How to take reliable measurements from 100Hz to 500Hz to cross over a 3 way

[Efalegalo]

Draki,

I'm not able to save measurements in the Demo version, but I've had no trouble exporting measurements.

[Efalegalo]

More VituixCAD questions

Kimmosto,

Sincere thanks for responding. Your responses were immensely helpful. I've read through the "preparing measurements" pdf but its going to take my non-technical mind a while to sink in :-).

And sorry for the typo. You are correct, I do NOT have a turntable setup just yet. I plan on building one with a lazy susan in the comming weeks.

This is how far I've gotten:

I've attached my ARTA settings below. I tried taking measurements at 48000 sampling rate but I get "pops" in the sound. I don't seem to have an issue at 96000.

"Dual-Channel Measurement Mode" is checked (actual connection is "semi-dual"). I'm not sure whether I'm also suppose to check "Filter dual channel impulse response"??

First and foremost, I'd like to mention that I'm a complete newbie to speaker design. I've been messing around with building my first TM bookshelf speaker. I've had decent success with WinPCD and Xsim, but I understand Vituix is the most feature-full of the three software.

In Xsim and WinPCD - all my measurements were taken on the design axis, minimum phase was extracted, z-offset was determined, location of driver was specified. When modeling crossover, the predicted response matched the measured response on the design axis to the T (< 0.5db/variation).

I'm hopping to achieve similar level of success with VituixCAD.

Today, I practiced the following (only ARTA-Demo + VituixCAD was utilized):

1) Measure tweeter on tweeter axis - 60cm from baffle - Exported gated measurements with measured phase into destination folder
2) Lower mic height and back it up by 24mm because the woofer is mounted on a 24mm thick stepped baffle (making sure the mic to stepped baffle is also 60cm). Before exporting gated response, make sure "marker" cursor was set to the exact point (467 in my case) as the one selected for tweeter's gating. Response was exported with measured phase.
3) Lower ARTA output volume by -20db and take near-field measurements of the woofer (<5mm from dust cap) and port (0mm from rear baffle). Selecting 1/12th smoothing, export measurements into destination folder - retaining measured phase.
4a) Launch Vituix - Merger Tool. In low-frequency part, load both the woofer and port near-field measurements. Specify 28mm for port diameter and 136sqcm for woofer radiating diameter
4b) Select "Diffraction Response" and launch diffraction tool. Specify woofers location on baffle. Center mic icon to be 100% in center of woofer location. Export diffraction simulation in destination folder and load into Merger Tool
4c) In high-frequency part, load woofer's on-axis fair-field response. Check "Axial". Adjust "scale" and "transition frequency" as appropriate.
4d) Under output section, select "Minimum Phase" to see how it matches measured phase. In my case, it was substantially off, so I did NOT check minimum phase. Check "Feed Speaker" and click "Save"
5) Specify woofers location on the baffle. In my case, X,Y,Z coordinates were 0,-167,-24, respectively. Which I am interpreting to mean: in-line with tweeter, 167mm below tweeter, and 24mm in front of the tweeter (due to stepped baffle mounting)
6) Specify tweeter location on baffle. In my case, 0,0,0 for XYZ
7) Under tweeter selection, enable view of "minimum phase" under SPL graph. Adjust tweet delay so measured phase response nearly matches minimum phase response.
8) Go back to woofer selection and enter the same value under as that determined for the tweeter in Step 7 above


Now, some questions:

1) Assuming a hypothetical world where one was only listening to the speaker on the tweeter axis (0 vertical off-axis, and 0 horizontal off-axis), do the steps above appear correct
2) After I spliced the diffraction adjusted near-field woofer response to the far-field woofer response (in the Merger tool), I found my measured phase (?) was significantly off from minimum phase. How do I know whether to select "minimum phase", "minimum phase with GD of HF" or leave both unchecked and move forward with measured phase. Which one is the right option? In my case, utilizing the measured phase (i.e, not checking "minimum phase") resulted in the better match between Vituix predicted response and the speakers actual measured response.

3) Somewhat generic question: does Vituix require specifying "tails" to derive minimum phase? Looks like the software does it automatically. But in other software, I specifically had to specify the low frequency and high frequency roll-off for of the drivers for the software to able to derive minimum phase.

[kimmosto]

I'm not able to save measurements in the Demo version, but I've had no trouble exporting measurements.

This will change when you start using turning table and spatial IR group measurement and export, because you have to save pir files in order to export response files.

[ergo]

I would consider ARTA one of the best value for money softwares for speaker and other audio measurements. It's such a versatile package that one can keep discovering new features for a while. Also this far all of the version upgrades have been free for the license owners. So it's well worth to consider the purchase.

[kimmosto]

I'm not sure whether I'm also suppose to check "Filter dual channel impulse response"?

You can check that, but not must.

I understand Vituix is the most feature-full of the three software

Despite of some limitations with passive crossover topologies, VituixCAD is fast and adequate due to power response & directivity control (which are essential in speaker design).

I'm hopping to achieve similar level of success with VituixCAD.

I'm quite confident that you could set targets much higher with VituixCAD, though software doesn't make final decisions and high quality measurements are needed. It gives important information and designing is fast, but human intelligence is still needed to make excellent speakers.

