Open baffle is another special case in Diffraction tool. It does not cause any diffraction effects because program assumes that edge is thin and sides of the radiating surface are identical. Rear diffraction fully compensates front diffraction and vice versa. So model contains just directivity of radiating surfaces and (shortest) delay of inverted signal from the other side. Signal from the other side is 90 deg off-axis.
Oh yeah, spinorama is all we need to know - no matter what ever happens above 85 dBSPL, what kind of acoustics, how speakers are positioned, how far listened, 100% THD or 100% IMD is also okay, terrible resonances <100 and >16k are just fine, shotgun soundstage created by terrible diffraction is perfect, bass may arrive half a day after HF etc. Long live ASR!yes, Of course the power & DI chart is all knowing
I think both ARTA and REW could have both options. One may have good enough sound card for cross-correlation of full dual channel but someone else may not. Response compensation is nice feature for example for tweeter protection, but not the only possibility. Also generated measurement signal can have DUT protection with shelving HP or 1st order HP. Known pre-filtering is very easy for response calculation.
Member
Joined 2003
What I meant was in regards to linear performance, I find the spinorama to be the best tool for evaluating overall response over interrogating any single axis of information alone, such as in the diffraction tool.Oh yeah, spinorama is all we need to know - no matter what ever happens above 85 dBSPL, what kind of acoustics, how speakers are positioned, how far listened, 100% THD or 100% IMD is also okay, terrible resonances <100 and >16k are just fine, shotgun soundstage created by terrible diffraction is perfect, bass may arrive half a day after HF etc. Long live ASR!
In ASRs defence, there is other information such as distortion at 86 and 96dB, CSD and individual driver responses provided for speaker evaluations. But I am not convinced in the subjective testing methods, only one speaker is ever used for example, so with ASR, and most any review for that matter, I come for the objective data, and skip over the subjective opinions. The problem always is with the more layman of reader who doesn't know how to interpret the objective data, so is relying on subjective impression for the interpretation, which can lead to some incorrect conclusions.
Agreed there, I found it quite odd that this feature didn't exist when switching from SoundEasy to ARTA. In SoundEasy, it was labelled as "pre-emphasis" for MLS measurements, but it is exactly that, a pre-filter. Software filter doesn't fully protect if you have issues like amplifier thump or live connection of RCA that creates a buzz, but for most operations it's good enough, and a much better transfer function in digital domain than what happens when a simple capacitor is placed in series with a tweeter.
The latest build of 2.0.83.1 (2022-02-04) uploaded few minutes ago uses frequency range limits when calculating response for G(f) block. Magnitude and phase of G(f) response are frozen outside frequency range limits specified by user. This is quite cheap trick because result does not necessarily maintain delay features.When I tried this, I am unable to limit the filter to a certain frequency range, despite setting it in the Optimizer window.
What end up is that it will try to hold a flat FR until 0Hz (or it seems so as step response never came down). That drove the woofer crazy.
Is there a way to have the FR limit to a certain range? It can do unity gain outside the range so it won't affect the response.
For example woofer equalized to target within 20...8500 Hz:
Tweeter equalized within 400...18000 Hz:
Bad measurement data outside frequency range limits does not affect to G(f) anymore. There's still possibility that taps is not adequate, but improvement may need quite much smarter logic and calculations or some user decision or both.
This works also with Axial response and Listening window options for whole speaker.
I understand that, but my criticism is focused especially to ranking products with Preference rating. Close to irresponsible imo. Timing plots and compression tests with practical signal spectrum should be there. IMD due to possible internal leak or vibration, diffraction (multiple sources with short delay) and better visualization than ERDI and SPDI about capability to produce acoustical resolution (how smooth and how much) are deeper, but may reveal something interesting standard measurements don't show.
Anyway, I'm banned again and no plan to return anymore.
Member
Joined 2003
This is a welcome addition, thanks! If this is a second build of 2.0.83.1 (originally 2022-02-02), I don't think the software flags the automatic update unless you change the version number.
Ah, yes we understand each other. I agree, it is irresponsible to determine a "preference rating" that considers linear distortion only. Differences between preference rating would have to assume no other factors change that aren't considered in the rating, such as THD and energy storage problems, which will never be the case when reviewing difference speakers from various manufacturers. I personally never look at or even consider that rating when reviewing the data available there, I have all the information I need in the other charts and graphs to form my own rating.
Unfortunate that they ban you there, given some of the poor behaviour that is allowed. I'm not active over there, but I'm sure that you were only encouraging intelligent discussion/debate as you do here and elsewhere.
Last edited:
Considers magnitude part of linear distortion. At certain constant (low) level only. That is radical simplification allowing very lame sound - and spinorama-believers will never know and hear the problem.
^ Very interesting! Any chance you could expand on the problem? I think you know exactly what is the difference between good and lame sound, while both would have very similar magnitude graphs, but I'm afraid this might be something you can't disclose? The missing piece
I'm interested because experimenting with mono prototype I've noticed sound can be boring or exciting even though the magnitude graphs in VituixCAD simulation look very good and very similar and the available ratings are as good as I've got. While my DSP settings or measurements or anything might contain error so the real systems are not as close each other as the simulations suggest there is difference in group delay and step response for example.
