So... what *disadvantages* can waveguides have?

[ernestcarl]

Using anecdotal mfr information to support your subjective view or support the mfrs limited info is the issue.

If you prefer your speakers, that is fine.

Going beyond that requires more than personal subjective opinion. Who can provide relevant information on performance, up-front, if not the manufacturer?

I avoid being emotionally attached to stuff. It does the job well for me or it doesn't. Preference is final over the opinions of others. I just don't claim to be superior re my personal preference or need to forensically defend it

I’m not sure why you think the advantages of preconditioning DSP in improving a speaker’s response is merely anecdotal. It is laid out quite clearly in the paper how it’s supposed to work — conditions and limitations included — the patent also goes into more detail -- look around at the citations and who has cited the patent. This system of signal processing and design has been used (in a limited sense or partially by others) for years. Yes, my subjective assessment of the speaker (S8) not to specifically have any apparent horn sound... well, obviously, that is very much anecdotal — but this should be provable with objective measurements.

Maybe you and others need the “final” verification... and that’s totally fine. Which is why I suggested to try contacting the people who actually use said tech if one is really interested. They have the measurements. They developed the wavelet analysis methodology. They probably have thousands of measurements for who knows how many horn waveguide speakers. Who better to ask for solid verification if not the manufacturer(s)/inventor. I doubt they are that stingy when it comes to keeping every single wavelet and impulse plot “secret” and under wraps. In fact, all the usual expected detailed spec sheets and product info are already readily available in their website. If you want more data, why not ask them nicely what specifically you want.

As for me, I personally don’t need Fulcrum Acoustic's verification since I can produce my own objective data. My speakers essentially use the same technology so to whatever limited extent I am able to use audio analysis software, I should be able to verify some of the claims being false or true myself. Which is why I asked how you want your “verification” served or in what form exactly. It should be possible to derive some objective data into the fore no matter how unsatisfactory (amateurish and incomplete) it may be to dyed-in-the-wool objectivists. As far as I know, although wavelet transforms have been out for a long time, this new kind of transient wavelet analysis is unique to Dave. Though, I reckon EAW uses something similar. The DSP implementation by RCF seems rather incomplete as there does not look like any improvement in the noise in their graph example:

Image from RCF

FYI: I don’t own any speakers made by Fulcrum. Nor have I gushed here emotionally about how good my S8s sound compared to other monitors. What the hell? I haven’t even mentioned any other speakers being inferior to my relatively cheapo, and butt-ugly Sceptre S8s. There are a over a dozen other speakers I’d be happy to exchange over these right this second — not because they sound bad, but because new and shiny audio gear always piques my interest.

My annoyance may have shown. Annoyance by what I see here as vague disinterest and quick-handed dismissal of something unfamiliar yet truly innovative -- few seem to mildly acknowledge that DSP preconditioning-optimization of horns/waveguides or speakers in general is even worth considering, yet alone real and already being implemented here and elsewhere (I'd be surprised to learn if most here think that this is science fiction) -- e.g. as in the case of improving speaker transients and reducing resonances whenever the sufficient conditions are met. Sure, it may be fine to use computers in the process of designing stuff, but when it comes to actually implementing complete speaker systems, people seem to be satisfied with the same old, same old techniques e.g. foam plugs, needlessly expensive exotic drivers, and passive crossovers, DSP limited to just xo and nothing else, leaving raw transients un-optimized/smeared as-is directly from imperfect horn waveguides. Granted, advanced digital signal processing isn't viable to every designer and speaker design out there, it is still worth considering and learning more about.

 

[ernestcarl]

I think that EQ would put the nail in the coffin on any "coloration".

Basic IIR EQ should be able to mitigate some excesses in the mid and HF response from a horn or even woofer -- even if only partially. If there truly are severe reflections and other horn resonances, then this type of simple equalization might not be enough without attenuating too much of the original sound.

 

[ctrl]

In this quote he seems to say that lots of people perceive "horn sound", that HOMs is something which we know to exist, and that HOMs can be an explanation for that.
But the thing which matters is still not so much what Geddes has or hasn't said, but to what degree those things are audible. As you say, we don't know much about that.

