A question from a beginer
Hi Kimmosto,
Thank you for this usefull software.
Let me ask a probably naive question :
I'm currently using your software to check the behavior of different midbass drivers.
And while the parameters that I give to vituixCad are extracted from the scanSpeak or Seas datasheets,
I get some efficiencies value significantly higher with vituixCad from those given by datasheets.
For example with 18WU/8741T00
VituixCAD says 88.9 db/2.83V
ScanSpeak says 85,4 db/2.83V
How to explain these differences ?
Kind regards
Stéphane
Hi Kimmosto,
Thank you for this usefull software.
Let me ask a probably naive question :
I'm currently using your software to check the behavior of different midbass drivers.
And while the parameters that I give to vituixCad are extracted from the scanSpeak or Seas datasheets,
I get some efficiencies value significantly higher with vituixCad from those given by datasheets.
For example with 18WU/8741T00
VituixCAD says 88.9 db/2.83V
ScanSpeak says 85,4 db/2.83V
How to explain these differences ?
Kind regards
Stéphane
^Scan-Speak probably measures to half space and calculates average USPL visible in datasheet. VituixCAD calculates with basic function by R.H. Small:
18WU/8741T00:
n0=4*pi^2*fs^3*Vas/c^3/Qes=4*pi^2*35^3*0.0371/344^3/0.44=0.003506
SPL=SPL0+10*log(n0)=112.1+10*log(0.003506)=87.55 dB/1W/1m
USPL=SPL+10*log(8/Re)=85.55+10*log(8/5.92)=88.86 dB/2.83V/1m
Obviously T/S parameters on datasheet do not give accurate SPL and USPL. Should be asked from Scan-Speak why they give T/S parameters which have that much error compared to actual device. Whole idea of T/S parameters is to get decent simulation result with simple math at LF.
Motor design causes part of the difference: impedance rises already at mid-range so USPL calculation with Re does not work well. SPL is closer to reality because actual impedance at mid is not so far from nominal 8 Ohms.
An externally hosted image should be here but it was not working when we last tested it.
18WU/8741T00:
n0=4*pi^2*fs^3*Vas/c^3/Qes=4*pi^2*35^3*0.0371/344^3/0.44=0.003506
SPL=SPL0+10*log(n0)=112.1+10*log(0.003506)=87.55 dB/1W/1m
USPL=SPL+10*log(8/Re)=85.55+10*log(8/5.92)=88.86 dB/2.83V/1m
Obviously T/S parameters on datasheet do not give accurate SPL and USPL. Should be asked from Scan-Speak why they give T/S parameters which have that much error compared to actual device. Whole idea of T/S parameters is to get decent simulation result with simple math at LF.
Motor design causes part of the difference: impedance rises already at mid-range so USPL calculation with Re does not work well. SPL is closer to reality because actual impedance at mid is not so far from nominal 8 Ohms.
Last edited:
USPL=SPL+10*log(8/Re)
I kimmosto,
After thinking a little on the subject. I found out why the
USPL is different and probably for the same reason why the frequency response in Enclosure analysis is a too high for a given input voltage.
The formula that you use : USPL=SPL+10*log(8/Re) hasn't a lot of physical meaning while the 8 is a completely conventional value.
May be it could be usefull to compute USPL and frequency reponse, to use cone motion and some aproximation of its radiation impedance.
I kimmosto,
After thinking a little on the subject. I found out why the
USPL is different and probably for the same reason why the frequency response in Enclosure analysis is a too high for a given input voltage.
The formula that you use : USPL=SPL+10*log(8/Re) hasn't a lot of physical meaning while the 8 is a completely conventional value.
May be it could be usefull to compute USPL and frequency reponse, to use cone motion and some aproximation of its radiation impedance.
^8 is conventional because it represents 2.83V in the formula i.e. in definition of USPL.
Anyway, Scan-Speak's web site has massive toolbox including single Excel file: VENTED-FDD.0.96.US_EN.zip
If you enter equal parameters to both VCAD Enclosure tool and S-S's Excel, you will get about 0.2 dB difference (black=VCAD, gold=S-S):
Excel file searches maximum USPL=88.01 dB @ 104 Hz from frequency response. USPL would be 88.77 dB/2.83V/1m without effect of lossy inductance with extended impedance model.
I'm not so keen to make any fancier than they and most of the other tools have.
Anyway, Scan-Speak's web site has massive toolbox including single Excel file: VENTED-FDD.0.96.US_EN.zip
If you enter equal parameters to both VCAD Enclosure tool and S-S's Excel, you will get about 0.2 dB difference (black=VCAD, gold=S-S):
Excel file searches maximum USPL=88.01 dB @ 104 Hz from frequency response. USPL would be 88.77 dB/2.83V/1m without effect of lossy inductance with extended impedance model.
I'm not so keen to make any fancier than they and most of the other tools have.
