

Not a really big problem, but I've noticed that polar map contours won't be drawn if the SPL data is below -30dB. I added +100dB to the driver data to bring it up to a more normal level and the contours appeared.

[kimmosto]

polar map contours won't be drawn if the SPL data is below -30dB.

Oops. Limit was in absolute value instead of dB. Next rev. supports Ymax down to -200 dB, and value hovering between Ymin - 20 dB and Ymax + 20 dB. Ymin=magenta, Ymax=red.

[kimmosto]

2.0.100.0 (2023-03-10)

Enclosure

• Crossover of driver -link removed.
• Added Filter tab with six optional active filter stages: HP, LP, SLP, PEQ, PEQ and L-T. DSP system selection with list box. Transfer function magnitude is shown in Group delay chart with dB scale (right). Copy button copies active blocks to clipboard. Blocks can be pasted to crossover in the main program.

Main

• Contour lines of directivity polar map drawn down to -240 dB.
• Value hovering range of directivity polar map extended to Ymin-20dB...Ymax+20dB.

[Justin Zazzi]

Thank you Kimmo!

[Reet]

Check for updates when opening VituixCAD doesn't appear to be functioning at the moment, at least on my end. Manually downloaded the latest. After 100 builds of 2.0, maybe it's time for a 2.1 milestone?

[kimmosto]

Check for updates when opening VituixCAD doesn't appear to be functioning at the moment, at least on my end. Manually downloaded the latest. After 100 builds of 2.0, maybe it's time for a 2.1 milestone?

Problem is more difficult. Something has probably changed with service provider.

2.0.100.1 (2023-03-12)

• Support for Tls 1.1 and Tls 1.2 security protocols enabled to bring auto update and online database in Enclosure tool back to life. Manual download and installation of 2.0.100.1 (2023-03-12) is probably required because my service provider has ended support for Ssl3/Tls1.

---

[Reet]

For the impedance "show raw impedances", currently it shows impedance for all drivers in the project, whether they are on the crossover or not, hidden or not. I would really like to see this function show raw impedance for only the drivers used in the crossover, and hide impedance for drivers with "hide traces" selected. Thanks!

[kimmosto]

I would really like to see this function show raw impedance for only the drivers used in the crossover, and hide impedance for drivers with "hide traces" selected.

2.0.100.2 (2023-03-17)

Main

• Impedance chart shows raw impedance for only the drivers used in the selected crossover variant.
• Auto update reads the latest version number asynchronously without blocking the program if internet/server is not available.
• Auto update downloads the latest VituixCAD2_setup.exe asynchronously without blocking the program if internet/server is not available.
• Initial project directory changed to Main directory in Options + \Projects.
• Download directory changed to Main directory + \Download.

[Nil L]

Kimmo, I measured the tweeter in REW, exported to txt file, uploaded to VituixCad. It seemed to me that the FR amplitude should be slightly different. I exported this response from VituixCad to a txt file, loaded it and the original file into REW and compared. They are different. On one image, strongly smoothed responses, on the other up to 1/48. What should I do to load the response into VituixCad as it is in REW? I also compared bass and mid responses. The mid is exported 1:1, the bass after 10 kHz differs by 10-15 dB. This difference in bass response is irrelevant. But the tweeter has to be accurate.

Update.
I myself found out that this situation only appears if you export from REW without smoothing. 1/48 smoothing fixes this problem. It's almost 1:1, but the levels are ok.

[Reet]

Something from my email today, though you guys might get a kick out of it. Might be the worst VituixCAD tutorial yet

https://www.minidsp.com/applications/rew/isobaric-polar-plots

[JonMarsh]

If anything, your comment is a diplomatic understatement.

[Quza]

Do you have any hints on creating the crossover for a speaker like the kii3?
Doing everything with linear phase crossovers and afterwards replace them with normal phase crossovers and all pass filters doesn't work as good as I thought it would. A lot more fiddling with the parameters is necessary.
Is there a way to abuse the optimizer of VituixCAD for this purpose?
I tried turning the speaker by the angle where the strongest cancellation should be and had the optimizer seek a low target level, but that obviously started reducing the overall level of the speaker pretty quickly.

