Is there such a thing as "box sound"?

[terryforsythe]

Also, you can probably correct a low-Q system to a high-Q without much penalty.

For the most part, I agree, as long as the Qtc still is within reason. The advantage of lower Qtc is increased damping, which shows up in the time domain plots. Some people think a Qtc = 0.5 is ideal, while some designs target 0.71. 0.5 is slightly better in the time domain, but 0.71 is better in the frequency domain around the normalized frequency f = 1 (see charts in post # 13).

The other way around, not so much: you'll need to apply a significant boost.

It depends - take a look at the frequency domain plot in post #13. With a Qtc = 0.5 you would need to add 3dB (double the power) at the normalized f = 1 frequency to get the same output as Qtc = 0.71. With a Qtc > 1 you would EQ by reducing power around that frequency.

But, the plot in post #13 may not be completely accurate below the f3 frequency in real world systems. In real world systems a lower Qtc tends to output more SPL at frequencies below the f3 frequency. This plot probably is a better representation of real world systems:

 

[Pareto Pragmatic]

I do appreciate the effort in this thread, it is very interesting. As interesting as it can be given that I will never go open baffle (in my particular room).

"it doesn't have a box sound, the bass is just there"

I can say that about pulling my "box" speakers out 10'/3+m from the wall. The bass is cleaner. There's also less of it. For all my speakers.


If I wanted to dial in what people MIGHT be hearing, I would need a very symmetrical room. I would build 2 ways, one open baffle and one ported ... with a port for each driver. To the extent that there is a "box" sound, I would think that would reveal it with proper matching. Same system, same drivers, same room, different design, but both firing sound forward and back.

That takes the people/OB fanbois out of the equation, and gets directly at the answer sought.

 

[Mort]

That takes the people/OB fanbois out of the equation, and gets directly at the answer sought.

But then we would have nothing to talk about.

 

[terryforsythe]

If I wanted to dial in what people MIGHT be hearing, I would need a very symmetrical room. I would build 2 ways, one open baffle and one ported ... with a port for each driver. To the extent that there is a "box" sound, I would think that would reveal it with proper matching. Same system, same drivers, same room, different design, but both firing sound forward and back.

It would be difficult to make it a fair comparison. Open baffle drivers tend to perform best using high Qts, e.,g., above 0.65, whereas the optimal Qts for a ported speaker is quite a bit lower - around Qts = 0,4 provides efficient use of enclosure volume in a ported design.

 

[voodooless]

But as, mentioned, this is applicable to compact midrange drivers within a 3-way system solely, and the differences are measurable via waterfall plot.

I would love to see those measurements, both frequency and waterfall.

 

[voodooless]

Some people think a Qtc = 0.5 is ideal, while some designs target 0.71

Ideal is a flat in-room frequency response. Only looking at the Q of your woofer/box combo is rather pointless, at least for bass response.

 

[Arindal]

Ideal is a flat in-room frequency response.

A flat in-room response is desirable solely for a particular frequency range, in which constant directivity from loudspeaker side, and constant absorption grade of the room can be guaranteed.

Open baffle drivers tend to perform best using high Qts, e.,g., above 0.65,

The quality factor of the bass driver becomes almost irrelevant in an open baffle configuration. The decay around resonance frequency alone is not audible under such circumstances, and frequency response as well as directivity are anyways defined by the dipole configuration.

whereas the optimal Qts for a ported speaker is quite a bit lower - around Qts = 0,4 provides efficient use of enclosure volume in a ported design.

Qts is measured under free-field conditions, the resulting system quality factor Qtc of a vented box is a different thing. The subjective decay, tightness of bass, impulse response, is anyways mostly defined by the resonator (port or PR), not by the active driver´s Qtc alone.

 

[ZolaIII]

But then we would have nothing to talk about.

