Hi Kimmo
What do you think is useless about ER and ERDI in CTA 2034A?
We have had a discussion about this in Audio Science Review and I expressed some doubts.
One has been resolved but others are still subject to discussion, I am interested to learn your view.
Best wishes
David
I think he was saying that vertical and rear measurements weren't weighed strongly enough in the original interpretation with just 3 rear curves and no weighting. I remember him saying something to that effect earlier in this very long thread
My opinions were and still are reflecting features of my designs including e.g. boxed coaxials, conventional 2...4-multi-ways and dipoles with and without coaxial.
Earlier calculation (without weighting with RMS grouping and full rear sector) ER and ERDI were hiding obvious problems in directivity. Those problems were clearly visible in power and DI responses. So possible consequence of ignoring power and DI, and designing by ER, ERDI, In-room and axial/listening window only was worse sound.
At the moment with full rear sector data in horizontal plane and equal weighting of front, side, rear, floor and ceiling bounces looks close to as revealing and useful as power & DI responses, but no more. Quick and hasty conclusion is that value of ER and ERDI is still low if adequate data for power & DI responses is available.
One possibility to increase weight of sides, verticals and rear is to remove front -30...+30 deg from ER and ERDI. That's already included in listening window so designer is aware of it. Maybe we should not emphasis problems by skipping some data if that action reduces correlation with listening impression.
Also travel distance of reflecting sound has effect to weight in practice, but that's not directly taken into account in ER and ERDI calculation. Actual weight of rear and front bounces are probably lower that floor, ceiling and sides in "typical room".
Last edited:
I remember him saying something to that effect earlier in this very long thread
Yes, it was few months ago and based on wrong interpretation and/or typos in standard.
Last edited:
***Thanks!
As far as I know, if I do not have off-axis measurements, then to build a polar map it is necessary to indicate the diameters of the cones of the drivers. Right? For example, as in LSPCad. Do I need to do this in the VC?
You can generate off-axis responses for measured full range axial response with Diffraction and Calculator tools, or use Enclosure and Diffraction tools to create simulated LF part for boxed or dipole speaker and Merger to merge with HF part with simulated off-axis if measured single far field response is time-windowed i.e. frequency range is limited at LF.
BUT accuracy with simulated ideal flat pistons is not very good and therefore I have left this procedure as little attractive as possible; complex and time consuming. Easiest, fastest and most accurate is to follow recommended design procedure. If that is out of question, measure design axis responses only and simulate XO without directivity.
***Thanks! I understood you
Tell me please! Can I see the calculated step response and square wave of the finished system.
Thanks!
Help documents probably tell something about step response
Search "step curve". Square wave is not shown anywhere because nothing new compared to step.Help documents probably tell something about step response
***Ok! Thanks!
The latest twist with dual added mass method and LOG model
Rev. 2.0.46.1 (2020-03-21)
Enclosure
* Creep function of LOG model normalized with fs. Parameters R0 and C0 removed from driver database, Edit/Add driver and T/S calculation windows.
Hopefully it's more stable and finds accurate solution with different drivers. Not guaranteed because feature is tested with single driver only. Feedback to e-mail with measured impedance responses (3/driver) is welcome.
Rev. 2.0.46.1 (2020-03-21)
Enclosure
* Creep function of LOG model normalized with fs. Parameters R0 and C0 removed from driver database, Edit/Add driver and T/S calculation windows.
Hopefully it's more stable and finds accurate solution with different drivers. Not guaranteed because feature is tested with single driver only. Feedback to e-mail with measured impedance responses (3/driver) is welcome.
Last edited:
New build of rev. 2.0.46.1 available.
Qms and Vas calculated without frequency-dependent compliance with Dual added mass & LOG model to reduce difference to traditional T/S values. This is just workaround to avoid separate R0 and C0 fields in driver database.
Qms and Vas calculated without frequency-dependent compliance with Dual added mass & LOG model to reduce difference to traditional T/S values. This is just workaround to avoid separate R0 and C0 fields in driver database.
^Oops, that was fixed 3 days ago to Paste command in Add/Edit driver window, and rev. 2.0.46.2 also to Update database... with xml file. Bug since 2.0.38.0 (2020-02-28) due to squared and '/' characters which are illegal in xml element name. Thanks for info!
Rev. 2.0.46.3 (2020-03-24)
Main
* SPL and delay of acoustical simulation normalized by distance from speaker's origin to virtual microphone no matter microphone offset X,Y mm in Drivers tab. Earlier normalization was by Listening distance set in Options window causing excess delay and reduced readability of phase responses if microphone was moved.
Enclosure
* Added paste from speakerbench.com in json format to Add driver window.
Main
* SPL and delay of acoustical simulation normalized by distance from speaker's origin to virtual microphone no matter microphone offset X,Y mm in Drivers tab. Earlier normalization was by Listening distance set in Options window causing excess delay and reduced readability of phase responses if microphone was moved.
Enclosure
* Added paste from speakerbench.com in json format to Add driver window.
Rev. 2.0.46.4 (2020-03-29)
* Impedance response files with less than 48 ppo read with cubic spline interpolation to reduce error especially with T/S parameter extraction.
Calculate T/S parameters
* Solving weights and tolerances adjusted to improve fitting of extended impedance model with Le, Leb, Ke and Rss.
* Checking of known Mms or BL enables Calculate button without loading second impedance response.
This is second build so check that date in about box is 2020-03-29.
* Impedance response files with less than 48 ppo read with cubic spline interpolation to reduce error especially with T/S parameter extraction.
Calculate T/S parameters
* Solving weights and tolerances adjusted to improve fitting of extended impedance model with Le, Leb, Ke and Rss.
* Checking of known Mms or BL enables Calculate button without loading second impedance response.
This is second build so check that date in about box is 2020-03-29.