VituixCAD v2

[Reet]

I have trouble understanding why the diffraction sim for merging would not be at typical listening distance of 2-3m? There is very small difference of course from 2500mm distance to 10m simulation, I am just trying to understand the reasoning behind using a longer distance than typical in-room listening.

Unrelated, I have noticed on two separate occasions a certain someone has suggested that the weighting of CTA-2034 is inappropriate for an omnidirectional speaker in which a mid bass driver is vertically oriented pointing at the ceiling, like Linkwitz Pluto. I disagree with this, my understanding of CTA-2034 weights appropriately for relation of speaker to listener in a typical room regardless of speaker type, driver orientation, etc. I’m interested in your thoughts/opinions on this.

[Reet]

I have trouble understanding why the diffraction sim for merging would not be at typical listening distance of 2-3m? There is very small difference of course from 2500mm distance to 10m simulation, I am just trying to understand the reasoning behind using a longer distance than typical in-room listening.

To answer my own question, am I right to assume that since measurement distance at 1m is too close for capturing diffraction at 2-3m, so the diffraction sim is then done too far at 10-30m to compensate, so end result averages out to closer to typical 2-3m result.

[kimmosto]

...why the diffraction sim for merging would not be at typical listening distance of 2-3m?

VCAD sums direct sound and diffracted rays as pressure values at the mic using pressure proportional to travel distance. Some other programs calculate with axis distance squared. So result calculated with VCAD is very sensitive to axis distance setting compared to other programs giving close to same result = full baffle loss between ca. 1 meter and infinity. Using the longest possible axis distance in VCAD is just a trick to get close to full 6 dB baffle loss which helps to avoid too thin sound balance.
Some day I might change axis distance calculation. After that measurement distance compensation added to the latest revision could be quite useless.

[tktran]

Any chance that there is an option to change the trace colour in SPL Trace? The red seems to be the exact same red colour as the on-axis for a certain manufacturer.

[kimmosto]

Maybe trace's hue -180 deg and 50% transparent. For black&white traces some bright color with transparency of 50%

[kimmosto]

...or 5px wide lime and cyan highlighter pen with 50% opacity

[kimmosto]

2.0.97.0 (2022-12-16)

Main

• 'SPL Horizonal.txt' and 'SPL Vertical.txt' multi-column files measured with Klippel NFS are splitted automatically to multiple frd files when .txt is renamed to .nfs and loaded to Drivers tab.
• Average on-axis SPL 100-12000 Hz added to Preference rating window.
• Added 'Trifonov Q Optimized Transient mk2' target responses to Optimizer.
• Added Active 2-way Trifonov TQ, OPA 2-way Trifonov TQ and Passive 2-way Trifonov TQ library blocks.

SPL Trace

• Added 'Green SPL, Cyan Z' command to Options menu. SPL trace is highlighted green and Z trace cyan instead of solid red and blue.

[Reet]

Interesting addition. Generally I don't try to hit any specific high pass or low pass target curve shape, but rather aim for balance of overall response with least complexity. Comparing a couple of my previous designs, I found it interesting that the low pass sections had very good agreement with the Trivonov TQ curve, however high pass section is more inconsistent from speaker to speaker.

If I could suggest a small change to VituixCAD, I was wondering if the file open dialog could include "All supported filetypes" by default, instead of having to manually select .txt or .frd etc. in a dropdown menu. It would simplify file open process to have txt, frd, lms, nfs files all shown without having to manually select from the dropdown.

[kimmosto]

if the file open dialog could include "All supported filetypes" by default.

That is very good idea. I've seen applications using combined extension filter, but never done it to my programs. So now it's there. I will fix if this causes problems with saving frequency or impedance responses. At the moment Save file dialog uses the same filter string with Open file dialog.

2.0.97.1 (2022-12-17)

• Frequency response file extensions (*.txt;*.frd;*.cal;*.mic;*.lms) listed together in Open File Dialog to avoid filter selection.
• Impedance response file extensions (*.txt;*.zma) listed together in Open File Dialog to avoid filter selection.

[draki]

2.0.97.0 (2022-12-16)

Main[*]Average on-axis SPL 100-12000 Hz added to Preference rating window.

Care to elaborate this one? You lost me here..........I mean I can't find it (the PR window in the Main)? It must be me therefore the question.

