Frequency responses to simulated directions are interpolated from measured responses (because measured directions are not necessarily exactly equal to simulated directions). Did I get it right?
VituixCAD versions from 0.1 to 2.0.20.0 (2019-08-20) did not have frequency response interpolation. Program just selected the closest measured angle, and compensated measured magnitude and phase with mechanical geometry ~ difference between simulated flying distance and length of design axis. That approach is okay especially with circular radiators if angle step while off-axis measurement sequence is small enough e.g. 5 degrees to avoid large jumps in response shape when position of virtual mic travels in the simulation or location of driver in crossover is adjusted or design axis is moved or nominal listening distance is changed.
Sparse angle steps e.g. 30 degrees while off-axis measurement sequence cause quite big jumps in response shape, and result between measured directions is not necessarily very logical and accurate.
The most dramatic flaw of missing interpolation is behavior with rectangular radiators. Rectangular radiator is not possible to tilt vertically or rotate horizontally in the simulation without quite high error and stepped jumps in response shape if measurement data includes horizontal and vertical planes only. Of course we could measure rectangular radiator in it's final & fixed tilted/rotated position, but that would be as inflexible because does not offer rotation of simulated planes/orbits or change of design axis or listening distance etc.
Free simulation of rectangular radiators require either balloon data all around the speaker or response interpolation between measured planes/orbits. The last one is selected to VituixCAD.
The most inflexible approach would be measuring everything at predecided listening point. That is accurate for sure, but requires large anechoic. Possibilities to adjust construction in the simulation are gone also with adjustable constructions so designer cannot optimize construction at the same time with crossover.
I'm sorry for all the questions, I just want to turn it off and if I cant, then I want the knowledge to control the result. If I can't then I will have to use other methods.
Will measured data every 5 degrees mean there will be no interpolation? or will there be some, diagonal angles calculated for power?
Will measured data every 5 degrees mean there will be no interpolation? or will there be some, diagonal angles calculated for power?
Last edited:
Interpolation is automatically "off" if location of all drivers in the crossover is X,Y,Z=0,0,0 mm and T=0 deg and R=0 deg and angle step while off-axis measurements equals to angle step setting in Options window.
Power response is calculated normally in horizontal and vertical planes only. That changes along with axis rotation; Planes deg parameter in Drivers tab.
Power response is calculated normally in horizontal and vertical planes only. That changes along with axis rotation; Planes deg parameter in Drivers tab.
I'd like know actual reason for this.
Optional selection of the closest measured angle would be easy to add, but I'm not very eager to restore old flaws and add more switches and instructions if measurement data producing acceptable interpolation result is possible to arrange. Magnitude interpolation is not a problem, but phase could be if timing of responses to be interpolated are "from different worlds".
I'd be interested in a possibility to select angles.
I am working on a prismatic speaker where the sides are angled 18 degrees (up and down). I planned to measure in 3 or 6 degree intervals, and then enter the position and angle of each driver accordingly. I am not sure if the interpolation to 5 degrees makes much of a difference to the result or if it's just my engineer's OCD here.
Anyway, time to send another huge thanks to Kimmo for continuously well thought through and generous work.
/G
I am working on a prismatic speaker where the sides are angled 18 degrees (up and down). I planned to measure in 3 or 6 degree intervals, and then enter the position and angle of each driver accordingly. I am not sure if the interpolation to 5 degrees makes much of a difference to the result or if it's just my engineer's OCD here.
Anyway, time to send another huge thanks to Kimmo for continuously well thought through and generous work.
/G
Ok, this must be zero if measurements cover the delays and don't require VC to do the physical rearranging. No problem.
Two reasons I would change this. One is when I want to see phase better so I delay both drivers to control wrapping. Other is when I plan to delay in software.
Angle windowing. I'd like to set the size of a slice and the energy in each slice, but not have numbers bleed over into the next slice (or over many slices).
Why?
Is something wrong with interpolation calculation? Or is measurement data incompatible with the program? It yes, why it is incompatible and can you do something about it in the future?No need to answer if driver's XYZRT=00000 is adequate trick at the moment.
T deg parameter of driver in XO is vertical rotation (tilt) around it's center point. That does not change delay by distance from driver to mic.
Global timing to e.g. improve readability of phase responses is set with Delay [us] parameter in Drivers tab.
It looks as though I have misinterpreted. However I do see interpolations (soft transitions) when angles are added.
Yes, to get control of measurement weighting so I can use data that doesn't conform to a sphere, a cylinder or half-space. Since this isn't needed for most users, I can calculate power myself, and DI if I can't re-import it.
Hi kimmosto ! (And all of you )
Yesterday, I was simulate some crossovers with parametrics EQ and I've found that...
Let say, I've got tweeter's 180° measurements and I add an param EQ.
I can see how the EQ affect the SPL, but it seems to me that 'Line Charts' windows doesn't reflect my EQ yet in tweeter band...
Am I wrong ?