It would be important that the discussion goes beyond OS waveguides, these are not the culmination in terms of waveguides. It's immediately clear that at +-10deg are massive (back) reflections at the horn mouth, which even the foam-plug cannot suppress.

This is shown by the link of @tuga in post#54 where different horn and waveguide loudspeakers are shown in the wavelet representation and the HOM's are particularly pronounced with the OS waveguide - if the shown are really HOM's. They are in any case problematic time-delayed artifacts with high sound level.
But in the presentation slide it's not explicitly stated that the same compression driver was always used. We have no information about the mic distance,...

The existence of (backward) resonances in WG and horns is a fact (as others said already), just as edge diffraction occurs on any baffle and the resulting secondary sound source also produces effects due to the time-delayed superposition.

Waveguides like a lot of things in audio are a matter of design and implementation. If there's a downside, it's that the designer can screw it up. Around here the Klippel will reveal all...

Yep, a horn or WG whose +-deg90 frequency responses show only minor resonances will very likely not have a 'honk' sound - with good crossover tuning.

Image from RCF

This wavelet gif is definitely misleading if it is supposed to "prove" an improvement of the decay behavior.
For this, the FR should not change or would have to be normalized, which is not the case. The case "withFIRPhase" shows a lower sound pressure and therefore decays apparently faster than "withoutFIRPhase".
If I can see it correctly, the axis values of the wavelets (x-axis with milliseconds and the right dB color bar) also change.
It only shows that FIR filter work as expected


But it seems to me that the selected speaker examples of FA and RCF that @ernestcarl shows in post#64 or in post#23 are particularly extreme cases. With normal sized wave-guides or well designed horns such behavior does not occur.

Here are a couple of wavelet examples from a speaker with the SeasDXT, with a DIY WG and the single measurement without crossover of the XT1464 CD horn (pay attention to the scaling, this goes from 0.5kHz to 40kHz!, it is normalized to the "peak energy" on each frequency):

There is quite a bit of resonance and cancellation in the range above 20kHz, but there aren't any noticeable resonances and time-shifts in the audible range (beside the "normal" phase shift caused by crossover or the tweeter dropout itself).
It is interesting to observe the effect of the phase shift / group delay in the wavelet, when the black dashed line of the "peak energy time" moves further and further to the right with falling frequency
.

 

[tuga]

This is from another JMLC document:

 

[Duke]

The existence of (backward) resonances in WG and horns is a fact (as others said already), just as edge diffraction occurs on any baffle and the resulting secondary sound source also produces effects due to the time-delayed superposition.

There is a tradeoff relationship between the radius of the "lip" around the mouth of the horn, and the strength of the resulting backwards reflection. The bigger the lip radius, the weaker the backwards reflection. In other words, a larger radius lip would have reduced the strength of the reflection shown for the oblate spheroid waveguide.

"Radius" is an imprecise term - the curvature of the lip may but often does not have a constant radius. For a given horn diameter, the Le Cleac'h horn has a very large (and optimized) "lip" of very large "radius". How much real estate to devote to "waveguide" and how much to "lip" is a tradeoff to be juggled when designing an Oblate Spheroid waveguide, and my undestanding is that Marcel Batik's waveguide design software allows one to optimize this aspect.

In my opinion one of the reasons small waveguides can be hard to get right is that there simply may not be enough real estate available for a proper lip radius/geometry without losing the pattern control which is the raison d'etre of the waveguide in the first place.

 

[ernestcarl]

But it seems to me that the selected speaker examples of FA and RCF that @ernestcarl shows in post#64 or in post#23 are particularly extreme cases. With normal sized wave-guides or well designed horns such behavior does not occur.

These were quick image captures I took from their marketing videos. We can safely presume that these were definitely pre-selected (as some of their "best examples" -- evidently not favorably in RCF's case ) in the service of marketing purposes. Doesn't mean their DSP corrected speaker designs do not work as intended. But, yes, we do have to take such examples with a grain of salt. But the theory behind DSP correction or the application of "preconditioning filters" as method of improving loudspeakers systems in general esp. in large scale implementations (i.e. complex live sound reinforcement) is pretty sound. Since one cannot totally eliminate distortion from acoustical drivers, the goal should be achieving significantly "higher gain before feedback" -- hopefully the feedback is audibly low/microscopic enough to be essentially benign.

