VituixCAD v2

[kimmosto]

In any case, latest release at least applies the calibration arithmetic across the entire measurement without windowing, so hopefully there are no further causes for concern. 🤞

Time window limit is fixed, but response compensation with loopback does not actually work. Program also jams with (much) longer than L=125ms, R=500ms time windows when "Merge loopback" option is selected. All this is written directly to AV Nirvana to avoid linking, repeating and "human messenger" I hope that also ea revisions would be properly tested before publishing. At least I don't like to be a mandatory beta tester for other than VCAD.

I could accept high-pass filtering in stimulus and mathematical compensations for protection. That would not need actual HP filter and loopback for calibration. Loopback would be needed for timing only which has been okay for a long time. More about manipulation of stimulus in some of my my earlier message. JohnM to consider this as an alternative to start sweep at midrange..low treble.

[Upsetter]

Sorry for being a noob, but are there a way of verifying if this issue with REW messes up my measurements? I did over 100 measurements with REW 5.20.13 and are getting strange results in my attempts to simulate in VituixCAD. I was certain it was my incompetence, but after banging my head in the wall for days I started searching and found the discussion around the REW issue. Frankly I don't totally comprehend the issue at hand, but I get some phase issues I think.

[Reet]

You haven't really provided enough information to know what you've done, or if there's even a problem. It should be obvious that D1 and D2 have differing phase though, from simply looking at top left graph, or interference null at 10kHz. it could be timing error, could be distance mic to speaker, possibly something else.

What was the loopback method used? What was measurement conditions for each D1 and D2 of same driver? Mic on axis with driver at 1m, or something else?

[kimmosto]

Drivers are in the same X,Y,Z location (which is impossible in practice when they use common data) and the other is inverted, but is another driver rotated with R parameter (or tilted with T parameter)? That would be my first guess, because as far as I know program does not have a bug in summing acoustical responses of two identical drivers in the same spot.

Also angle coverage of measurement data might be narrower than selected space for simulation.
Anyway, should be nothing to do with REW.

[Reet]

I didn't notice the inverted one! 😩

[kimmosto]

(which is impossible in practice when they use common data)

Actually this is technically possible, but not recommended in measurement instructions: For example, if concept is symmetrical dipole with two monopoles, rotation center can be between the drivers inside the box, and excess phase (=delay from driver to rotation center) of measured one DUT should be included in frequency responses i.e. reference time while FFT must be at rotation center too. With that kind of setup location of both driver instances in XO can be X,Y,Z=0,0,0 mm, and rear driver inverted and rotated back with R=180 deg, T=0 deg.
If complete construction with both drivers connected to the power amplifier is measured, crossover schematic has only one driver instance because it has frequency and impedance response data from both physical drivers.
In all cases when driver instance is rotated with R or tilted with T, measurement data should include sector covering also effect of rotation and tilting. Typically simulation having any driver rotation is best to have full space angle coverage to provide e.g. on-axis response for drivers which are rotated back with R=180 deg. Because after rotation with R=180, 180 deg response become 0 deg. Challenge increases with half space designs having rotated or tilted driver, because data is needed also from off-axis > 90 deg.

[Upsetter]

Thank you for replies, I will try to provide more sufficient information.

- All measurements taken at 1 meter, semi dual channel, 4ms gating. All settings followed instructions for measurements with REW, except I started measuring at 400Hz instead of 5Hz because of no DUT protection (I would not think extrapolation in low frequencies to cause issues in other end of spectrum)
- Driver in example is a GRS open back AMT tweeter (NEO 3 clone), measured nude, no baffle
- Measured 0 - 180 degrees in 10 deg. steps. Filename example: "GRS PT2522 raw hor 10"
- Checked impulse response on all measurements, some had REF on reflection peaks around 90deg, so I shifted those before exporting
- In simulation the "second" driver is inverted, Rotated 180degrees and tried delayed with Z=driver half dimension(38mm)

I get that 10k dip is cancellation due to phases being 180deg apart, but shift in Z dim only make matters worse. In the first example indeed X,Y,Z was identical, R was 180. If I plot in a positive number in Z, more severe dips occur. Negative Z gives more useful result so I suspect some kind of timing issue.
I did check all the impulse responses from my measurements to se if something was reversed, but they look correct to me. Surely I have done something wrong, I just can't find the culprit it seems

Z=38mm

Z=0mm

Options settings

[Upsetter]

I did some more brain-storming around my issue, and maybe I found something in my measurements. When Z=0 I get that dip around 10k, which is around 34mm full wavelength. In my measurements delay in relation to loopback varies from around the same approx. 30mm, except the ones around around 90deg, which is several meters. I don't understand what setting I did wrong, but I thought when gating all measurements before room reflections I would not get impulse Reference at a peak at 9-10ms (with 4ms gating). Driver is 9mm thick and centred at rotation point, so to me it makes no sense that 0deg has 15mm delay, 180 has 34mm delay. 20deg has 5mm(!) but corresponding 160deg has 34mm (same as 180, 170, 150 etc).