See? This is why you shouldn't bother with measurements. Just use ruthenium plated, titanium dipped speaker cables and the sound will become warm, brilliant, and with superb PRaT, and you'll begin to hear details you never knew existed. Also make sure your speakers are facing the moon when listening to complex passages.
Good to see you here.
kimmosto,
an observation I'd like to share. I've noticed that the last version of the software is pretty slow in converging to the target curve (if converges at all) compare to LspCad at similar initial conditions. Is it a different solver scheme in your software? I am not 100% sure but I think some older version (like a year old version) was at least as good as LspCad in that regard.
an observation I'd like to share. I've noticed that the last version of the software is pretty slow in converging to the target curve (if converges at all) compare to LspCad at similar initial conditions. Is it a different solver scheme in your software? I am not 100% sure but I think some older version (like a year old version) was at least as good as LspCad in that regard.
^Solver has not changed much since the beginning. Parameter and impedance constrains are added but they should not affect so much. The fastest version 2.0 was probably rev. 2.0.3.1 (2018-06-10) having dense matrix circuit solver, narrower internal frequency range and no response interpolation. Now there are many actual improvements and new features reducing performance. For example each calculated response is weighted average of four measured responses and you can set angle step for power & DI to 5 deg (and full space in two planes) and internal frequency range is wide 5Hz-40kHz with response extrapolation. Minimimum phase, excess phase, listening window average, excess GD and so on are calculated without asking as well as impulse response and power dissipation if those windows are open.
The fastest settings in Options optimize about three times faster than the slowest settings, though the fastest method includes axial response only while optimizing acoustical domain.
Version 1.1 was clearly faster than 2.0 for sure because network was simple formal ladder calculated with impedances and Ohm's law and responses were not interpolated etc...
Calculation routine is not very optimized at the moment. I will try to eliminate at least repeating while single calculation cycle. That might double the speed I hope.
The fastest settings in Options optimize about three times faster than the slowest settings, though the fastest method includes axial response only while optimizing acoustical domain.
Version 1.1 was clearly faster than 2.0 for sure because network was simple formal ladder calculated with impedances and Ohm's law and responses were not interpolated etc...
Calculation routine is not very optimized at the moment. I will try to eliminate at least repeating while single calculation cycle. That might double the speed I hope.
thank you!
I see and understand how the current version is becoming a fundamental and powerful tool for the simulations. I think it is one of most, if not the most, powerful tools out there. Some time I just need a quick and easy solution to try something without involving all its capability and just doing quick single measurements. I guess V1.1 is good for that, with only aspect missing is being able to smooth the response in the driver tab. Why it is such a big problem for me? FuzzMeasure doesn't export using smoothing settings, so I guess it is more of their fault, but they are not on the forum haha
I see and understand how the current version is becoming a fundamental and powerful tool for the simulations. I think it is one of most, if not the most, powerful tools out there. Some time I just need a quick and easy solution to try something without involving all its capability and just doing quick single measurements. I guess V1.1 is good for that, with only aspect missing is being able to smooth the response in the driver tab. Why it is such a big problem for me? FuzzMeasure doesn't export using smoothing settings, so I guess it is more of their fault
, but they are not on the forum haha
Few tips to improve optimizing performance of ver.2 without compromising Power response result too much:
1) Select Angle step = 20 deg in Options window. This is the most significant parameter giving 3 times faster optimizing than 5 deg with full 0-180 deg measurement data. Accuracy of Power response is still okay.
2) Hide Normal phase, Minimum phase and Excess phase curves in SPL chart.
3) Hide also Listening window curve if not optimized as axial response.
4) Hide Normal and Excess group delay curves in GD & Phase chart.
5) Disable Directivity chart, though it's not refreshed while optimizing.
6) Uncheck Minimum impedance and Maximum gain constraints in Options window if not actually needed.
7) Select Passive component snap = No in Optimizer window.
8) Reduce size of main window to minimum.
9) Close Impulse response and Power dissipation windows and Enclosure tool.
Program could force most of those off while optimizing but so far I have trusted that user eliminates extra features and windows if more speed is needed.
1) Select Angle step = 20 deg in Options window. This is the most significant parameter giving 3 times faster optimizing than 5 deg with full 0-180 deg measurement data. Accuracy of Power response is still okay.
2) Hide Normal phase, Minimum phase and Excess phase curves in SPL chart.
3) Hide also Listening window curve if not optimized as axial response.
4) Hide Normal and Excess group delay curves in GD & Phase chart.
5) Disable Directivity chart, though it's not refreshed while optimizing.
6) Uncheck Minimum impedance and Maximum gain constraints in Options window if not actually needed.
7) Select Passive component snap = No in Optimizer window.
8) Reduce size of main window to minimum.
9) Close Impulse response and Power dissipation windows and Enclosure tool.
Program could force most of those off while optimizing but so far I have trusted that user eliminates extra features and windows if more speed is needed.