Thanks for clarification
I was considering to buy SE instead of ARTA but sadly they do not offer an eval-version to check their software first.
What exactly do you mean with point 1.)
Michael
I was considering to buy SE instead of ARTA but sadly they do not offer an eval-version to check their software first.
What exactly do you mean with point 1.)
Michael
I'm not familiar with ARTA, but if the two software packages I've used extensively, SpeakerWorkshop and SoundEasy, both use a reference impulse, which is from the signal driving the DUT. Since this is never a perfect signal, it's always necessary to subtract this effect from the response of the DUT to obtain a more accurate response of the DUT. I selected SoundEasy based on the fact that it had the most complete speeaker design features for a very affordable price, improvements are made at least once a year, the designer really spends time to consider recommendations from users. It's amazing to see the designer put in continued effort for so many years. The full manuals are on line for download if you wish to lear about it.
Allow me to question your explanation, Earl.
We most certainly would see more improvement with regard to that non Gaussian beam behaviour of the OS contour in the region around 4kHz with increased mouth round over as showing up actually in my simus
http://www.diyaudio.com/forums/showthread.php?p=1924924
This "non Gaussian beam behaviour" effect does not enter the picture with LeCleach horns - and also not with the pure round over "horns" I'v e shown:
http://www.diyaudio.com/forums/showthread.php?p=1927587
http://www.diyaudio.com/forums/showthread.php?p=1927588
http://www.diyaudio.com/forums/showthread.php?p=1927590
(at least not in the same magnitude and clarity)
Michael
Last edited:
Ahh - OK !
No - I never do CSD with reference impulse - not even know if ARTA could possibly do (think so). I'm fine with what I got until - I guess you should not ask for tooo high of a resolution - possibly you run into artefacts or consistency problems easily - but haven't explored that any further...
Might be a differnt task if you use valve amps for measurement purpous...
Michael
Last edited:
I actually learned a lot because of the high resolution.
But the reference impulse does not effect interpretation of CSD so much. Items 2&3 have more impact.
But the reference impulse does not effect interpretation of CSD so much. Items 2&3 have more impact.
Last edited:
Hi
Cone width is 37mm
Surround included is 44mm
Funny numbers 🙄
Cone depth, I would say about 5mm
Im not sure about sensitivity
Seems like it needs a big series resistor not to sound harsh
With a resistor it sounded smooth and airy
But too early for that
Hope to do more experiments soon
Attachments
Last edited:
If looks very much like the W2-800SL. My instinct is that probably zobel circuits are what is necessary to make them sound smooth, but the breakup modes are also a concern.
There is not enough data posted to make an objective comparison and he can't have ever made a subjective one either.
It begins here:
http://www.diyaudio.com/forums/showthread.php?p=1516725
and picks up again here:
http://www.diyaudio.com/forums/showthread.php?p=1518047
Earl provided impulse data from earlier measurements in there somewhere, as well....
Stick with ARTA for measurements. I did. Get LspCAD if you need crossover design software. SE does a bit of everything but I could never get past the user interface. The 'reference impulse' is just the 2-channel measurement that we all know and love. 🙂
For a superior CSD-type display, try ARTA's Burst Decay. It plots the CSD in periods rather than seconds and that seems more useful. Mathematically it's doing Linkwitz's shaped toneburst test except it's doing many frequencies at once instead of one at a time.
Of course all of these are just different ways of displaying the same old information that's present in the impulse response or the frequency response so we don't want to give them too much extra importance.
Dennis SE does Burst Decay waterfalls too. Looks similar to the CSD's. I actually wanted to use that instead because it is referenced to periods which I find less misleading at lower frequencies for most people. I must be missing something though because I could never get plots that looked right. So I used the CSD.
That said I like ARTA's measurement suite a bit better with the Farina method, etc. Just haven't got around to using it much to see if there is any advantage in my situation to using it. Now they just need to add some sort of matched filter for recovering low frequencies and I'd probably use all the time.
Speaking of CSD's I'll look for the Geddes ones. What I can remember is that it was as clean as anything else I tested.
That said I like ARTA's measurement suite a bit better with the Farina method, etc. Just haven't got around to using it much to see if there is any advantage in my situation to using it. Now they just need to add some sort of matched filter for recovering low frequencies and I'd probably use all the time.
Speaking of CSD's I'll look for the Geddes ones. What I can remember is that it was as clean as anything else I tested.
Last edited:
Be careful, I've never posted CSD that I have actually done myself so you have to always question the source. Believe it or not there are people out there would would like to discredit me!!😱
Yeah, I didn't mean to badmouth SE. I didn't really give it a fair chance and the stuff JohnK does with it is amazing so we know it's very powerful.
SE is a bit of an integrated package. Generally, the user interface follows a design sequence from the left-side menu to the right. The details of the user interface needs some time to getting used to, and there are a few things that could be improved. I'm sure there are other heavy weight's out there, and there are individual packages specifically for certain functions that may perform better in their specialized areas. But if you are looking at the total system design package, it's probably the best value on the "affordable" market.
The point Micheal is that if you irradiate an aperture with a given radius with a spherical wave, if the square root of the sum of the squares of these dimensions exceeds the aperture radius plus the wavelength of the source then a central minimumcan develope, this is called Fresnel diffraction.
In cases where this does not happen either the wave illuminating the aperture does not sufficiently resemble a spherical wave, or the above stated condition is not met, or some combination of both.
In the Le Cleche horn the first condition is not met and there is no central minimum, but it has a narrow beam that narrows with frequency.