I see that my issues is probably not related to VituixCAD, but some other issues in REW or my execution. If I am cluttering this thread please let me know, and I will address it elsewhere. Thanks in advance.

PS; I tried attaching zipped Mdat file, but file size is above limit

[kimmosto]

Maybe it's best to wait for next REW revision (ea27) where at least most of the problems related to dual channel measurements should be fixed. After that I can update measurement instructions.

I would probably measure AMTs as mechanically and electrically complete package of two drivers = as a single dipole driver. This eliminates need to add second driver instance to XO, rotate/tilt, invert and set some Z mm to rear driver. This should not change much with correct measurement process, but it eliminates quite many variables from the equation. Variables are sensitive to incidents, accidents and mistakes.

[Upsetter]

I would probably measure AMTs as mechanically and electrically complete package of two drivers = as a single dipole driver. This eliminates need to add second driver instance to XO, rotate/tilt, invert and set some Z mm to rear driver. This should not change much with correct measurement process, but it eliminates quite many variables from the equation. Variables are sensitive to incidents, accidents and mistakes.

Just to be 100% certain I understand correctly - Does this mean since my AMT driver is open back and radiates about exactly the same response to the rear as the front, then it's possible to just use single driver in XO sim, with the 0 - 180deg measurements?

Anyways, I will look for REW update, and your updated measurement instruction. Thank you

[kimmosto]

Just to be 100% certain I understand correctly - Does this mean since my AMT driver is open back and radiates about exactly the same response to the rear as the front, then it's possible to just use single driver in XO sim, with the 0 - 180deg measurements?

Sure. That's how measurement and simulation should be done with every naturally dipole radiator. I thought you have built a dipole radiator from two separate sealed AMT drivers.

[kimmosto]

Next time photo first that we can evaluate is simulation (XO network and all parameters of driver instances) okay for mechanical and acoustical construction.

[kimmosto]

What would you think if recent project list is in alphabetical order and without directory name? Directory would be visible in tooltip.

[Upsetter]

Sure. That's how measurement and simulation should be done with every naturally dipole radiator. I thought you have built a dipole radiator from two separate sealed AMT drivers.

Ahh, that explains a lot! I've been hung up on what information on dipole simulation I have been able to piece together from various forums etc. I never totally understood why I would use multiple driver instances for dipole of already measured dipole constructions.... I have a habit of overthinking/-complicating things.

Thanks for your patience, and sorry for making guess-work of it. I will remember to provide pictures next time. I have my construction in pieces now and no good photos at the moment

These are from early iteration, no longer MTM mid cone drivers. Now the AMT tweeter sits on top of 8" AMT mid, rest is identical

[Upsetter]

What would you think if recent project list is in alphabetical order and without directory name? Directory would be visible in tooltip.

My project list is not very long yet, but seems like a good way to list them. Directory path is long and not needed in that list, for me at least

[Reet]

What would you think if recent project list is in alphabetical order and without directory name? Directory would be visible in tooltip.

I prefer order of last accessed to be honest (how it currently operates) and is consistent with most any other application "recent" list, I believe.

[Reet]

Ahh, that explains a lot! I've been hung up on what information on dipole simulation I have been able to piece together from various forums etc. I never totally understood why I would use multiple driver instances for dipole of already measured dipole constructions.... I have a habit of overthinking/-complicating things.

Thanks for your patience, and sorry for making guess-work of it. I will remember to provide pictures next time. I have my construction in pieces now and no good photos at the moment

These are from early iteration, no longer MTM mid cone drivers. Now the AMT tweeter sits on top of 8" AMT mid, rest is identical

Interesting speaker. To echo Kimmo's comments, just measure full 180 degrees of the speaker as you normally would, there's no need for special tricks for the baffle-less dipole arrangement.

[kimmosto]

I prefer order of last accessed to be honest (how it currently operates) and is consistent with most any other application "recent" list, I believe.

I'm extending list for 30 files. Something should be done because certain project is quite difficult to find.


