VituixCAD v2

[technodanvan]

You're always welcome here! Hopefully someday I'll understand everything your software can do.

[DaveFred]

draki and Dave are hoping that smug and frustrated Finnish Vituix Man speaker dictator would leave diyaudio for good and start posting here Let's see what happens...

OT comment: My background is woodworking, and my day job is selling woodworking machinery to people with widely varying skills and knowledge. Much like off axis measurement are necessary for a "proper" crossover, face jointing (the operation of making the face of a board flat on a long cast iron bed surface planer) is a first necessary step to proper dressing (taking from rough cut to useable) lumber. Every day I have to deal with people who say they just run both sides through a thickness planer and that is good enough. I have to hold my tongue and nod when inside I am thinking, "You are an idiot".

[kimmosto]

draki and Dave are hoping that smug and frustrated Finnish Vituix Man speaker dictator would leave diyaudio for good and start posting here Let's see what happens...

Just happened that smug dictator moved from diyaudio to here ASR forum would have XenForo and https, but review approach and 'Desperate Dealers Forum' i.e. journalistic flavour does not fit for design tools and human beings possible working also for commercial manufacturer(s). So let's continue here!

[kimmosto]

My last words on diyaudio referred to 'some conditions'. This forum has basic rules. Personal hopes including predicted reactions concerning this thread start with:

1. Users should try to follow recommended measurement method, tools and settings to capture timing differences of drivers, and recommended design procedure to enable crossover simulation with quasi full space (dual plane) response data. Using response files downloaded from manufacturers (measured to half space possibly/probably unknown volume) or tracing responses graphically from datasheets will be actively discriminated in this thread regardless that VituixCAD includes also tools for capturing and pre-processing that kind of driver data.

2. VituixCAD includes features of many other design tools targeted for DIY, and offers some extra/professional features too. Some tools VituixCAD is able to replace already: BoxSim, Edge, FPGraphTracer, FRD Response Blender and Minimum Phase Extractor, Frequency Response Modeler, splcopy, PCD, WinPCD, spltrace, UniBox, WinISD, XDir, XSim and XSim 3D. People are free to like whatever, but I'd like see VituixCAD the most preferable and recommended also outside this thread so my target is to improve if some individual feature is not the best. So do not write here or near by that "XSim is popular" or "I prefer The Edge because it's easier and I don't like to study anything new". Obscurity, half-truths and laziness are not valuable for anybody.

3. I'd like to avoid RTFM answers so read feature list, measurement instructions, user manual, changelog and check history on diyaudio and here to avoid repeating and copy-pasting from existing public documents, please.

4. Speaker projects simulated with VituixCAD could/should be in separate threads. Messages about features and difficulties with the simulator can be separated here. Measurement and design method published outside this thread should also follow recommendations to be correct and educative for other members - especially for beginners. Simulation of "serious" speaker project with single axial responses - measured, downloaded or traced will rise my blood pressure and generate disapproval so just don't do it.

5. Audio technology is close to science so SI units instead of imperial.

may continue...

[kimmosto]

6. Other platforms than Windows are not supported at the moment. Win7 or Win10 on virtual machine or multi-boot works.
I have investigated and studied Java which looks possible and interesting too, but time, energy and financing are the problems. Without brilliant GUI designer I would be fired and divorced and probably burned out in a year. For what or who as long as Windows is alive? wxWidgets doesn't look an option though performance could be significantly better in some functions.

7. Features which will not be added (without source code quite ready to just include in the project):
- Wavefront simulation
- Horn and waveguide simulation
- Transmission line simulation
- Special direct radiators such as resistive output port for cone
- Leaking radiators such as resistance enclosure
- Damping of enclosure
- Higher order diffraction, diffraction from bevel/facet. Ok I know how to do these, but laziness vs. value and performance...
- Off-axis from conical and flexible radiators (not just flat rigid circular)
- Optimizing of XO connection.

[JonMarsh]

I can live with that...

But these days I can’t live without VituixCAD! I’m hooked!

[DaveFred]

Geez, what's left to talk about?

[kimmosto]

Measurement instructions cover just very basic concept with small direct radiators measured with time window in limited space at home i.e. for the most limited and difficult conditions. Real life has much more possibilities to improve. For example few...some users already have decent acoustic lab or Klippel NFS with proper options or large free space or good conditions for ground plane or at least more space for longer than 100 cm mic distance without need to shorten time window.
User manual is full of big holes and ambiguities based on my assumptions and hopes about users' knowledge.
Some features possible for me or code available as open source or by private contributor can be added. Enclosure, diffraction and merger tools could be combined into single massive and complex pile - at least in theory, etc.
More philosophical discussion; what is important, when (=what cases) and why, and what looks just overshoot in my or your opinion.