I have already posted a simulation of a scheme that goes some way towards giving nearly constant directivity with a spherical wavefront out to 20kHz,(the "Oskugel"), but this needs a 20mm. diameter piston to drive it, (I originally modelled an annular diaphram compression driver with a 20mm. throat to drive it), and not a standard compression driver.
rcw.
In cases where this does not happen either the wave illuminating the aperture does not sufficiently resemble a spherical wave, or the above stated condition is not met, or some combination of both.
In the Le Cleche horn the first condition is not met and there is no central minimum, but it has a narrow beam that narrows with frequency.
I have already posted a simulation of a scheme that goes some way towards giving nearly constant directivity with a spherical wavefront out to 20kHz,(the "Oskugel"), but this needs a 20mm. diameter piston to drive it, (I originally modelled an annular diaphram compression driver with a 20mm. throat to drive it), and not a standard compression driver.
rcw.
I see...
I never cared about sub 1ms decay – I mean what resolution down there could we expect with usual 44.1 or 48kHz sampling anyway. I'm surprised to hear that you discovered interesting things down there.
Maybe you can share some examples?
I was (and sill am to some degree) a big fan of CSD plots as they provide a intuitively understanding of decay effects – basically what we see in the tail of an impulse response - but with the advantage to be decoded in frequency domain and also better displayed by the log scaling.
My priorities changed dramatically at the point when I understood what John Kreskovsky *really* told us with his min phase behaviour concept of speakers.
His claim is that (single driver) loudspeakers - meaning the driver plus the cabinet – or the baffle – or the horn, in practical terms *always* can be considered to be of min phase.
This results in the conclusion that we – theoretically – are able to perfectly equalise the complete device towards perfect impulse response / frequency response / CSD.
With nowadays available DSP and PC power such advanced equalising is a breeze to do.
The – practically - most sincere limitation of above concept is that this is precisely valid for one point in space only.
To stretch the benefits of this approach over an as wide as possible room angle, it is a *must* to keep directivity under control in such a way that it does change as less as possible over frequency band of interest.
Which specific kind of directivity one likes to preserve (omni, cardioid, dipole, whatever) is of minor interest here - key is to keep it stable over frequency.
Looking into the polar response of direct radiators we immediately become aware that with direct radiators this is not going to be happen (at least not in the upper frequency department).
Simply put - diffraction effects simply do not allow for defects free sound fields.
Following that path we come to the conclusion that we need kind of diffraction alignment to finally avoid time of arrival consistency distortion.
Bottom line – for "old school" XO cooking CSD measurements of speakers may be interesting – for advanced equalised speakers the only thing in CSD that is left to be displayed (for the single speaker) is the decay of the acoustic XO filters chosen.
Well sure, and of course also the impacts of deviations due to non perfect min phase behaviour over room angle if we perform CSD at several off axis positions...
Michael
I think we would be getting too much off topic. So I will try to explain things as close within topic as I can.
I don't thing anything related with acoustics can be called Minimum Phase, but as with all engineering problems, we must have some assumptions that will give us predicatable results of a reasonable degree. Therefore, if we assume MP, we find that results of engineering calculations are close enough for the basic purpose. If speakers were truly minimum phase, we would get exactly the same response regardless of distance. In my measurements, this does not happen, which brings me to study horn/guides with hope that the wave front is more pure, and hopfully has a certain pattern such that there is a large region where listeners can have the same experience of sound quality.
The CSD is a bit difficult to explain. But we need to understand one thing. A tone has to continue for a certain amount of time till we are able to have an audible impression on what tone it is. Thus, CSD will dominate sound coloration when two systems have similar SPL response. This is explained quite well in the Newell and Holland book. To allow faster decay, there must be ways to dissipate the stored electrical and mechanical energy in the driver. In the process of trying to dissipate these, sometimes there will be energy shifts through the spectrum. I think Earl has also mentioned this somewhere. If we do not have enough CSD resolution, it becomes almost impossible to see this shift.
Does a fast decay always sound the best? Well, the cleaner the speaker is, the more room reflections will effect the final listening experience. This is why we need to understand how we would like a speaker to be used.
The one serious issue with this approach is that it can hurt more than help if the system response off-axis is not also smoothly varying. This is too often the case for most direct radiator systems because few people will take the steps required, such as adequate diffraction control. IMO, DSP is the easy way out that can easily be a case of the cure is worse than the malady for the very reason that you state, the off-axis.
If that is the primary concern, then one must consider all that enters into it. Earl's position on an OS waveguide as satisfying that condition is probably the most concise, as much because it also helps to ameliorate three issues at once, those being diffraction (secondarily due to directivity control), non-concentric drivers issues (low crossover Fc) and smooth response due to reasonable directivity (nothing is perfect, of course). The use of DSP quite often ignores the off-axis.
That's not exactly true. One need only closely examine Dunlavy's best products to disprove this. It's all in the execution. In fact, it is better above 10K for direct radiators as I think even Earl would admit, but his position is that it's importance is questionable. The directivity control can be excellent horizontally for direct radiators, including diffraction when controlled. The limits of control for polar response is not due to diffraction, it is due to non-concentric drivers with Fc too high for the CTC spacing and the vertical off-axis.
I am not interested in going the OS route myself, but the fact of the matter is that for directivity control, at least in the front 2-pi space, low Fc is essential and that is the strength of the OS waveguide approach due to the need for only two drivers and the exceptionally low Fc that may be used.
Dave
I mentioned this ONLY in the context of extremely LF sound fields in rooms where the modal structure is discrete. Once there are enough modes that the energy can be carried at random, then this cannot occur.