Have a nice day ...
)Yesterday, I was simulate some crossovers with parametrics EQ and I've found that...
Let say, I've got tweeter's 180° measurements and I add an param EQ.
I can see how the EQ affect the SPL, but it seems to me that 'Line Charts' windows doesn't reflect my EQ yet in tweeter band...
Am I wrong ?
Have a nice day ...
Never mind. 180 deg is still visible without Mirror missing angles if driver's X,Y,Z,R,T=0,0,0,0,0. But position requirement could be very strict. Y+1 mm or X+1 mm drops 180 deg response out with 0-180 deg measurement data without Mirror missing because 180 deg axis will be in unmeasured shadow.
Magnitude response interpolation works okay but phase response interpolation sucks especially to rear sector. Biggest challenge for simple interpolation with complex numbers is too big (>180 deg) phase difference between measured source responses. Interpolated phase could have temporary or more permanent error in polarity and group delay.
I have tested phase interpolation also as group delay and unwrapped phase interpolation, but they have own uncertainties related to sparse frequency axis.
To eliminate interpolation:
1) Measure with angle step VituixCAD supports at the moment; 5, 10, 15, 20 or 30 deg. Select measured angle step in Options window.
2) Set listening distance in Options window to maximum 99900 mm to avoid need of interpolation due to driver layout. Maximum distance keeps angle difference between measured and simulated <0.5 deg if drivers are located max. ~870 mm from speaker's origin.
To minimize possible negative side effects of phase interpolation:
1) Prefer simple designs with low XO frequencies. Create high or omni directivity with single radiator, not with driver arrays. Do not cross within top two octaves.
It's not probable that we could control significant majority of off-axis responses with some filter topology if design is complex multi-driver array so measuring with very small angle step does not provide solution.
If this has been brought up in the past, just direct me to that.
I am working on LR-4 filters and realize in need time alignment for the tweeter, passive ofc. Did some googling and found that one can indeed make such a circuit, they are called; Ladder delay network - I have some reading to do, but does VCAD have this function ? and of not, can it be added ?
I am working on LR-4 filters and realize in need time alignment for the tweeter, passive ofc. Did some googling and found that one can indeed make such a circuit, they are called; Ladder delay network - I have some reading to do, but does VCAD have this function ? and of not, can it be added ?
In the Crossover tab, open the library (LIB) and the dropdown offers many modules, the All-pass included.
I was actually going to write that ... LOL. Discovered that they are also called Lattice delay network which is the same as the All-pass. But thanks for the advice.
help needed with first measurements.
Hi,
after playing with the demos, I finally got around to perform some measurements. (scarlet solo, emm6, REW)
I choose a commercial 2 way I own since the 1980, it has a 8" woofer and is crossed at 3.1 khz. It beams a lot, I presume this shall show in the measurements, therefore it is a good practice candidate and confidense builder.
I performed measurement at driver heigth and 1M, and only did 0-10-20-30-40 angles.
I measured near field of dust cap and Port (without gating).
I also took 3 measurement on tweeter level for acoustic delay at 1 M
I took one measurement on tweeter level with cross over connected for comparison.
Now the questions:
1) in merger tool, do i have to compensate for the fact I used less than 2.83 volt or is it ok to leave number as they are even if lowish?
2) I simulated the crossover and the on axis spl trace does not match the measurement I made with the crossover, it has a dip of circa 3db from 1.2Khz to 6 khz, if I apply the delay it is even worse. Is it a phase issue that shall be dealt with in the merger?
Big Thanks to everyone who can help me get a better understanding of this amazing software!
Hi,
after playing with the demos, I finally got around to perform some measurements. (scarlet solo, emm6, REW)
I choose a commercial 2 way I own since the 1980, it has a 8" woofer and is crossed at 3.1 khz. It beams a lot, I presume this shall show in the measurements, therefore it is a good practice candidate and confidense builder.
I performed measurement at driver heigth and 1M, and only did 0-10-20-30-40 angles.
I measured near field of dust cap and Port (without gating).
I also took 3 measurement on tweeter level for acoustic delay at 1 M
I took one measurement on tweeter level with cross over connected for comparison.
Now the questions:
1) in merger tool, do i have to compensate for the fact I used less than 2.83 volt or is it ok to leave number as they are even if lowish?
2) I simulated the crossover and the on axis spl trace does not match the measurement I made with the crossover, it has a dip of circa 3db from 1.2Khz to 6 khz, if I apply the delay it is even worse. Is it a phase issue that shall be dealt with in the merger?
Big Thanks to everyone who can help me get a better understanding of this amazing software!
A basic question: i am simulating an enclosure for Seas W26FX001. I have measured the impedance and TS-parameters using DATS V3. In enclosure tool, at what Voltage can i see Cone excursion and port velocity exceeding the woofers limit? This woofer is rated at 150 Watt long term power handling. Do i use the power graph to see when i have raised the voltage to 150 Watt?