So I'm quite confident that we have enough to discuss

[Quza]

Seems I have missed you by a few hours for my first question on DIYaudio

Is it valid to correct ground plane measurements by the difference of the diffraction simulations with normal baffle height and twice the baffle height? And if so, am I to apply that correction to the off-axis measurements as well? (https://i.ibb.co/ydgFD3y/diffraktion.png)
I got lucky and have epoxy screed floor in my garage, so a twice as long window for quasi anechoic measurements with the speaker on the floor compared to the speaker at half the room height.

And thanks for everything you've done for the DIY audio community. VituixCAD was a game changer for me!

[kimmosto]

Is it valid to correct ground plane measurements by the difference of the diffraction simulations with normal baffle height and twice the baffle height?

Ground plane measurements are far field measurements => goal is to measure so that mathematical corrections to response shape are not needed i.e. responses are directly valid for HF response list of Merger tool with level scaling by -6 dB only. One error source with ground plane is that (small) speaker on the floor is not exactly the same as speaker on final stand due to differences in baffle loss and diffraction. One workaround is to make small stand to cause diffraction and baffle loss also below bottom panel, and tilt speaker down so that DUT points to microphone while measuring 0 deg response. Rotation center should stay in the center point of DUT on baffle surface, and manual rotating table should be thin or extended to mic or some other invention to avoid diffraction from ground plane. This kind of tricks are easier with small speakers. Tall constructions need supports, turning upside down etc.

[kimmosto]

Diffraction simulation is needed for conversion from near field LF measurement to far field LF response to full space. Sometimes for generating directivity for single axial LF response of boxed driver. Diffraction simulation with double height could be useful to imitate change in baffle loss caused by vicinity of the floor if position of the woofer is very close to floor e.g. 20 cm. Diffraction tool cannot seal gap between bottom panel and floor so double height is cheap trick to reduce error.

[kimmosto]

...tilt speaker down so that DUT points to microphone while measuring 0 deg response. Rotation center should stay in the center point of DUT on baffle surface

Problem with this is that hor 180 deg measurement is not 0 deg in vertical plane if rotating table is horizontal on the floor. So rotating device should be short axis which is tilted front - towards mic

[Quza]

Thanks for your quick response(s), Kimmo!

You're partially answering my questions while I'm writing my answer. Witch! 8O

I'm planning on getting a square board and add some height-adjustable feet on the back, so I can angle the speaker at the microphone and rotate the speaker on the board manually while staying in that plane and distance. I can't imagine an easy way to make this with a stand :/ With a second, thinner board on the first one and a nail I might even create a turntable, if turning it manually gets too annoying.
So there will be changes in frequency response and while they are minor as the simulation shows (at max 1dB), they are broadband. It's easy to correct the FR for this deviation, so why not? After all - cheap trick or not - it should be closer to the actual response of the speaker when used 'normally', right?

/edit: Attachments are full size and can't be deleted, Jesus! Sorry!

[kimmosto]

Exactly. Here is ground plane setup with bottom loss for mid-woofer of small standmount/bookshelf speaker.

[kimmosto]

Measurement instructions for ARTA, REW and SoundEasy updated and pdf files renamed 2021-04-30. Added few comments about large constructions, deep horns, long planars/ribbons and ground plane measurement. Not much better that earlier, but hopefully helpful for someone.

[bnesic]

Thanks Kimmo!

[Bear]

Hopefully someday I'll understand everything your software can do.

I won't, and I am very much okay with that. I do enjoy what I can learn along the way, though. I just need a calibrated turntable for the measurements. My Earthworks M30 finally needs to be used in anger, a decade later.

[Bear]

Thanks for this, Kimmo, truly. The fact that it is free for us is still mind-blowing. I am just learning, and I will never realistically devote the time/energy to fully appreciate all that this does, and the ocean of talent and capability that has gone into it. For people like me, though, I'd like to tack on a couple of comments so that we don't lose you here (I expect that our moderators here are more sympathetic to your way of thinking anyway...).