I (agree too) disagree... He can have both more bass and cleaner one, just has to go 2.2. Reinforce what you need to get, let it sum on it self and make sure it can't propagate more than good to the ritam main bass section (120 Hz for the sake of ELC) which should be on the mains anyway and stressed out that are pulled out so that you get better back to front (or depth of the field) ratio (ISO 3382 to room what's posible). Similarly woffer's on mains should be contacted and cut similarly to beginning of mids (250 Hz), rest is to talk about for what ever preference (from flare to dispersion angle).

Regarding design I whose characterised as someone who hate port's. Truth is I just don't see the point of them on 99% of shelf designs (I understand the intention of designer and that's it). Ask your self simple question. Does port have function when woffer is cut healthy above it so that it doesn't get excited much? Where does it end like that? Somewhere in between open and ported regarding decay (18 dB). And where room decay leads us? The same towards 18 dB per octave. That's what I have in mind when I say aim towards Butterwort regarding Q. You can use port as either or. In 90% cases it's used to reinforce some response from woffer which is over to it's -6 dB and to a point this works as those are small woffer's but really not good. 9% are those that try to keep it up to - 5.5 dB of driver and partly even to offset part of standing waves and non linear woffer excursion to it's Fs (more serious woffer's 8" and up). Then there is 1% that uses it just to woffer Fs (really big subwoofer's) and to time delay (still cut under it) with properly done cabinet of course. With compression closed enclosures you again balance between time coherence and extension and it again ends the same way (18 dB slope). Cutting close enclosure one's above Fs to where phase/impedance meets corresponds to driving them easy and to great time domain but you need quite big and capable one's to do that. Anyway you can't beat physics eventually argue about what's better compromise and to what extent.

 

[Pareto Pragmatic]

It would be difficult to make it a fair comparison.

The question for me would be is it MORE fair than the OB fans opinion, or comparing different speakers/drivers?

Pick a Qts in the middle, it would be off in different directions from optimal (too high too low), but would that overwhelm the box effect? Or not? No idea, but I would enjoy learning.

 

[voodooless]

A flat in-room response is desirable solely for a particular frequency range, in which constant directivity from loudspeaker side, and constant absorption grade of the room can be guaranteed.

That's why the comment is mainly about bass response: there is no directivity there with a box, all in omni.

 

[terryforsythe]

Qts is measured under free-field conditions, the resulting system quality factor Qtc of a vented box is a different thing.

Correct, except Qtc typically is used for a closed cabinet, not a vented cabinet. To be clear:
Qts is for total Q factor for the driver where Qts = (Qms * Qes) / (Qms + Qes).
Qtc is the Q of the system in a closed cabinet.
QB is the total system loss for a vented cabinet where 1/QB = 1/QL + 1/QA + 1/QP. But, QA and QP tend to be not very significant, so QL commonly is used.

The subjective decay, tightness of bass, impulse response, is anyways mostly defined by the resonator (port or PR), not by the active driver´s Qtc alone.

As you probably are well aware, when designing a vented cabinet you should to design the cabinet around the driver's Thiele-Small parameters, including Qts. The point I was getting at is that drivers with very high Qts (e.g., >> 0.4) are not ideal for ported enclosures - in comparison to using a driver with lower Qts, the cabinets become excessively large to achieve a desired QB and the frequency response is less linear. But, with an open baffle design, drivers with high Qts (e.g., 0.7) tend to work well since there is no enclosure and the driver is operating in free air, while the bass response of a lower Qts driver rolls off prematurely.

 

[terryforsythe]

Pick a Qts in the middle

The way I would approach it is to get a pair of drivers with a Qts of maybe around 0.44. Use one as is in the vented enclosure. Then use the old school method of applying shellac to the spider of the other one to increase the Qts (specifically, Qms) for use in the open baffle. It also will increase the fs, though.

Another way to approach it would to find two different drivers with the same fs, but one with a Qts of around 0.4 and one with a Qts around 0.7. Likely, the higher Qts driver will be larger.

 

[DVDdoug]

My take-away is that "box sound" is not defined and we should avoid the terminology. If somebody else uses it, we can either ask what they mean or ignore them.

There is a LOT of "audiophile terminology" that I ignore.