 

[ernestcarl]

BTW, if you modify a loudspeaker that uses special FIR filtering to create reverse/inverted signal images from the expected reflected backwaves -- you might just make things worse -- i.e. adjusting the external waveguide like turning it by 90 degrees or adding some foam, or even by modifying the internal damping significantly.

Here's an example:

While there clearly are benefits to such a speaker mod as you go lower in frequency... Notice how adding a significant amount of damping as well as blocking the port instead increased the amount of reflections.

To me, this seems like a strong indicator that this speaker does use such special FIR filtering.

Fortunately, my pseudo-anechoic EQ significantly reduces this unwanted side-effect.

If you don't like the black and white image or my use of such low-res settings, well, increasing the resolution or adding color in REW doesn't necessarily make things any clearer:

 

[617]

DSP is almost never a good solution for acoustical problems in my experience. Some interesting filters here though.

 

[dfuller]

DSP is almost never a good solution for acoustical problems in my experience. Some interesting filters here though.

Agree. Better to not brute force a mediocre design to flat with DSP. Instead, use DSP as a way to improve already-good engineering.

 

[ernestcarl]

DSP is almost never a good solution for acoustical problems in my experience. Some interesting filters here though.

Well, "never a good solution for acoustical problems" sounds pretty vague. It's just question of application.

Better to not brute force a mediocre design to flat with DSP. Instead, use DSP as a way to improve already-good engineering.

There's always a trade-off in any kind of design. But if adding additional signal processing can improve something with little to no trade-offs, there is no sense in avoiding it like the plague.

 

[617]

Well, "never a good solution for acoustical problems" sounds pretty vague. It's just question of application.



There's always a trade-off in any kind of design. But if adding additional signal processing can improve something with little to no trade-offs, there is no sense in avoiding it like the plague.

Acoustic phenomena are three dimensional. Manipulating a 2d signal to fix acoustic issues is a bit like trying to unscramble an egg.

 

[ernestcarl]

Acoustic phenomena are three dimensional. Manipulating a 2d signal to fix acoustic issues is a bit like trying to unscramble an egg.

Yeah, there are limitations. I get your point...

column 5 of the patent I previously linked to is clear in describing what are "good" candidates for such kinds of correction:
Creating Digital Signal Processing (DSP) Filters to Improve Loudspeaker Transient Response

"It should be understood that a loudspeaker mechanism to be identified for digital correction needs to be a mechanism resulting from stable, correctable behaviors. Specifically, the transient response of a loudspeaker typically represents the combined effect of a multitude of physical behaviors of components forming the loudspeaker. Some of these behaviors are nonlinear, time-variant, or spatially variable and therefore are not good candidates for digital correction. Other behaviors are sufficiently linear and time-invariant (LTI) and further sufficiently consistent spatially (or directionally) to be largely correctable with specialized digital filters. The present invention is directed to creating DSP filters to correct only those sufficiently LTI and spatially consistent behaviors, to thereby improve the transient response of a loudspeaker."


The S8 mentioned before just happens to tick all the requirements:

No point in showing the vertical phase responses of the KH120...

Impulse Response

'Clarity' at every angle is actually better with the S8. But the KH120 has a significantly smoother, more linear magnitude response. S8 can reach higher SPLs -- and so on and so forth...

Yet again, as mentioned: there are always trade-offs... Which design demonstrates better engineering? Assuming they cost $$$ essentially the same... The answer depends on which metric you're judging by. In reality, often aesthetic tastes and subjective opinion trumps even the raw metrics.

Though, I have to say, even super cheap smart devices with speakers like the Amazon Echo Dot still benefits from DSP -- e.g. going above volume five or six automatically applies a limiter for the bass to reduce distortion. So... yet again, the application intended for the correction needs to be considered as always.