1) Assuming a hypothetical world where one was only listening to the speaker on the tweeter axis (0 vertical off-axis, and 0 horizontal off-axis), do the steps above appear correct

Steps look okay, but this question may be not. You should measure polar responses and simulate and optimize directivity and power response (and DI) in order to design healthy sounding speaker also for one listener on axis. You are hearing much more than axial response, though reflections won't effect so much if room is acoustically very damped or listening distance is short.

2) After I spliced the diffraction adjusted near-field woofer response to the far-field woofer response (in the Merger tool), I found my measured phase (?) was significantly off from minimum phase. How do I know whether to select "minimum phase", "minimum phase with GD of HF" or leave both unchecked and move forward with measured phase.

You could check both minimum phase options if phase difference at splicing frequency is more than about 30 degrees. Do not mind high and low frequencies. Minimum phase extraction should be avoided also with ARTA by using longer time window. That is possible by increasing distance to nearest boundaries, adding damping material on the floor and ceiling or decreasing measurement distance. The other possibility is to increase splicing frequency closer to maximum. Adjust splicing point with cursor and monitor how meas. to min. difference behaves. Select splicing frequency which gives good magnitude and phase merging.

Which one is the right option?

You should prefer measured phase because it can save possible non-minimum phase features of the driver.

3) Somewhat generic question: does Vituix require specifying "tails" to derive minimum phase? Looks like the software does it automatically. But in other software, I specifically had to specify the low frequency and high frequency roll-off for of the drivers for the software to able to derive minimum phase.

Slope estimation is automatic in VituixCAD. Usually automation is adequate and allows very fast processing of hundreds of measurements at once. Minimum phase extraction is secondary/back-up function if something is not done according my recommendations. Manual trimming and slope setting would be valuable in some (rare) cases, but I've avoided more complex user interface.
P.S. Some internal functions have to use minimum phase extraction. For example excess group delay and phase response calculation of circular piston or rectangular radiator to off-axis.

[Efalegalo]

Steps look okay, but this question may be not. You should measure polar responses and simulate and optimize directivity and power response (and DI) in order to design healthy sounding speaker also for one listener on axis. You are hearing much more than axial response, though reflections won't effect so much if room is acoustically very damped or listening distance is short.

Kimmosto,

Thanks once again for taking the time to respond to my (very basic/elementary) questions.

As you suggested, I'll make it my A1 priority to build a turntable within the next couple weeks.

D

[Efalegalo]

turntable

Turntable under way. Its far from finished. I still need to build the stand that would perfectly align the speaker flush on the rotational axis.

The base has marking in 10 degree increments. I've installed heavy duty ball bearings that I had on hand. I first attempted with wheel casters, but they do not swivel around as nicely as the ball bearings. I was concerned that any wobbling might cause the speaker to fall off its stand, especially if the stand is tall (with the monitor placed 1/2 way been ceiling and floor). The bearings make rotational movement incredibly smooth.

I had thought about using a lazy susan, buy given that (i) the load will always be on one side, and (ii) eventually I would like to build a large floor standing speaker (with my preferred building material of 1" baltic birch ply), heavy duty ball bearings seemed more appropriate.

At any given rotation, the platform is supported by at least 4 bearings, each capable of handling 75lbs.

All in all, the platform is cable of supporting a speaker that is 16" wide, 18" deep, and weighing up to 300lbs.

(please excuse the mess in the background - just had basement waterproofing done).

[Juhazi]

My turntable consists of two 80x80cm pieces of coated furniture board, connected with a wooden pin at center. About six synthetic felt pads in-between and upper plate has a rounded edge for 180 degrees, markings between 5¤. I use a kitchen ladder as a stand when needed. Sorry, no picture.

[Efalegalo]

My turntable consists of two 80x80cm pieces of coated furniture board, connected with a wooden pin at center. About six synthetic felt pads in-between

Ha - that would have been a lot simpler :-)

[Efalegalo]

Simulated vs Measured

Progress update: built a turntable, and have spent additional time learning about VituixCAD - what awesome piece of software!

I do have one question/request: Can anyone explain why my simulated/predicted response is about 1db less than measured response?

See attached pics - Blue = Measured, Red = Simulated

Software utilized = VituixCAD, and Arta (Demo Version)

I did NOT touch the volume control during pre-crossover and post-crossover measurements.

I measured each individual crossover components (e.g., caps, coils, resistors) with Dayton DATs and measured values were entered into Vituix

Please disregard the magnitude of the db scale (nothing here was tested at 120db :-). That being said, the delta between measured vs simulated is still valid.

Thoughts?

[kimmosto]

^That is XSim - not VituixCAD. Anyway, one possible reason could be impedance response of drivers, which depends on source voltage and impedance while measurement. With XSim it is easy to forget wrong value to internal resistance of coils and capacitors. Default initial value for caps is 50 mOhms.

[Efalegalo]

^That is XSim - not VituixCAD.

Correct. The above application is Xsim; however, it was only used to display the frd files.

I used the “export” function in VituixCAD to generate the simulated repose displayed above.

Again - Xsim was just used for display purposes.

Not sure is there is a feature in Vitiux to import response.


Sent from my iPhone using Tapatalk

[kimmosto]

^It is called Open response because file format conversion is not needed. VituixCAD has at least 11 places where you can load frequency responses. The most suitable for comparing purposes are Reference angle and Power response overlays of Power & DI graph in the main program, and Calculator tool (Scale, Delay, Invert A function).