I'm trying to create different setups for the speaker:
- constant directivity >100Hz, from 0-90° for indoor use in front of heavy theatre curtains
- maximum cancellation at 90°, constant directivity from 0-40°, for outdoor use with other activities next to the stage
- maximum cancellation in a large area behind the speaker and constant directivity from 0-60°

[Nil L]

I have a repeat question that I asked in post 826 I loaded the characteristics obtained from REW with 1/48 smoothing into VituixCad. In VituixCad they are displayed correctly. But after exporting impulses from the impulse view window, the tweeter, for example, is 22.5 dB higher than it was, middle is 17.5 dB higher. What could I misconfigure? After post 826 I checked, and then everything happened correctly. The levels stayed the same.

[kimmosto]

Do you have any hints on creating the crossover for a speaker like the kii3?

https://www.htguide.com/forum/mission-possible-diy/development-tools-techniques/42827-vituixcad-v2?p=835082#post835082

[Reet]

I have a repeat question that I asked in post 826 I loaded the characteristics obtained from REW with 1/48 smoothing into VituixCad. In VituixCad they are displayed correctly. But after exporting impulses from the impulse view window, the tweeter, for example, is 22.5 dB higher than it was, middle is 17.5 dB higher. What could I misconfigure? After post 826 I checked, and then everything happened correctly. The levels stayed the same.

Make sure "normalize impulse" is unchecked when exporting impulse from REW.

[Nil L]

I don't export impulses from REW, but frequency responses. And impedances. After exporting from REW to a txt file, the frequency response in VituixCad is the same as the one I observe in REW. By level.

[Reet]

Sorry, I misunderstood your post. For impulse export from VituixCAD, file format other than PCM wav must be used to avoid normalization.

[Nil L]

32-bit IEEE mono (.wav)(*.wav) is used. In the options, the 'Normalize SPL' checkbox, turned on and off, does not change the impulse export.

[tktran]

Hi Kimmo,

In the manual for taking measurements for REW with VituixCAD2, you stipulate measuring at 1m in the far field. and adjusting time delay to

"Timing offset: 2.907 ms with 1000 mm measurement distance from mic to rotation center, with analog electronics without processing delays. Delay [ms] = measurement distance [mm] / 344 + possible processing delay [ms] of DSP"

Are there any problems of measuring at 50cm, or 2 ft etc, and adjusting the timing offset?

As long the measurement condition for being in the far field are met (eg. >2 times cabinet width && >3 times driver diameter)
Of course for really big speakers measuring out at 1m or more is better, but for tiny 2 ways...

BR,
Thanh

[JonMarsh]

Check for updates when opening VituixCAD doesn't appear to be functioning at the moment, at least on my end. Manually downloaded the latest. After 100 builds of 2.0, maybe it's time for a 2.1 milestone?

On my Mac Pro running WIN10, always kept updated, the newer security settings prevent downloading and running VituixCAD updates. Had to manually update this morning. I was wondering about this, too, because I hadn't seen any updates in a month or so, but I hadn't checked the Changelog status, either.

And MS makes it hard to run the update after download- says it's protecting you from an unrecognized App.

[theSven]

On my Mac Pro running WIN10, always kept updated, the newer security settings prevent downloading and running VituixCAD updates. Had to manually update this morning. I was wondering about this, too, because I hadn't seen any updates in a month or so, but I hadn't checked the Changelog status, either.

And MS makes it hard to run the update after download- says it's protecting you from an unrecognized App.

Windows wants all applications to be code signed these days. That would required getting a publicly trusted code signing certificate that can be used to sign the program.

[Reet]

Just tell windows to take the training wheels off and install your software anyway.

FWIW Mac has a similar problem with software, they call it "notarization", so you also have to specifically allow installation of software that hasn't been notarized.

[JonMarsh]

That's what I always do when that happens, anyway, but MS's "Nanny" efforts have become more annoying, and they don't even have a direct dialog choice for that- just a "Get Info" button. Well, I'm used to dealing with this BS, but it doesn't mean I have to like or admire it.

[Reet]

Are there any problems of measuring at 50cm, or 2 ft etc, and adjusting the timing offset?

Measurement distance is really up to you, there must be a balance as farther is better to capture real diffraction at listening distance, but closer is better as far as window length goes. 1m is generally a good distance for most speakers indoors, until you are building big cabinets where 3x cabinet width is >1m. Closer than 1m distance may allow for slight increase in window length, but quite minimal difference to actual measured data resolution.