[Reet]

2.0.97.1 (2022-12-17)

• Frequency response file extensions (*.txt;*.frd;*.cal;*.mic;*.lms) listed together in Open File Dialog to avoid filter selection.
• Impedance response file extensions (*.txt;*.zma) listed together in Open File Dialog to avoid filter selection.

Thank you !

[kimmosto]

Care to elaborate this one? You lost me here..........I mean I can't find it (the PR window in the Main)?

Sorry I was lazy and grouped Optimizer and Preference rating under Main (because they co-operate with main window only). This is more traditional presentation in changelog and hopefully helps to find the value.

2.0.97.0 (2022-12-16)

Main

• 'SPL Horizonal.txt' and 'SPL Vertical.txt' multi-column files measured with Klippel NFS are splitted automatically to multiple frd files when .txt is renamed to .nfs and loaded to Drivers tab.
• Added Active 2-way Trifonov TQ, OPA 2-way Trifonov TQ and Passive 2-way Trifonov TQ library blocks.

Optimizer

• Added 'Trifonov Q Optimized Transient mk2' target responses.

Preference rating

• Added average on-axis SPL 100-12000 Hz as SPL_ON to bottom left corner.

SPL Trace

• Added 'Green SPL, Cyan Z' command to Options menu. SPL trace is highlighted green and Z trace cyan instead of solid red and blue.

[kimmosto]

Interesting addition. Generally I don't try to hit any specific high pass or low pass target curve shape, but rather aim for balance of overall response with least complexity. Comparing a couple of my previous designs, I found it interesting that the low pass sections had very good agreement with the Trivonov TQ curve, however high pass section is more inconsistent from speaker to speaker.

Few words about Trifonov Transient filters. Generally my attitude is positive for improvements of timing. My solution has been FIR with linear-phase slopes for few years, but many years ago I tested Le'Cleach XO with line arrays (KS-160-21). That approach has symmetrical 3rd order Butterworth slopes, delay and inverted HF. Result was okay though significance was low due to high XO and 2-way only.

Theoretical problem with passive application with separate drivers is directivity effects in horizontal plane due to different Z position. Horizontal plane with Trifonov TQ looks like this with XO=2293 Hz, c-c=150 mm and constant omni directivity of both ways:


Obviously this has some effect to early reflections and power responses:


In practice woofer's directivity usually increases close to XO and tweeter's own directivity could be zero right above XO so it's very close that phase mismatch of this kind of XO helps to balance directivity so that on-axis could be kept quite flat.

Active DSP version of Trifonov TQ with digital delay for "Z movement" is smoother in horizontal plane because mechanical Z movement is 0 mm:


Result depends on XO frequency, Z difference, Y difference and directivities so this example with c-c = 1 wavelength at XO is just one case.
Time domain


===========

Trifonov TP has lower Q factors so phase matching is also different.

Power dip is a bit wider and deeper in theory with the same parameters (XO=2293, c-c=150mm)

Time domain is a bit better than with _TQ due to shorter delay.

[kimmosto]

Le'Cleach XO with line arrays (KS-160-21)

Memories from 2009

[Reet]

Few words about Trifonov Transient filters...

Thank you for this additional insight.

I find many of your posts such as this provide valuable insight, however over time get buried in past pages of single thread discussion. I hope you don't mind, that I have copied some of past information provided into a "random tidbits" document that I have compiled for easy reference of this useful information. This document can be found here:

https://drive.google.com/file/d/12xUt4vHfakLBP1x-_eZDuC2-PrYwyehP/view?usp=share_link

Complete link to my documentation relating to speaker design with strong focus on VituixCAD specifically, can be found here:

https://drive.google.com/drive/folders/1KDyECkAIVAuNEtKVobKmsfvtFczGqOQw?usp=share_link

In that shared folder, documents I have compiled are under "My Documents", and other supporting documentation that I find particularly useful is in the "Other Documents" folder.

Most recent addition was the "Single vs Dual Channel measurement" document, which attempts to highlight the clear benefit of a 2-channel measurement process for loudspeaker design. Next plan is for a document about minimum phase and its (rare) use cases. This one is proving to be quite complex as there is quite poor understanding of minimum phase and phase in general in DIY community and will require me to set up some measurements to illustrate / prove some points with real measured results. It also reviews the various software available for response merging, which can be damaging to phase when done incorrectly. Spoiler - VituixCAD is by far and away the most functional and featureful tool for response merging

[kimmosto]

2.0.98.0 (2022-12-22)

Graphs

• Chart legends sorted to alphabetical order on Copy and Export.