 

[617]

Yeah, there are limitations. I get your point...

column 5 of the patent I previously linked to is clear in describing what are "good" candidates for such kinds of correction:
Creating Digital Signal Processing (DSP) Filters to Improve Loudspeaker Transient Response

It should be understood that a loudspeaker mechanism to
be identified for digital correction needs to be a mechanism
resulting from stable, correctable behaviors. Specifically, the
transient response of a loudspeaker typically represents the
combined effect of a multitude of physical behaviors of com-
ponents forming the loudspeaker. Some of these behaviors
are nonlinear, time-variant, or spatially variable and therefore
are not good candidates for digital correction. Other behav-
iors are sufficiently linear and time-invariant (LTI) and further
Sufficiently consistent spatially (or directionally) to be largely
correctable with specialized digital filters. The present inven-
tion is directed to creating DSP filters to correct only those
sufficiently LTI and spatially consistent behaviors, to thereby
improve the transient response of a loudspeaker.

The S8 mentioned before just happens to tick all the requirements:

View attachment 109068

View attachment 109069

No point in showing the vertical phase responses of the KH120...

Impulse Response
View attachment 109070

View attachment 109071

'Clarity' at every angle is actually better with the S8. But the KH120 has a significantly smoother, more linear magnitude response. S8 can reach higher SPLs -- and so on and so forth...

Yet again, as mentioned: there are always trade-offs... Which design demonstrates better engineering? Assuming they cost $$$ essentially the same... The answer depends on which metric you're judging by. In reality, often aesthetic tastes and subjective opinion trumps even the raw metrics.

Though, I have to say, even super cheap smart devices with speakers like the Amazon Echo Dot still benefits from DSP -- e.g. going above volume five or six automatically applies a limiter for the bass to reduce distortion. So... yet again, the application intended for the correction needs to be considered as always.

The manipulation of directivity of both microphone and speaker arrays is one of the most interesting uses of dsp. I believe most smart speakers use this technology.

 

[Savi]

Hi. Sorry if I missed the answer to my question in the previous posts. I was wondering was it the effect of a waveguide on tweeter distortion ? Do you know a harmonic distortion measurement comparison of the same tweeter in the following conditions:
- Without waveguide
- With waveguide
- With waveguide and the frequency correction inherent to the waveguide use
?

 

[Chromatischism]

Hi. Sorry if I missed the answer to my question in the previous posts. I was wondering was it the effect of a waveguide on tweeter distortion ?

Reduced distortion due to increased efficiency at and above the crossover point.

 

[Ro808]

Waveguides are part of a system.
A waveguide is a horn and an acoustic horn (other than a horn instrument) combined with a compression driver constitutes a single system.

Some basic underlying physics (Houterman, 2020 (edited)):
"Acoustic impedance Z is defined as the ratio of the sound pressure p at a boundary surface to the acoustic volume flow U through the surface. The acoustic impedance is the property of a geometry and the characteristic impedance is a property of a medium.
The acoustic impedance is a complex number and exists of a resistive part (real part) and a reactive part (imaginary part) and the inverse (1/Z) is called the admittance. The resistive part represents the energy transfer in an acoustical wave. And the resistive part is the acoustical resistance and is due to viscosity, which may be considered as friction between adjacent layers of air. When air travels through a tube, the velocity of the particles at the boundary are zero and a maximum at the centre of the tube, this causes friction between adjacent layers of air. Acoustic impedance also exists of an imaginary part, the reactive part, is associated with the reactions of forces of inertia (masses) or elasticity (compliance). If a tube is small enough it also has inertance, this results in an increasing of the reactive component with increasing frequency. Therefore, at a higher frequency the acoustical impedance is higher.
The the larger the mismatch between the acoustical impedance and the impedance of air, more sound will get reflected. To obtain minimal reflectance the circumference of the opening should be bigger or equal to the wavelength.