As long the measurement condition for being in the far field are met (eg. >2 times cabinet width && >3 times driver diameter)
Of course for really big speakers measuring out at 1m or more is better, but for tiny 2 ways...

I would suggest, minimum 3x cabinet width. Size of driver is irrelevant.

[tktran]

Measurement distance is really up to you, there must be a balance as farther is better to capture real diffraction at listening distance, but closer is better as far as window length goes. 1m is generally a good distance for most speakers indoors, until you are building big cabinets where 3x cabinet width is >1m. Closer than 1m distance may allow for slight increase in window length, but quite minimal difference to actual measured data resolution.

Thanks for this. I spent the last few Sunday night at the local basketball court from 9PM to MIDNIGHT! taking measurements at great heights.
Which is great as far as low noise floor (30dB) and window lengths (>10ms) are concerned.


Except that setting and packing up takes SO LONG:


I'd rather do it from the comfort of my own spare room, but it's only practicable at about 50cm for a nice window length.

I would suggest, minimum 3x cabinet width. Size of driver is irrelevant.

Do you know the math behind behind this?

Up until recently I had gone with Jeff Bagby's suggestions from White Paper - Accurate In-Room Frequency Response to 10Hz.pdf (claub.net)
where I understood far field to be '>2 times cabinet width && >3 times driver cone diameter'

Only since when VituixCAD's Merger tool started merging with both the diffraction simulated far field (3-20m) using Diffraction tool, and measured distance at 1000mm have I changed my process...

[Reet]

Do you know the math behind behind this?

up until recently I had gone with Jeff Bagby's suggestions

With a big basketball court at your disposal, you don't need math. Just measure yourself at many distances and compare, find the point where an increase in mic distance no longer affects the measured response. Realistically, the furthest you can place the mic to actual listening distance the better, 2m distance is more ideal but not practical in many rooms, big basketball court might be able to work though, of course with greater effort to hike the speaker up high and be able to rotate it safely and easily.

In any case, you'll find the 3x baffle width mentioned in VituixCAD measurement guide(s). Jeff's comment about 3-5x the driver width is about being "effectively in the far field", without consideration for the baffle size. That paper also is describing a single channel method with USB mic, so perhaps some generalization to prevent measuring the "on-axis" woofer response too far off-axis with mic located at tweeter axis, not driver axis.

[Reet]

Today's adventure. Compared measurements of a bookshelf speaker at various distances. Tweeter is disconnected so only woofer is measured. Speaker is 17.5cm wide, so 2x baffle width is 35m, 3x would be 52.5cm. Here I measured in some small arbitrary increment of 11cm, all the way out to 99cm.
This is just the far field response. Normalized amplitude at the bump at 1300Hz so they are all represented at about the same 4pi amplitude.



I then merged the whole set with nearfield + diffraction response at 300Hz.


For the responses here, overall bass balance was not greatly affected until at 22cm distance, however there is difference in the response >500Hz all the way up to 1kHz, likely early reflections from my very non-ideal office space used for this test.

[Reet]

I should mention as well, that each response at each distance had the window length set to about the same location on the first reflection. For each measurement, the following right window length was used:

3cm - 1000ms
11cm - 5ms
22cm - 5ms
33cm - 4.8ms
44cm - 4.6ms
55cm - 4.35ms
66cm - 4.1ms
77cm - 3.87ms
88cm - 3.7ms
99cm - 3.5ms

[Reet]

Looking back on the above measurements, the closer mic distance has greater influence of near reflections from whatever table the speaker is sitting on, and the increased window length really provided very little difference to overall measurement resolution. 1m distance reduces impact from early reflections, so I’ll stick with that.

[Nil L]

For the responses here, overall bass balance was not greatly affected until at 22cm distance.

All the same, 33 cm. Since 5 out of 8 measurements fit one into one. And 33 cm is 2 * speaker diameter.
And these are measurements of a speaker that plays up to 100 hertz. And what will measurements look like with a speaker that plays at least up to 50 hertz? I think 2*speaker diameter won't work. 1.5 m, and preferably 2 m will be required.

[tktran]

Very interesting!!

I will try a similar experiment this weekend, outdoors, at 2m height.

[Nil L]

Very interesting!!

I will try a similar experiment this weekend, outdoors, at 2m height.

Wow, I was thinking of asking you to do this. So phones are not needed.