To others like me...
1) In reading the DIYA threads, people seem to lose sight of the realities of manufacturing. In the real world, quality costs money, and a manufacturing line doesn't have the level of precision that lay people believe. And that's before factoring into that process that speaker drivers often require parts like "glue" that are applied by hand. At some point, people want to pick at issues that are well beneath the "noise floor" of the manufacturing process, and that's silly. Your patience with these types of issues questions was Job-like.

2) Numbers are self-credentialing. People want to believe numbers. This is actually a cognitive trap that many engineers fall into with particular fervor because they forget that numbers themselves are generated by a process that has error and probability (human, quantum, etc.). In my own field, Predictably Irrational, by Dan Ariely, is a fun and easy read that should be humbling for all of us (even the title alone!).

3) TANSTAAFL - I love things like PCD and being able to trace factory curves (since manufacturers won't publish the data sets for the masses, with a few exceptions). It helps for things like doing bench racing and desktop R&D for those without a dedicated R&D team (Jon/ET!!), but no more than that. If one persists, though, will one's amazing no-measurement approach make noise? Hopefully!! Will you be able to convince yourself that it's better than a 30 year old Bose Acoustimass system you found at the garage sale? It may actually be better! But it is not in any approximation an "optimal" outcome for the component parts.

4) Much like wine tasting, so much of audio plays DEEP into the flaws in human perception and cognition (numbers being self-credentialing is but one flavor...). Don't believe me? Read this: https://en.wikipedia.org/wiki/Cognitive_bias. If you still don't believe me, then well.... Confirmation Bias and Cognitive Dissonance are truly some of your very best friends, so go play with them.

Since I tossed a Heinlein reference into the mix above, I'll throw Orson Scott Card here: I am not the speaker for the dead. Others can and will disagree, but I'd share the aspiration to improve the signal to noise ratio here.

[kimmosto]

I expect that our moderators here are more sympathetic to your way of thinking anyway...

Difficulties on diyaudio and earlier on some other forums are kinda my problems. I could just ignore everything else than features and using of VituixCAD; stay away from other threads etc. I've tried that, but not succeeded very well.
Phrases such as "experienced designer knows and is able to handle restrictions of tools" or "everyone has own learning pace" are just empty words. Certainly don't stand for "we should primarily recommend the poorest methods and tools especially for beginners". "Experiences designer" may have significant number of designs and the first one designed decades ago, but it doesn't mean that designs are good and process was effective, controlled and repeatable with other concepts and suitable for other listening environments/acoustics. Recommending the simplest and poorest methods and tools for beginners sounds just intentional harm or assumption that all beginners are too ignorant, stupid or lazy to rise both personal/learning and design targets.
VituixCAD will never be all inclusive and the best and most accurate in everything so target is still mediocre, and this is not a mission with too big thoughts. Just something above the poorest crap.

[kimmosto]

I love things like PCD and being able to trace factory curves ... It helps for things like doing bench racing and desktop R&D for those without a dedicated R&D team...

Tracing is not needed for checking are sensitivities of drivers compatible for passive crossover or are drivers compatible with selected enclosure type. Active filters do not have sensitivity limits. So "value" is to produce something real - a preliminary filter where designer can start subjective tuning by listening or measuring (whole construction) or leaving as is. In addition, method is slow and not usable for special radiators because simulation for conversion from manufacturer's data (to half space in some/unknown volume) to far field in full space is not available. Some demand exists for sure, but how much attention is enough.

[kimmosto]

Some internal changes done to Diffraction tool in the latest build of 2.0.67.0 (2021-05-03). Function should be unchanged so please report if something strange is found.

[draki]

I am using ARTA for measurements but have recently tried REW, and all works fine.

However, as far as I can tell, there is no auto signal generating sequence (which ARTA has) so it makes for many jumps from the measuring stand for the (manual) next rotation and back to the PC desk for hitting the start button.

Or there is such an option with REW that I am not aware of?

[kimmosto]

There is no auto continue with pause, but someone could ask it from JohnPM. Needs Pause and Last number parameters to auto numbering window. Would be close to ARTA and ARTA Recorder. Hitting start is required if pause = 0 s (or separate Auto/Manual radio buttons or listbox). Something like this

[kimmosto]

Locating laptop (on the floor) close to turning table is quite adequate with manual table.

[Bear]

Tracing is not needed for checking are sensitivities of drivers compatible for passive crossover or are drivers compatible with selected enclosure type. Active filters do not have sensitivity limits. So "value" is to produce something real - a preliminary filter where designer can start subjective tuning by listening or measuring (whole construction) or leaving as is. In addition, method is slow and not usable for special radiators because simulation for conversion from manufacturer's data (to half space in some/unknown volume) to far field in full space is not available. Some demand exists for sure, but how much attention is enough.