 

[Duke]

I have taken part in an experiment comparing different enclosure sizes for the midrange drivers, under anechoic conditions and corrected for on-axis frequency response. The difference is astonishingly huge...

Dude, please don't leave me hanging like that! WHAT was the huge audible difference; what did you learn from this?

I have experimented with different enclosure stuffing strategies without changing the enclosure size for a midrange driver, but have never experimented with different enclosure sizes. If you are free to share what you learned, that would be very valuable information to me. Or if you can't share it, please point me towards where I can find/buy that information.

We are talking about wavelengths which make it difficult to time-window the measurement, or separate the measurements of different sources...

Yes!

I think that the resonances (whether internal or panel) which result in "boxiness" begin to arrive AFTER the direct sound, so there is a little bit of a time gap. I'm indulging in speculation now: BECAUSE this kind of resonance has its onset a little bit later in time, the ear/brain system's "masking" characteristic doesn't mask it as well as if it was simultaneous with the direct sound. Something like this happens in diffraction horns, wherein the diffracted energy arrives a little bit later in time and therefore is not masked and therefore is more audible and objectionable than we would suspect from the frequency response curve alone.

 

[Pareto Pragmatic]

The way I would approach it is to get a pair of drivers with a Qts of maybe around 0.44. Use one as is in the vented enclosure. Then use the old school method of applying shellac to the spider of the other one to increase the Qts (specifically, Qms) for use in the open baffle. It also will increase the fs, though.

Another way to approach it would to find two different drivers with the same fs, but one with a Qts of around 0.4 and one with a Qts around 0.7. Likely, the higher Qts driver will be larger.

Sounds reasonable. I do think that would provide a better match in terms of real world designs.

Could try both. Control for the driver in one case, for the fs in the other. Not a big deal for measuring differences in output, but we're up to a lot of conditions for human subject/perception testing.

Would you eq to match levels? Thinking bass roll off making sure the slopes match... whatever that took. Or would you design to match levels?

 

[Salt]

Thinking about Horn, TML and others ... the calculation becomes even more ... difficult.

 

[terryforsythe]

Would you eq to match levels? Thinking bass roll off making sure the slopes match... whatever that took. Or would you design to match levels?

I would match the levels with the electronics. I would spend time simulating the bass responses before buying any components. I use VituixCAD, but there are a number of other programs that should serve the purpose of simulating bass response.

 

[Arindal]

WHAT was the huge audible difference; what did you learn from this?

That decay and resonance issues in bass and lower midrange play a crucial role in perceived tonality, transparency and timbre, completely independent from frequency response and directivity.

Three loudspeakers could show absolutely identical frequency response with only their midrange enclosures differing, we will nevertheless perceive voices reproduced by the high-Q midrange as ´bloated´, ´overly warm´, ´lower midrange dominant´, ´resonant´, the low-Q one as ´warm, but balanced´ lower midrange, while the open baffle/cardioid version sounds comparably lean, more transparent and more like you would expect it from a recording of an acoustic event. Note, that the former does not necessarily mean a ´boxy´ sound, but something into that direction.

AFAIK this was not widely published back than, but discussed a lot, when several loudspeaker manufacturers (MEG, Amphion, Kii, D&D) decided to go for cardioids in the lower midrange, and some in the bass as well. Interestingly, in recent years we see a wave of cardioids hitting the market.

I think that the resonances (whether internal or panel) which result in "boxiness" begin to arrive AFTER the direct sound, so there is a little bit of a time gap.

That sounds like an explanation for lower frequency effects, but I cannot really imagine this to be the main effect for lower midrange ´boxiness´, or the effects described above in the 300-700Hz band. I mean, we are talking about midrange enclosure sizes of roughly 0.5-3l.

If group delay is not the main issue here, my personal take would be that decay is crucial, independent from frequency response. Particularly for typical fundamental frequencies of human voices, we seemingly have a sense for sonic events which are either decaying quickly, or decaying much longer, with the latter being recognized as ´sonorous´, ´warm´, ´full-bodied´.