[Reet]

And what will measurements look like with a speaker that plays at least up to 50 hertz? I think 2*speaker diameter won't work. 1.5 m, and preferably 2 m will be required.

use nearfield for low freq measurements…For far field measurements, baffle width is limiting factor on mic distance, not speaker diameter.

[tktran]

LIVE!

stay tuned…

[tktran]

Far field measurements first



Edit: Scale and colours changed for clarity

[Nil L]

Is this a tweeter? Not indicative. Do you have a driver in the enclosure that plays below 50Hz? If you could measure such a driver from a distance of 1m, 1.5m, 2m, it would be interesting.

[tktran]

This is a midwoofer with response down to below 50Hz, yes.

However, to get a reflection free (anechoic) response down to 50Hz; we need a gate time of 20ms
The math is T = 1/f where f is the frequency in Hz and T is the period in seconds.

So if you want to measure to 50Hz, you need a window of 1/50 = 0.02 secs = 20ms.

There's a tool included in VituixCAD2:

Auxiliary -> Time window that you can use to calculate this-

eg. to measure down to 50Hz, at a distance of 1.5m; that means the speaker and microphone have to be off the ground by more than 4.15m:







To measure at 2m, the speaker has to be off the ground at least 4.4m



If you're NOT using VituixCAD2 you can try it for yourself online:

https://mehlau.net/audio/floorbounce/

I do have the ability to measure at 5m above the ground, but I'm too lazy- to set up and pack up is even more bothersome than the basketball court!

I prefer to do it indoors in my lab and merge the farfield with the nearfield, as per Kimmo's instructions:
page 14 of:

VituixCAD_Measurement_REW.pdf (kimmosaunisto.net)

[Reet]

Some more information on "how far is far enough" from Audio Precision:

https://drive.google.com/file/d/1ICLgcjqGtyvNaSva4jPQ8Z3vYwFN9_Ex/view?usp=share_link

So how do time selective techniques work in practice? Consider the measurement setup depicted in Figure 4. A measurement microphone is positioned on-axis in front of a loudspeaker in a semi-reverberant room. To be in the far field of the loudspeaker, the distance d is chosen such that it is more than 3 times M, the largest significant dimension of the loudspeaker. The path length of the direct sound from the loudspeaker to the microphone is d, and the path length of the first reflection from the nearest reflecting surface (the floor in this case) is 2d_R. Based on this geometry, the time difference, T, between the direct sound arrival at the microphone and the first reflection is T = (2d_R - d) / c, where c is the speed of sound.

[kimmosto]

2.0.101.0 (2023-04-01)

• Added EASERA .etx file format option to impedance response exports. f, Ohm Re and Ohm Im columns. Constant frequency step directly for complex FFT. Sample rate 44100 Hz. 8193 data lines (16k FFT).

Convert IR to FR

• Added WAV 32-bit IEEE mono impulse response file format option to Export command. Smoothing not yet supported.
  dBSPL in Pascal so scaling of +94dB is needed if reader does not assume that values are in Pascal.

[kimmosto]

Are there any problems of measuring at 50cm, or 2 ft etc, and adjusting the timing offset?

This is very case dependent. I had to make instructions shorter than few hundred pages which would be required to speculate (all) different speaker concepts and measurement conditions. Line arrays, long (or wide) planars, horns, deep waveguides etc. could require different approach. Simplified theoretical limit for far field is 6a i.e. 3 x diameter of "radiating surface". 1000 mm is nice rounded value which enables 4.5 ms time window in almost any living room, and supports cones up to ca. 15" or baffle width or height or longest dimension up to 33 cm - depending on how radius of radiating surface is interpreted. But 1000 mm is not carved to stone. I have measured 18" at 1 m. Some tractrix horns at 1.2-1.5m due to very different wavefront and possible frequency dependent far field point and acoustic center quite far from rotation center which is different compared to other ways with direct radiators.
Using near field measurement, diffraction simulation and merging allows violating simple far field rules. For example Audio Precision's AN assumes that radiating surface includes all diffractions (~full baffle loss) in the measurement data. That is not strictly required in VCAD measurement instructions because it's impossible to achieve with large constructions by measuring indoors at home.
Life is full of compromises, and user makes final decisions.

[kimmosto]

Unapproved message waits for moderation. Why that feature is still active?