In defense of the proletariat among us, what you've just described is about 19 steps ahead of what many of the 'hobbyists' want to do with such things. I play around with PCD (because I'm an Excel geek) and factory data to get a feel for how the bits and pieces interact and to play (e.g., What would a 250Hz passive filter look like with these drivers? Do I need $100 worth of capacitors to save $60 in driver costs vs an alternate approach with a different driver that had a broader usable range?). The useful part, to me, is supporting my ability to learn by seeing things in action, and then to play around with what would otherwise be called "rough order of magnitude" (ROM) budgeting numbers.

But that's the view from a single prole.

[kimmosto]

This is enough for PCD and other Excel tools or anything already covered in VituixCAD.

As mentioned, target of VituixCAD is to replace e.g. Bagby's sheets so we can discuss here what should be better / why someone prefers something else and the reason is related to features or usability - not knowledge, motivation or personality of the user or me.

[kimmosto]

2.0.67.1 (2021-05-05)

Convert IR to FR
* Added support for wav exported from Praxis. Accepts extra format bytes before data chunk ID.
* Added support for MLSSA txt exported from Praxis. Accepts " 1" as the first line.

[kimmosto]

2.0.68.0 (2021-05-06)

Convert IR to FR
* Added Energy time curve (ETC) to impulse response chart.
* Added Step response to impulse response chart.
* Added context menu for impulse response and frequency response charts including Traces, Save as overlay, Open overlay, Copy image and Export image commands for both. Show ETC and Show Step for impulse response. Show Minimum phase and Show Excess phase for frequency response.
* Added scaling of frequency response maximum and span with mouse wheel.
* Added overlay with drag&drop to frequency response chart.
* Removed snapshot buttons and checkboxes.
* Removed square root, MP and XP checkboxes.

[JonMarsh]

2.0.68.0 (2021-05-06)

Convert IR to FR
* Added Energy time curve (ETC) to impulse response chart.
* Added Step response to impulse response chart.
* Added context menu for impulse response and frequency response charts including Traces, Save as overlay, Open overlay, Copy image and Export image commands for both. Show ETC and Show Step for impulse response. Show Minimum phase and Show Excess phase for frequency response.
* Added scaling of frequency response maximum and span with mouse wheel.
* Added overlay with drag&drop to frequency response chart.
* Removed snapshot buttons and checkboxes.
* Removed square root, MP and XP checkboxes.

Big update!

[kimmosto]

2.0.68.1 (2021-05-07)

Impulse response
* Added Energy time curve with fixed 0...-60 dB scale. ETC checkbox to show.

[draki]

There is no auto continue with pause, but someone could ask it from JohnPM. Needs Pause and Last number parameters to auto numbering window. Would be close to ARTA and ARTA Recorder. Hitting start is required if pause = 0 s (or separate Auto/Manual radio buttons or listbox). Something like this
[ATTACH=CONFIG]31129[/ATTACH]

I asked and JohnPM responded. He says he is looking into it but may take a while until actual implementation.

[kimmosto]

^Ok. Hopefully pause time will also be before the first measurement to allow movement closer to manual table before the first rotation.

[kimmosto]

2.0.68.2 (2021-05-09)

ETC curves in Convert IR to FR and Impulse response windows were not official/standard ETC so they are renamed to 'Impulse dBFS' which is compatible with naming in REW.

[kimmosto]

2.0.68.2 (2021-05-10) part II

Main
* Error messages due to conflict between frequency response file naming and plane & angle parsing settings extended. Naming requirements are written to five different places in instructions, but it hasn't been adequate in practice. The most common RTFM subject in my inbox. Let's see whether six places is enough.

[kimmosto]

Most of you probably follow also VituixCAD thread on diyaudio. Some messages might be interesting or short or incorrect so we can extend discussion here. Direct replies should be there of course, but at least I won't write them.

For example:

Quickly went through the Paul Carmody pages. Unfortunately the procedure described there only works for so called on axis measurements. If that is good enough for you XSim might be easier for starters.

I do not agree. For example value/parameter tuning is much easier in VituixCAD. In addition, XSim does not show source file and settings of driver responses, and requires response loading for each driver instance in XO. That slows down designing especially with multi-driver projects and "forces" designer to verify origin and settings of the responses in practice.
I recommend to avoid all single axis tools (XSim, Bagby's workbooks, ...) due to serious lack of features and need to study new tools if targets will be higher some day.