Optimizer

• Added 'On-axis linearity' checkbox and text box. Limits on-axis deviations into dB value within optimized range of ON/LW.

Preference rating

• Added On-axis linearity in dB. Difference between maximum and minimum SPL within 100-12000 Hz.

[kimmosto]

Example how 'On-axis linearity' constraint can be used. Target is to make as linear predicted in-room response (PIR) as possible so we optimize Preference rating with Custom equation and set all weight to SM_PIR and others to 0. That alone could lead to high deviations in on-axis and listening window responses if directivity slope is not linear. Or some unwanted tilt. Now we can set On-axis linearity constraint to keep deviations acceptable and slope close to 0 dB/oct. For example on-axis should stay within 2.5 dB 100-12000 Hz if possible:



This feature is an alternative to target line in SPL chart. Advantage is that user does not have to play with target line(s) or slope target(s) in Preference rating and weighting between ON/LW and PIR/SP to get on-axis response within desirable tolerance window. One optimized response such as PIR or SP, and on-axis constraint is simple to set up.

[Reet]

I have a minor bug to report. When opening a MLSSA text file in the impulse response tool, you are prompted to select the sample rate, with options of "OK" or "Cancel". Selection here should only be "OK, as cancel button will cause VituixCAD to lock up.

[Reet]

In the "Documents for Newbies" category, I've put together a document highlighting the usefullness (or uselessness) of minimum phase. This document in combination with my other on single vs dual channel measurements should highlight the reason why dual channel measurements and using measured data including phase should be done where ever possible when designing with VituixCAD.
https://drive.google.com/file/d/1dq7...usp=share_link

Complete set of documents is found here:
https://drive.google.com/drive/folders/1KDyECkAIVAuNEtKVobKmsfvtFczGqOQw?usp=share_link

[kimmosto]

I have a minor bug to report. When opening a MLSSA text file in the impulse response tool, you are prompted to select the sample rate, with options of "OK" or "Cancel". Selection here should only be "OK, as cancel button will cause VituixCAD to lock up.

Ok, thanks. Problem was that all samples in impulse response buffer were 0 when Convert IR to FR window opens. Calculation or graphs cannot handle absolute silence. Lock up is now avoided by initializing first sample with 1.0 Pa. Not totally avoided because legal IR file can be totally silent. I will try to handle that later.

[kimmosto]

2.0.98.1 (2022-12-25)

Convert IR to FR

• Cancel button removed from 'Illegal MLSSA format' window asking Sample rate.
• Y scale minimum of IR chart limited to 1E-6 to avoid crash with silent IR.
• Status of Show Impulse response added to user config.

[Nil L]

For three drivers, I have loaded FRs that have a relative time between them. Then I made a crossover. Can I see how the relative timing between drivers has changed with the crossover?

[Upsetter]

Greetings.

I was wondering if I could kindly ask some advice, since I'm kind of chasing my tail on a topic here.

If I am simulating a driver in dipole but with no baffle, will there be some other factor to consider in regards to delaying rear driver?
In my mind the total driver dia. will be "baffle size", but if I'm using this number divided by half plus thickness of "baffle" (driver basket flange), I'm getting severe dips in the response. To avoid this I have to reduce delay distance to a very low number. Simulated response should resemble on axis response when no filters is applied, no?

[Reet]

For three drivers, I have loaded FRs that have a relative time between them. Then I made a crossover. Can I see how the relative timing between drivers has changed with the crossover?

Phase plot is in the top right, there's also the impulse response under the view menu.

Merry Xmas

[Reet]

Merry Christmas kimmosto!

Xmas feature request

I'd love to be able to manually adjust Y axis scale in the impulse response view. There is a "freeze Y" button to prevent auto-scaling, but I think it would be better for some manual adjustment ability here to be able to zoom in/out.

[Nil L]

Phase plot is in the top right, there's also the impulse response under the view menu.

Merry Xmas

I tried this menu item. but did not understand how I can see, for example, the impulses of drivers with their crossover branches on one graph. I loaded data at 48 kHz, but the program always opens the pulse view window from 96 kHz. What does the time on the bottom left mean, why doesn't it change if I choose a different driver to view. The scale of the graph does not change. The instructions don't say anything about it.