Something like this happens in diffraction horns, wherein the diffracted energy arrives a little bit later in time and therefore is not masked and therefore is more audible and objectionable than we would suspect from the frequency response curve alone.

Again, absolutely can confirm your experience with diffraction horns, and it sounds like an explanation, but there might be other ones. I would argue that such delay between two events of the same frequency, does have an impact on frequency response as well, and our ears rather perceive the resulting comb-filtering effect and its peaks. With horns in general, it is pretty difficult to separate any of the effects taking place at the same time, may it be in listening tests or measurements.

 

[voodooless]

Three loudspeakers could show absolutely identical frequency response with only their midrange enclosures differing, we will nevertheless perceive voices reproduced by the high-Q midrange as ´bloated´, ´overly warm´, ´lower midrange dominant´, ´resonant´, the low-Q one as ´warm, but balanced´ lower midrange, while the open baffle/cardioid version sounds comparably lean, more transparent and more like you would expect it from a recording of an acoustic event. Note, that the former does not necessarily mean a ´boxy´ sound, but something into that direction.

And yet we have this:

BLINDTEST: Midrange 360-7200hz, NO audible difference whatsover.

Just realized no proper thread was made to explain in a simple way the blindtest I organized a little while ago: So here it is: Blindtest. Rotating cube (enclosure) with 2 different drivers per round, one driver at a time, on-axis to the listener. Bandpass crossover 360hz @ 7,200hz (48db/oct...

www.diyaudio.com

The Q itself cannot be the issue here. Because if you correct the frequency response, you correct the Q. This is just a minimum-phase property. What is left is the fact that you might put a driver in a too-small box, which is poorly damped. If there were audible differences, it's because of what happens in that box, not because of Q.

 

[Arindal]

And yet we have this:

So, in this case different midrange drivers have been tested under usual circumstances (enclosure volume) against each other after frequency response correction? That is not quite what I was referring to, in my case, the very same drivers were used, the only difference was how they interacted with the enclosure volume.

The Q itself cannot be the issue here. Because if you correct the frequency response, you correct the Q. This is just a minimum-phase property.

As mentioned, the whole thing is not a minimum-phase system. You could correct for the frequency response, but particularly decay, as visible in the waterfall plot, persisted being completely different, particularly with the open-baffle arrangement.

If there were audible differences, it's because of what happens in that box, not because of Q.

Qtc is a result of driver + enclosure. In the particular case, the open-baffle variant stood out, producing a different perception of timbre despite from identical anechoic response.

That's why the comment is mainly about bass response: there is no directivity there with a box, all in omni.

But there are different absorption grades in the bass bands (think of unwanted resonators of all kind), and there is different decay at different frequencies. Both make it illogical that a perfectly flat response would always be the right goal, and would guarantee identical sound perception. That is not the case in my experience.

except Qtc typically is used for a closed cabinet, not a vented cabinet.

Correct, because for a vented cabinet, it is almost impossible to calculate or measure a precise Qtc, as you have several sub-systems which are resonators in their own rights.

when designing a vented cabinet you should to design the cabinet around the driver's Thiele-Small parameters, including Qts. The point I was getting at is that drivers with very high Qts (e.g., >> 0.4) are not ideal for ported enclosures - in comparison to using a driver with lower Qts, the cabinets become excessively large to achieve a desired QB and the frequency response is less linear.

Very much depends on other driver properties, like Vas. There are some pretty common compact midwoofers with Qts>0.5 but rather restricted Vas, which are suitable for vented systems and are commonly used for this purpose (can check for examples by Wavecore and Peerless).

with an open baffle design, drivers with high Qts (e.g., 0.7) tend to work well since there is no enclosure and the driver is operating in free air, while the bass response of a lower Qts driver rolls off prematurely.

Yes, high-Q drivers work well in open baffle, but the precise Qts itself is not really relevant in my understanding, can be even much higher than Qts=1. The rolloff is usually dominated by the acoustic short circuit and calls for correction anyways. Such drivers are not very common, as high-Q models were in first instance designed for car-hifi use in some kind of free-air/indefinite volume closed box design.