Next step is try and reduce possibility of configuration error and then try and figure out what is it that makes sound boring (lame?), if it wasn't an error. Still, not knowing whats the target, what to look for, is unknown to me, many experiments ahead I guess
I'm interested because experimenting with mono prototype I've noticed sound can be boring or exciting even though the magnitude graphs in VituixCAD simulation look very good and very similar and the available ratings are as good as I've got. While my DSP settings or measurements or anything might contain error so the real systems are not as close each other as the simulations suggest there is difference in group delay and step response for example.
Next step is try and reduce possibility of configuration error and then try and figure out what is it that makes sound boring (lame?), if it wasn't an error. Still, not knowing whats the target, what to look for, is unknown to me, many experiments ahead I guess
Last edited:
Energy dispersion is not a secret and effect shouldn't be difficult to understand. Sound wave transient can be like a hammer blow, and there will be reduction in audible and sensible effect if total energy is distributed for example from 1 ms to 5 ms with crossover. Normalization to max 1 or 0 dB is typical problem in visualization of pressure(time) and energy(time) data so comparing momentary pressure/power/energy loss is not usually very direct and clear visually. Evaluation of strength/weakness of transient by listening requires suitable music and possibility to manipulate phase response without any other change. Low SPL is as revealing as high because compression may weaken 'hammer blow'.Any chance you could expand on the problem? I think you know exactly what is the difference between good and lame sound, while both would have very similar magnitude graphs, but I'm afraid this might be something you can't disclose? The missing piece
Ideal vs. excess group delay seen in some speakers. The most lame and thin sounding industry has greated.
Here is an example of my diy 4-way speaker with all-LR4 vs. all-LR2 crossovers (response and delays optimized) Measured indoors at roughly 1m, so some reflections visible. The LR2 version sounds better with transients, eg. piano recordings. Later on I've made two 3-way speakers and ended up usig LR2 xo with them too, despite of slightly worse spl-response and distortion than with LR4.
I have understood that kimmosto doesn't like the sound of Genelecs very much, perhaps because of very steep xo causing softening of ETC?
I have understood that kimmosto doesn't like the sound of Genelecs very much, perhaps because of very steep xo causing softening of ETC?
I'm trying to avoid naming, but at least the latest Ones and probably other SAM models too have FIR stage for phase linearization at MF...HF. Improvement is clearly audible in my opinion though cannot compare properly with FIR stage bypassed. Latency requirement limits possibilities with FIR - especially if the lowest XO must remain steep due to features of the radiators. Shallower slopes would help with taps, but that is just different compromise and not necessarily acceptable.
Member
Joined 2003
Just looking at this again. I think the flat line transfer function makes sense for avoiding compensation of the speaker natural low frequency response, left highlight of the first image. However, for the right highlight of the first image, or the left highlight of the second image, I would want a filter response that follows the slope of the filter there, so the filter continues for n beyond the measurement noise floor for example, the sharp corner in the filter response is a bit disconcerting. This is why my previous suggestion was for use of HBT for this task, but it sounds like HBT creates some phase errors. Maybe I’m just nitpicking once the response is say 40dB down, just something I thought I’d mention if there’s a practical solution for it.
i tried searching this thread prior to posting, but I have a question about the actual crossover frequency. in VituixCAD, there are 2 (of the 6) graphs in question... the SPL and the Filter. When modeling a 2-way system, which of these two, where the drivers cross, is the actual Fc? neither matches each other, so i want to make sure i'm understanding this correctly. thanks!
Short repeat. HBT would not change much or anything. For example if you manipulate driver's measurement below equalized range with HBT, phase response will be minimum phase there. But if your target slope is linear phase, there would be sharp and high phase jump in filter TF without extending equalized range down to two decades below driver's natural low limit, and always if HP slope is not 4th or 8th order having 0 degrees phase shift at 0 Hz.
Earlier I wrote that HBT cannot be used at equalized range because result phase response will not be at target (HBT replaces measured phase). So that method is out of question at least with linear phase targets.
This trick is quite close to perfect for magnitude response because user can and should monitor acoustic responses and ensure that gain is low enough at LF with HP filters to avoid unnecessary excursion. Weakest point of freezing TF outside equalized range is phase response which starts to follow measurement with some offset. Tweeter's target may include long delay so it's best to extend equalized range up to Nyquist with HP filter. Not stop at 18 kHz such as my example (which was fortunately just an example).
Member
Joined 2003
Well the top left is the acoustic response, and the right middle is the electrical response. Both are "correct" in some context of crossover frequency depending on what you are talking about, but for most cases when speaking of the crossover frequency of a speaker, we are speaking of the acoustic result.
Member
Joined 2003
Yes, I understand the phase problem with HBT in this case, I was focusing on the horizontal 0db/oct slope that is applied with the TF frequency limit, with the suggestion that there may be some benefit to apply a slope that follows the filter function in certain situations rather than flat line at 0dB/oct.