[Reet]

I tried this menu item. but did not understand how I can see, for example, the impulses of drivers with their crossover branches on one graph. I loaded data at 48 kHz, but the program always opens the pulse view window from 96 kHz. What does the time on the bottom left mean, why doesn't it change if I choose a different driver to view. The scale of the graph does not change. The instructions don't say anything about it.

It will show you the same information as if you were to measure your drivers in REW or ARTA for example, and view impulse, step, ETC, etc.

sample rate relates to file export, and of course FFT length is in samples.

Signal to show can be a bit confusing. "Driver in D1" for example will show impulse of filter response (filter transfer function), while "Driver SPL D1" will include the driver SPL data.

Step response can be quite informative of speaker "timing", for example here is a 3-way speaker of mine. I cycled through the driver SPL and saved as overlay, finally the solid red is the total system step response, so you can see how the audio from each driver is delayed due to both physical location and crossover implementation.

[Nil L]

Reet,
thank you. This is what I wanted. I will try to do the same as you. Maybe you have an account on Youtube, and you will record a video of how you do it? Or you can attach a video file here.
I can export impulses from this window and then open them in REW. I can save as overlay. But how then to look overlay?

[Nil L]

Found a problem with impuls export. I set the sample rate to 48k. I choose to export the driver 'with SPL'. Two drivers open normally in REW, one at -30 dB.

[Reet]

You have to use IEEE formatted wav, not PCM to capture the amplitude. Not sure what you're trying to do with REW though.

[Reet]

In the "Documents for Newbies" category, I've put together a document highlighting the usefullness (or uselessness) of minimum phase. This document in combination with my other on single vs dual channel measurements should highlight the reason why dual channel measurements and using measured data including phase should be done where ever possible when designing with VituixCAD.
https://drive.google.com/file/d/1dq7...usp=share_link

Complete set of documents is found here:
https://drive.google.com/drive/folders/1KDyECkAIVAuNEtKVobKmsfvtFczGqOQw?usp=share_link

Hi All,
Most of documents I've written available above in Google Drive, are now posted to the Articles section of HT Guide. Easily found in the DIY category, or select the "vituixcad" tag.

[Nil L]

You have to use IEEE formatted wav, not PCM to capture the amplitude.

I used 32 bit IEEE mono (.wav). For three drivers. Woofer and tweeter are normally exported and then imported into REW. And the midrange amplitude is 30 dB lower. I tried many times, restarted VituixCad, but the midrange graph does not look right.

Not sure what you're trying to do with REW though.

I don't know how to view overlay in VituixCad. Graph scaling is not clear to me. In REW, everything is easy and clear for me, and I can also compare the results of real measurements of drivers with those predicted by VituixCad.

[Reet]

I used 32 bit IEEE mono (.wav). For three drivers. Woofer and tweeter are normally exported and then imported into REW. And the midrange amplitude is 30 dB lower. I tried many times, restarted VituixCad, but the midrange graph does not look right.

Unfortunately you may have to show your work a little more to understand what the problem is. I went through my usual steps for impulse response export and import to REW looks exactly as what's in VituixCAD.

I don't know how to view overlay in VituixCad.

Just right click on any graph and "save selected as overlay". It's one of the most useful features in VituixCAD because you can do this anywhere in the program for providing comparative information easily.

[Nil L]

Looks like I figured out my questions. Here's what I found out. I opened a new project, loaded the data, and then the export of all drivers became normal. Something went wrong in the project that I wrote about. Exactly what, I couldn't figure out. Next, the 'IR center' checkbox was checked all the time, I unchecked it, and set the time there to 125ms, assuming it's the left window like in REW.
Further, I assume that in the impulse view window, VituixCad arranges the impulses of the drivers not relative to each other in time, but by setting the peaks at the same time, which is indicated in the checkbox. The same happens when importing an impulse. In REW, all imported pulses appeared to be peak-centered at t=0. Do you have it like that?

[Reet]

Looks like I figured out my questions. Here's what I found out. I opened a new project, loaded the data, and then the export of all drivers became normal. Something went wrong in the project that I wrote about. Exactly what, I couldn't figure out. Next, the 'IR center' checkbox was checked all the time, I unchecked it, and set the time there to 125ms, assuming it's the left window like in REW.
Further, I assume that in the impulse view window, VituixCad arranges the impulses of the drivers not relative to each other in time, but by setting the peaks at the same time, which is indicated in the checkbox. The same happens when importing an impulse. In REW, all imported pulses appeared to be peak-centered at t=0. Do you have it like that?