The beauty of VCad however is the capability to optimize the power response on the basis of off axis measurements... Those -2 channel- measurements and the optimization on that basis is superior, but will take you into a steep learning curve.

Since revision 2.0.66.0 (2021-04-20) VituixCAD is able to optimize also predicted in-room response / CTA-2034-A. That is combination of listening window (12%), early reflections (44%) and sound power response (44%). This enables weighting also for early reflections which can be significant, though healthy designs of common concepts have close to same features in early reflections and sound power i.e. early reflections should not include much new or different. Exceptions exist so this gives more options.

off axis measurements with 5 degree increments

Coverage and angle step of measurements could be anything constant or random or subjectively selected. For example 0, 5, 10, 15, 20, 30, 40, 50, 60, 70, 90, 120, 150, 180 deg. Program is able to calculate just measured off-axis angles or interpolate to predefined angles with constant step using four (4) measured responses.

[kimmosto]

Monologue continues...

2.0.68.3 (2021-05-12)

Main
* Fixed smoothing of group delay curves to allow also negative GD.

Convert IR to FR
* Added Impulse response envelope (ETC) to IR chart.
* Added Show impulse response to context menu of IR chart.
* Added Right window mm text box.

It's not yet mentioned in user manual, but there is window function in ETC calculation. IR peak is automatically detected for ETC (not using Reference time set by user) so positioning of window function will fail if reflection peak is higher than direct sound. I will change if this causes too much trouble. Left window is Half Hanning from 0 ms to impulse peak. Right window is also Half Hanning from impulse peak to the end of ETC buffer. Maximum number of samples in envelope buffer is 32768 to maintain FFT and IFFT performance while scrolling file list up/down. ETC is shorter and faster if measurement sequence is shorter than 32k. Window function is the reason why ETC could be below impulse dBFS within 0 ms and impulse peak, and in the end of ETC buffer.

[kimmosto]

2.0.68.3 (2021-05-12) part II in addition to previous

Convert IR to FR
* Right axis of impulse chart changed from % to dB for Impulse dBFS and ETC.
* dB span of impulse chart adjustable by pressing Ctrl key while clicking expand/compress buttons.

- Color of impulse dBFS curve changed to dark cyan.
- Distance (mm) to Reference time point is visible in tooltip while dragging Right window.

[Reet]

What is the importance of ETC when it comes to converting impulse response to frequency domain? Is it just to aid in locating reflections?

[kimmosto]

^Reflections are much easier to locate from ETC than impulse in dB because ETC does not have numerous dips due to phase variations. ETC calculation includes Hilbert Transform which is simple in frequency domain. Required math is explained e.g. in user manual of ARTA: 6.3 Impulse Response Envelope (ETC – Energy Time Curve) on pages 102-103.

[Reet]

Thank you!

[kimmosto]

2.0.68.4 (2021-05-14)

Convert IR to FR
* Added IR time scale minimum text box.
* Added IR time scale maximum text box.
* Added Skip samples text box.

Ignoring of sample padding is done while reading IR file so correct number e.g. 48 samples with Praxis at 48k is fastest to enter with keyboard (not with up/down keys or mouse wheel). Length of impulse buffer is not truncated -> value of remaining samples in the end will be 0. Program remembers previous Skip value for wav and txt files. ARTA's pir has 300 extra samples and CLIO crp, mls and mlsi have 0.
Everything in time axis scaling is reprogrammed due to new min and max text boxes. Major grid interval is more dense and adaptive (previously fixed span/4).

Function of expand/compress buttons is harmonized within this tool. Other tools may work opposite Logic is imitated from CLIO.

Window function was accidentally doubled for ETC in the previous build, but it did not have much effect. Just level drop in the beginning and end.

[JonMarsh]

Would you like me to turn this into a "sticky" thread so it's always at the top?

[kimmosto]

Thanks, but I believe this stays at the top automatically or easy to find as long as upgrades are informed. Otherwise it's quite silence. Maybe my toxic conditions were too scary

[JonMarsh]

OK- that's why I asked, instead of just doing it!

[kimmosto]

Pinning is valuable for popular and important "read only" threads to avoid sinking and disappearance. VituixCAD was - and still is the most popular thread (by number of messages) on diyaudio's software area so it stayed at the top without pin. That could happen also here if we have something to discuss. Automatic update feature should guide to read changelog2 so plain copies from that file are not very important.