VituixCAD

Just another Moderator
Joined 2003
Paid Member
Wintermute, for the off axis angles, how did you measure them?

If you measured off axis angles by rotating the speaker, and keeping the mic at the same position, you most likely are not measuring the angle you think you are measuring.

This is a very subtle mistake that I've never seen mentioned anywhere, but it makes a fairly big difference. If you are simply rotating the speaker, you are most likely rotating at the wrong point of rotation. If you rotate the speaker like how a turntable would rotate the speaker, then the point of rotation is at the center of the speaker. This is wrong, because the center of rotation needs to be at the driver you're measuring. You can easily see why this is wrong. Place the speaker on axis, and have the mic point directly at one of the drivers, say the tweeter. Then rotate the speaker 45 degrees. Notice the mic is no longer directly pointing at the tweeter. In fact, it is probably not even pointing at the front baffle! The angle between the mic and the tweeter is not 45 degrees, but an angle greater than 45 degrees.

The rotation point is around the front baffle, the center of rotation in line with the center X direction and should be zero in Z direction. Pictures of measurement setup attached. The front of the baffle is flush with the 90 deg line on the card (which in turn runs through the centre of the pivot bearing). I think that this is how it is supposed to be?

Tony.
 

Attachments

  • rotation.jpg
    rotation.jpg
    302.7 KB · Views: 171
  • setup.jpg
    setup.jpg
    968.7 KB · Views: 178
Just another Moderator
Joined 2003
Paid Member
Few main things imo:
* Timing of all measurements are normalized to known physical origin i.e. center point of driver on baffle surface or distance differences from mic to driver's origin are known, that measurement data can be located (with X,Y,Z mm parameters) to virtual 3D space in simulation. Semi-dual channel measurement mode or constant latency gear without IR peak normalization is must.

Hmmm this try bit I do not fully understand. All of my measurements were locked to time zero in holm impulse so are all relative to the original on axis mesaurement but maybe this is not correct.

Tony.
 
Just another Moderator
Joined 2003
Paid Member
OK I have something to go on. (though not sure it is going to help) will report back.

It appears that my very first measurement of the tweeter, which l locked time zero on (and then did all subsequent measurements relative to) is off. I'm not sure how it happened as I did at least three or four test measurements after it to make sure everything was working well. The funny thing is that every measurement after that (woofers and tweeter) is actually aligned to the LAST tweeter measurement I did. I'll have to go through and reprocess all of these to be aligned to the last tweeter measurement, and see what the results are.

A quick check replacing the 1st tweeter on axis with the last and not changing anything else, didn't make any difference however...

I may need to redo the measurements taking individual off axis measurements of the two M's on axis with them rather than all measurements on axis with the tweeter (with both M's playing). I knew that would be more accurate, but I took the lazy way out).

edit: actually the impulse is shifting back as the angle increases, is this expected?

Tony.
 

Attachments

  • time_shift.png
    time_shift.png
    61.2 KB · Views: 189
  • time_zero.png
    time_zero.png
    60.3 KB · Views: 168
  • time_shift_2.png
    time_shift_2.png
    56.7 KB · Views: 158
  • time_shift_3.png
    time_shift_3.png
    57.8 KB · Views: 65
Last edited:
edit: actually the impulse is shifting back as the angle increases, is this expected?

Yes, because the nearest sound source - edge area of the cone is moving closer to mic. Therefore starting point of time window should be set with off-axis measurement to 70...90 deg of the largest radiator. Finally, timing of all measurements is worth to bias with negative delay (common for all measurements) in the simulator to shape raw/unfiltered phase response of tweeter close to minimum phase. Too much phase wrapping in axial response looks fuzzy, and some point it will interfere calculation too.
Another option to compensate too early starting point of time window is to use "PreDelay" which decreases group delay of measurement. That parameter is available at least in ARTA and Clio. REW has "left window" which takes care of small variations in IR starting point due to speaker rotation. Other programs should have that kind of windowing system too.
 
Thank you Kimmo. In order to create power response:

1. can I use any measurements as A curves to be added in calculator? What is the name and the path of this file when saved? Is it enough to have proper names tested with proper ending for angles?

2. Can I insert that file as power response of a driver?

3. I tried to add more than two frequency responses for a driver but when I click the second file it replaces the previous one and cannot be added. I tried also to drag and drop all files together but also it cannot work. What to do?

When i click a frequency response it is not added but replaces the existing one.
When I try to drag and drop all the frequency response files together, it is not working. What to do?
 
Just another Moderator
Joined 2003
Paid Member
Perhaps a screenshot will help.

Click on the folder for the driver in question.
select the first file from the file dialog.
hold down the shift key
select the last file from the dialog.
click open

I'm assuming here that you are using windows.