Like I said, exported impulse will be exactly as shown in VituixCAD upon export, with window length based on number of taps selected. To simply "export everything" choose high number of taps.

Centre IR is really only needed for specific FIR filtering to ensure long pre-delay. You only need enough pre-delay to capture any pre-ringing of the impulse, generally 1-2ms is more than enough. You can see that I used 1ms in my previous image.

VituixCAD does not set the peaks at the same time, that would be counter-intuitive. Adjust z offset of one of your drivers while looking at the impulse response window "Driver SPL for D#" to see the time delay change in real time. Also, previous step response image shown from myself shows 3 drivers with peak of impulse at different locations in time.

[Nil L]

On your screenshot - Step Response. Not Impulse. Step Response is clearly visible. Impulse of some drivers (tweeter) is clearly visible, others (bass) are almost invisible. I have verified that when changing the Z parameter in the driver data, the Impuls and Step Response graphics in the Impulse Viewer shift in real time. But when exporting, the time t=0 is set to the peak of each graph. Therefore, exporting to REW does not make sense for the purposes of a more detailed analysis of the driver impulse graphs + crossover. And in the VituixCad impulse view window, the tweeter impulse is clearly visible, worse than the middle, and the bass is barely visible along the vertical axis. The vertical axis is not scaled. Therefore, I have no idea how to analyze the timing relationships of drivers in VituixCad as well.

[Reet]

Try updating to latest REW early access release:
https://www.avnirvana.com/threads/v5...s-build.11154/

step response is just the integral of the impulse, perfectly useful to show same time relations.

[Nil L]

Try updating to latest REW early access release

Your advice gave me an idea. In the REW analysis settings there is a parameter 'for imports set t=0 at impuls peak/first sample'. I chose the first option. I changed to the second and was able to see the timing relationships between the driver graphs.
So things should work out. Thanks a lot.

[Reet]

Same information as previous screenshot, shown in REW.

Impulse response:


Step response:

[Reet]

Your advice gave me an idea. In the REW analysis settings there is a parameter 'for imports set t=0 at impuls peak/first sample'. I chose the first option. I changed to the second and was able to see the timing relationships between the driver graphs.
So things should work out. Thanks a lot.

Glad you found the setting. I had forgot about that, set to "first sample" of course

[Nil L]

Another strange situation when importing the impulse of the midrange driver. Bass impulse import and tweeter impulse import work well at FFT length 65536 and 32768 taps. To obtain maximum data in a midrange driver impulse, FFT length 32768 and 16384 taps are required.

[mv8]

I have a question for Kimmosto. I have calibrated ARTA for SPL for 2.83V, but when I import the PIR files into VituixCAD, I have to scale the SPL up by about 9 dB. Is there any function I should change in the program options so that this correction is not necessary?

[kimmosto]

I have a question for Kimmosto. I have calibrated ARTA for SPL for 2.83V, but when I import the PIR files into VituixCAD, I have to scale the SPL up by about 9 dB. Is there any function I should change in the program options so that this correction is not necessary?

This was discussed in October on diyaudio. Few quotes from there:

Preamp. gain is not included in PIR file so it must be 1. Gain of soundcard input should be included in mic sensitivity in order to deliver correct sensitivity value to PIR file. This happens automatically when sensitivity is measured with microphone calibrator and 'Estimate Mic. Sensitivity' button, and accepted with Accept button.

Main purpose of frequency response measurement is not to show dB/2.83V/1m of the driver or speaker. It should show sound pressure level at the mic - no matter voltage in terminals and measurement distance. And this value should be delivered to frequency response graph or post processing for example with VCAD.

[kimmosto]

Xmas feature request

I'd love to be able to manually adjust Y axis scale in the impulse response view. There is a "freeze Y" button to prevent auto-scaling, but I think it would be better for some manual adjustment ability here to be able to zoom in/out.

Happy New Year! Here is "soft" stocking filler for you. Zero line stays in the middle and dB scale stays constant. It's simple but a bit limited solution.

2.0.98.2 (2022-12-28)

Impulse response

• Freeze Y checkbox renamed to IR max.
• Added IR max text box for manual scaling.

[Reet]

Thanks Kimmo and Happy New Year to you as well !