Here the last one is up, but without directory name:

[kimmosto]

I will add this for myself

[Upsetter]

Interesting speaker. To echo Kimmo's comments, just measure full 180 degrees of the speaker as you normally would, there's no need for special tricks for the baffle-less dipole arrangement.

Thanks! The MTM configuration was flawed I think, but there was definitely qualities in baffle-less sound that I never heard before.

[Reet]

Thanks! The MTM configuration was flawed I think, but there was definitely qualities in baffle-less sound that I never heard before.

I did a "hybrid" speaker with monopole bass, and dipole from >300Hz. It is a different sound, overall it is more spacious from higher reflected energy, but I find the image was lost compared to a standard monopole.

[kimmosto]

Maybe I leave this

[Upsetter]

Maybe I leave this

I think that looks good, being able to switch off "clutter" is a nice feature in my eyes, but different people have different preferences

[Upsetter]

I did a "hybrid" speaker with monopole bass, and dipole from >300Hz. It is a different sound, overall it is more spacious from higher reflected energy, but I find the image was lost compared to a standard monopole.

Good looking speaker! Is that baffle from bamboo? I've read about dipole imaging not being sharp as monopoles but they have been on my list of things I want to experience, so I had to try it out. Did you use VituixCAD for simulation of your design? If you have any tips I would love to hear them. The more I use and understand VituixCAD, the more I think it seems invaluable in a proper speaker design process.

[Reet]

Thank you! Yes, it is a bamboo countertop, 1.5". Bamboo is very strong, dimensionally stable and surprisingly little chipout when routering, though you will wear out blades quite quickly, similar to plywood it is full of adhesives to keep the small strips of bamboo together. If you like the look it's a great option for speakers IMO.

This speaker was originally designed before my VituixCAD days, started in SoundEasy, but eventually I did complete full spatial measurement and tweaked the filters with VituixCAD. It uses a Hypex AS2.1D plate amp, a DSP + passive filter approach, so there is a simple passive filter between the tweeter and midrange, and DSP covers the crossover between woofer and the top end as well as all the response shaping.

One thing learned through the process is that no speaker is truly dipole in the sense that the forward radiation is the same as the rear radiation. There will be difference in cone shape, driver frame and motors in the way, etc. One "trick" to easily see what the response of your speaker looks like from behind, is to simply check the "rotate 180 deg" box in the room tab. That said, I would not treat the crossover design process any different from any other speaker.

[kimmosto]

2.0.98.8 (2023-01-21)

Main

• Added 'Show directory' command to File Recent menu.
• Added 'Sort by pathname' command to File Recent menu.
• Added 'Sort by filename' command to File Recent menu.
• File Recent menu extended for 27 project files.

[kimmosto]

Hopefully this is not too early - again I tested REW V5.20.14ea27 quite shortly so there can be some combinations giving unexpected results. Radical bugs should no longer be found when measuring according to the instructions.

2023-01-21: Quick manual to produce measurement data with REW updated for revision V5.20.14ea27 (or later): VituixCAD_Measurement_REW.pdf

[Upsetter]

Hopefully this is not too early - again I tested REW V5.20.14ea27 quite shortly so there can be some combinations giving unexpected results. Radical bugs should no longer be found when measuring according to the instructions.

2023-01-21: Quick manual to produce measurement data with REW updated for revision V5.20.14ea27 (or later): VituixCAD_Measurement_REW.pdf

I might be missing something, but clicking the pdf link I can't see anything updated compared to the old document:

[Upsetter]

One thing learned through the process is that no speaker is truly dipole in the sense that the forward radiation is the same as the rear radiation. There will be difference in cone shape, driver frame and motors in the way, etc. One "trick" to easily see what the response of your speaker looks like from behind, is to simply check the "rotate 180 deg" box in the room tab. That said, I would not treat the crossover design process any different from any other speaker.

That is actually a very neat trick. Thanks! I did measure several nude drivers 0-180deg , and concluded exactly what you say - the response is never truly identical dipole. It's one of the reason I wanted to try the open back GRS planar drivers. The tweeter is somewhat disappointing though, transition from dipole to plain beaming is quite messy.

That bamboo is really pretty indeed. I will keep that in mind fot future projects.

[kimmosto]

I might be missing something, but clicking the pdf link I can't see anything updated

Refresh. Everything comes from some cache without refreshing

[Upsetter]

Refresh. Everything comes from some cache without refreshing

Off course, that was the case! Thanks

[kimmosto]

2.0.98.9 (2023-01-25)

General

• Phase wrapping removed before smoothing of phase responses.