Tony.
 

Attachments

  • select_multiple.png
    select_multiple.png
    110.7 KB · Views: 357
1. can I use any measurements as A curves to be added in calculator? What is the name and the path of this file when saved? Is it enough to have proper names tested with proper ending for angles?

2. Can I insert that file as power response of a driver?

Calculator tool is able to calculate power response, but it has some limitations (which main program does not have): angle step should be constant and full 0-180 deg covered (except with 2pi wall speakers) to get close to correct result.
Angle coding from filenames is defined in Options window, and the result is visible in A response list. But that response wouldn't be valuable for crossover simulation with VituixCAD because main program does not support power summing and calculates everything: power response, DI spectrum, individual off-axis responses, driver responses, phase responses, group delay, impedance response etc. with simulated crossover and loaded off-axis responses and impedance responses.
I suppose you are planning to simulate speaker with power responses of individual drivers? There may not be any workaround available for that method.
You can load 1 or 2 power responses as a reference response to Power & DI chart, but I don't see good reasons to do that.
 
Kimmo, Tony and Pavel, thank you all for your prompt replies.

Opening a choice of files to cheat the Power response works! After using that some time I get even more excited!!!

What I want to do is to use the two targets trying different references, e.g.
Listening window instead of 0 degrees, or
rough Power using only 14 measurements every 45 degrees or
only horizontal power, or
early reflections instead of power.

Once optimizer achieves smooth responses with smooth DI, all above methods may have similar results with changing the tilt of the power target, or changing the weighting factors.

My goal is to save some time, combining these functionalities in order to optimize multi drivers placed in various shapes within 360 degrees placement and not only on the front baffle in order to control further the directivity.

The more I use the program, the more excited I get! Many thanks again!

Vassilis
 
Version 2.0.1 with smart library blocks is out.

Blocks are still inserted as exploded - not physically grouped, but they are possible to select and move together by visiting in Tune block window. Components in the block are linked to "block header" with GUID. Block header contains common data for the block. Block attributes could be user questions, constants or math expressions.
Attributes are possible to group to Options. For example different filter types such as Bessel, Butterworth etc. could have own questions, constants and expressions.
Parameters of crossover components have new Expression field for math expression or variable name (if expression in the block).

Block tuning window:
An externally hosted image should be here but it was not working when we last tested it.

Purge command cleans expressions and links to block header if they are not needed anymore.

Parameter design and maintenance window:
VituixCAD2_BlockParams.png

Expression parsing is tested with Test button or by changing Option.

Almost all library blocks are parameterized.

More information in User manual on pages 9...14.
 
Hi,


1. How to optimize one active filter for a powered speaker with two way passive crossover?

We can put the same active filter in front of both ways but when running the optimizer they should both take the same values. Is there another way?

2. I had to change from active PEQ to active IIR , but after that opening the enclosure tool it cannot calculate SPL with crossover. Other tools like merger and calculator are working! Is there a solution?
 
We can put the same active filter in front of both ways but when running the optimizer they should both take the same values.

Are you using version 1.1? That can also handle common active filter for multiple passive ways but only with tricky workaround: whole speaker should be built inside the same way by locating components to "driver nets".

Version 2 is more logical because you can split output of active section to multiple passive "ways".

Attached/linked crossover schematic would help me to understand what kind of connection (and version) you have.

2. I had to change from active PEQ to active IIR , but after that opening the enclosure tool it cannot calculate SPL with crossover. Other tools like merger and calculator are working!

If you mean that SPL graph in Enclosure tool is not changing when you modify the crossover, the reason is probably that correct driver is not selected in the main program, or Crossover of current driver is not checked.
But, I may not understand your question because merger and calculator have nothing common with that feature.
 
Just another Moderator
Joined 2003
Paid Member
I've just tried the spl and impedance tracing in vituixcad. It is awesome!! I can't believe how good it is!! Even a terrible image it just nails it, and the few places where it gets it wrong, a couple of mouse clicks and it is corrected!

Seriously good, and will save a heap of time! :)

To get this to trace with the tool I had been using, I needed to go over the curve in another colour first in paint to have it even have a chance of finding the curve!

Tony.
 

Attachments

  • scw636_spl.png
    scw636_spl.png
    148.3 KB · Views: 254
  • spltrace.png
    spltrace.png
    215.5 KB · Views: 251
Scanned graphs could be quite much rotated and also bent but that looks decent without editing.

SPL Tracer has some difficulties with crossed lines, but image rubber helps. Once I tried to make forecast function for guiding over the crossroads but it wasn't success. Forecasting won't work well with rapid changes in direction.