When the gap between the impedance at the opening of the horn and the impedance of air is bigger more reflectance occurs. If reflectance occurs the power output drops, due to a part that is reflected and to the standing waves created by the reflected part. A standing wave is a wave that remains at a constant position. This occurs when two waves with the same frequency are travelling in opposite direction and interfering with each other. When they cancel each other out the amplitude of the wave drops. Standing waves causes drops in the output signal for certain frequencies, which influence the IR measurements. The standing waves can be reduced by not having the walls parallel to each other, for example creating a conical shape. However, sound will no longer propagate along the tube without any change if the tube is converging. For cylindrical tubes, the surfaces of constant phase are planar (plane wave), while for conical tubes, the waves are spherical. If the width of a tube is changing slowly, a sound wave propagates without reflection. If the area of the tube is changing in a large ratio, most of the sound power is reflected. For a conical shape, the area of the tube is changing quadratic. Hence, the shorter the wavelength, it will reflect further out in the horn.

Another phenomenon which causes irregular frequency response and sound power drop is distortion, distortion arises due to high particle velocity. If the inverse horn is converging, the particle velocity will rise as well. A combination of the sound power generated by the driver and a rapidly converging horn causes distortion. Distortion caused by too high sound power, can easily be tested by lowering the sound power level and compare the frequency response and directivity. More distortion also occurs when increasing the length of the horn. Distortion occurs in the propagation of the waves itself, it is most noticeable in waves that expand slowly, as in horns, where distortion increases with the length propagated."

Besides, compression driver distortion (2nd and 3rd harmonic) through the midrange (500 Hz to 1.5 kHz) is typically higher compared to modern cone midrange transducers.
Traditional compression drivers with high compression ratio’s generate strong harmonic distortion when high acoustic amplitude levels exceed the capability of the air (trapped between the diaphragm and the phase plug) to act as a linear spring. High compression ratios do improve efficiency; however the penalty comes in the form of increased distortion.

So "when a 2-way loudspeaker system with a waveguide + compression driver sounds weird" it makes sense not only to zoom in on a limited number of measurements, but to consider the interdependence between different electro-acoustic (sub)systems.
Moreover, loudspeaker drivers are limited bandwidth devices, which necessitate well-considered choices between various trade-offs. This is particularly true for 2-way systems.
If a low crossover frequency is desired for a 2-way speaker system with waveguide, it's important to select a suitable compression driver (and woofer).

During the rise of Gedlee's and Pi Speakers' 2-ways with waveguide, as well as similar systems from the DIY community (Econowave, DIYSoundgroup) 10-15 years ago, the B&C DE250 was the driver of choice.

Let's take a look at some distortion plots of the DE250 in combination with a slightly deeper horn.

Compared to the best of class 1" compression drivers, the DE250 suffers from considerable amounts of higher odd harmonics that become downright disturbing above 90 dB at 1m.
Now what could be a possible effect if such a driver is mounted behind a shallow waveguide and used in a 2 way system from 800 Hz?

 

[eddantes]

Would a waveguide (horn) covered in (or manufactured from) a textured surface provide some adavantage? Not unlike some speaker designs where the tweeter is surrounded by a fabric... I'm thinking old Spicas or Wilsons.

 

[tuga]

Would a waveguide (horn) covered in (or manufactured from) a textured surface provide some adavantage? Not unlike some speaker designs where the tweeter is surrounded by a fabric... I'm thinking old Spicas or Wilsons.

Golf ball dimples?

 

[eddantes]

Errr ... I was thinking felt... or velvet... Idea being that difraction might keep the modes at bay... I'm just spitballin' ... I have no idea.

 

[Ro808]

All horns/waveguides are finite and most of them require some kind of mouth termination in order to reduce reflections and edge diffraction. Tractrix, spherical and in particular Jean-Michel Le Cleac'h and MinPhase horns excel in this respect. They open up to a right angle, or, in the case of JMLC and Minphase horns, include a generous round-over.

A baffle acts as a kind of mouth extension for smaller waveguides and is considered by most manufacturers to be an adequate means of reducing reflection and diffraction.
Depending on the shape of the horn mouth and the application, damping materials such as foam are used.
2 examples are Peavey's Quadratic-Throat Waveguide designed by Charles Hughes and Frazier's CAT series.

Frazier published a whitepaper on the topic.