SPL Trace

• Offset in magnifier with some file types (jpg) fixed by converting image to bitmap on file open and drag&drop.

[Quza]

Hopefully this is not too early - again I tested REW V5.20.14ea27 [...]

Too early as in 5.20.14 is not yet released? 5.20.13 doesn't seem to have the option for Tukey 0,5/0,75.

[Nil L]

5.20.13 doesn't seem to have the option for Tukey 0,5/0,75.

Yes. But in the assembly of early access 5.20.14 there is.

[kimmosto]

2.0.98.10 (2023-01-28)

Convert IR to FR

• Calibration file extrapolated outside frequency range as flat 0 dB/oct.

General

• .mic and .cal files extrapolated outside frequency range as flat 0 dB/oct.

This may look small and simple modification, but it's not. Control parameters for response file reader, frequency point resampling and response extrapolation are changed. >10 classes/forms and >35 functions are modified so heads up. Everything is tested, but something small but unpleasant may remain.

[kimmosto]

Too early as in 5.20.14 is not yet released? 5.20.13 doesn't seem to have the option for Tukey 0,5/0,75.

Please note that instructions are compatible with V5.20.14ea27 or later. That is the first version in REW's history where full dual channel measurement works so instructions were updated immediately for that. Do not try to use it with any older official release because they are not compatible. The latest available is V5.20.14ea28 built on Wednesday at 2:52 PM. These early access builds are available at

https://www.avnirvana.com/threads/v5-20-14-early-access-build.11154/

[Quza]

Thanks for the link and the answer!
So the full dual channel measurement is the added "Use loopback as cal [...]" I assume. Then I'd argue that your old instructions were already totally fine since I don't know any non-condenser measurement microphones and even cheap audio interfaces are almost perfectly flat in FR. Some examples from input measurements of Focusrite, Motu, Steinberg, Behringer, Presonus, Audient, Swissonic, Mackie 😀


















But it surely doesn't hurt to correct these tiny deviations, too.

[kimmosto]

But it surely doesn't hurt to correct these tiny deviations, too.

It's not about tiny deviations. Full dual supports equalization of stimuli i.e. you can add either passive or active high pass or shelving filter to signal and get automatic compensation with loopback reference so that transfer function of protective high pass or shelf is compensated and result is compatible with flat stimuli.
Traditionally REW has had possibility to start sweep at mid-range, but that usually increases error at the lowest measured octave. And everything below starting frequency is just extrapolation.

[Reet]

One nice addition the latest REW has as well is the ability to include high pass or any filter in digital domain as part of the sweep function using REW's EQ. This feature has many applications in speaker evaluations, and is a much better solution for speaker measurements that simply limiting the sweep frequency range wit a hard stop.

[kimmosto]

2.0.99.0 (2023-02-11)

Enclosure

• Added solver for passive radiator parameters. Target is measured impedance response loaded to impedance chart as overlay.
  
  

---

2.0.98.12 (2023-02-08)

General

• Root folder of folder browse dialogs changed to Desktop to show also Network locations.
  
  
• Commercial licensing changed.

---

2.0.98.11 (2023-01-30)

SPL Trace

• Tracing algorithm modified.
  
  =>
  Tracing of thick traces with multiple color hues and saturations and high slopes improved.

---

[PANDINUS]

Do I understand correctly that these two connections cannot be entered and simulated?

[Reet]

Did you try?

[Justin Zazzi]

Question:

I am trying to simulate two speakers in an enclosure while also using the crossover/dsp filters.
When I enable to crossover filters by checking the box in the screenshot below, the number of drivers is locked at 1.
Is there a different way to do this?

[Justin Zazzi]

Possible bug?

Please see screenshot below.

[kimmosto]

I am trying to simulate two speakers in an enclosure while also using the crossover/dsp filters.
When I enable to crossover filters by checking the box in the screenshot below, the number of drivers is locked at 1.
Is there a different way to do this?

Driver instance in the crossover is just one driver. Transfer function with given generator voltage is sent from the main program to Enclosure tool, and Enclosure tool sends simulated impedance response back to main program for passive crossovers. So it's the most logical and easiest to handle when Enclosure tool has one driver, and it's terminal voltage comes from the main program. Purpose of crossover link is to enable testing of power and excursion with some active bass boost and/or subsonic filter. Simulating with single driver provides that information. After this test you can scale the system up by adding drivers and volume, and keeping system resonance constant and vent capacity adequate.
Purpose of Enclosure tool is to select driver and calculate box and vent/passive parameters. Not to simulate whole project with crossover.