Causes of "harsh" sound in speakers (and tweeter material)

[dfuller]

So if you equalize different tweeters in the free field to an identical frequency and phase response, they will sound identical.

Eh, yes and no. Soft domes have markedly different off-axis behavior from on-axis. That can sound very different in room.

 

[ctrl]

Eh, yes and no. Soft domes have markedly different off-axis behavior from on-axis. That can sound very different in room.

Please read my post#18 again. In free field on-axis they will sound identical. What you mean is listed in 2 (and 3, 4).

 

[Prana Ferox]

Tweeters are normally less able to handle power than woofers

The human ear tends to be a lot more efficient at higher frequencies (or at least away from the bass) and there is considerably less energy in most musical content in the higher registers. Tweeters in general are more efficient by design (less mass to move around, mostly.) This is why tweeters tend to be padded (i.e. resistance added in series/parallel in the crossover circuit) so the woofers can keep up, and why we don't talk about pants-flapping treble.

At high output levels you will get power compression as the voice coils heat up and increase their own resistance, causing their output to drop (often causing the operator to increase the volume, making them heat up more.) There are a lot of conditions here but generally this will affect bass drivers more, which can cause a speaker to sound 'brighter' as it heat soaks. Further, strained drivers distort more, showing up as increased high-frequency signal.

 

[kemmler3D]

With a good dome and surround design, the drive motor of the tweeter should be decisive. In post#18 I used three Bliesma tweeters with identical drive motor and front plate but different dome as an example - you can find the reviews here.

The measurement of the harmonic distortion (about 90dB@1m) is almost identical in the intended operating range above 2kHz and at least in this example practically without influence:
View attachment 324527 View attachment 324528 View attachment 324529
Source

In addition, harmonic distortion is extremely difficult to hear due to masking effects.

With a good dome and surround design, the drive motor of the tweeter should be decisive

I would tend to agree with that... but comparing Bliesma tweeters for breakup is like comparing Michael Jordan vs. LeBron James, it doesn't give you much sense of how average competitors look. IIRC the Bliesma tweeters are notable for having an ultrasonic breakup mode in most units. I think if we go lower down the spectrum (or at least leave the peak of the mountain) we can probably find silk dome tweeters with audible breakup?

 

[MaxwellsEq]

1. The first point is that the cone material itself does not affect the sound (there is no "ultra fast" material). So if you equalize different tweeters in the free field to an identical frequency and phase response, they will sound identical.
There is a master's thesis from Germany that investigates exactly this. I have discussed this work in detail here.

2. However, different materials have different radiation characteristics with identical tweeter designs, as the break-up behavior is different. A silk dome tweeter breaks up very early but in a controlled manner, a beryllium dome tweeter breaks up very late and often in an uncontrolled manner, but the tweeter shows almost ideal piston behavior up to high frequencies - a silk dome does not.
You can find more details here and following posts.
As an example, here is an identically constructed tweeter with the cone materials silk, berylium and diamant. The FRs are normalized to the on-axis FR. Each diagram show the measured FR for different angles compared to an ideal piston like behaving tweeter:
View attachment 324506 View attachment 324507 View attachment 324508

3. Even small changes (small offset or tiny "waveguide") on the front plate of the tweeter lead to changes in the radiation. You can find more details here.

4. The overall radiation of a loudspeaker plays a major role (apart from direct sound) in how the sound is perceived - especially when the SPL increases.
Speakers with identical on-axis frequency response can sound completely different. Therefore, complete measurements as required by the CTA-2034-A standard are essential to identify potential problems.

For me personally, for example, in the frequency range around 2-4kHz in the sound power, early reflection and PIR, there should be no hump at all (small dip is okay, but can sound "lame", too laid back). Otherwise I perceive the speaker as harsh and aggressive.
View attachment 324515
Since SP, ER and PIR are all "averaged" curves, even the smallest changes can have a major impact on the sound.

A flat on-axis FR (and LW) is usually not enough for a good speaker sound in a normal listening room - except the speaker has perfect radiation, then yes (near field listening in heavy dampened studio is another matter).

Update:

A speaker with design flaws, like below with such radiation can never be fixed by EQ - it will never sound correct. Therefore one needs full measurements of a speaker.

View attachment 324518 View attachment 324519

This is a very informative analysis. Thank you.

 

[MaxwellsEq]

The human ear tends to be a lot more efficient at higher frequencies (or at least away from the bass) and there is considerably less energy in most musical content in the higher registers. Tweeters in general are more efficient by design (less mass to move around, mostly.) This is why tweeters tend to be padded (i.e. resistance added in series/parallel in the crossover circuit) so the woofers can keep up, and why we don't talk about pants-flapping treble.

At high output levels you will get power compression as the voice coils heat up and increase their own resistance, causing their output to drop (often causing the operator to increase the volume, making them heat up more.) There are a lot of conditions here but generally this will affect bass drivers more, which can cause a speaker to sound 'brighter' as it heat soaks. Further, strained drivers distort more, showing up as increased high-frequency signal.

I don't disagree. But I'm fairly confident that if you apply increasing but equal levels of power to a woofer and a tweeter, the latter will fail at a lower power than the former.

 

[DSJR]

I think the issue is multi-faceted really. Some existing tweeters can sound most 'harsh' if crossed over too low in frequency (to get uniform directivity) even if superficial response plots tell otherwise (I learned this first hand from a designer who demonstrated this aspect to me). Simple first order (a series cap) 'crossovers' can make a tweeter exhibit this if the cap value is too high, but reducing said value can then develop a nasssssty tinselly peak instead, something I've also had demonstrated to me.

Don't forget the often ragged response in the upper hundred Hertz region of interfering ports, cone surrounds not fully optimised and general resonances and phase issues around 1 - 3kHz which can also add to issues right where our ears are most sensitive... many such boxes seem to get good reviews here as the Klippel doesn't always seem to show it (why good speaker designers LISTEN as well, once they have the basic tech side sorted out and they then fine tune as best they can), or if it does show said uneveness, these less then smooth upper hundred Hertz regions are all but ignored unless it's a port issue which seems the case in many wonder-boxes tested here.

Lastly, no idea as to the age of the OP, but our (male?) ears do go off big time once over mid fifties and some like mine, far worse than others sadly. In addition to age related high frequency hearing loss, our sensitivity to potentially harsh tones can increase because of the hearing loss I gather. I cannot stand to be in the same room as Naim driven LS3/5A's (or many ice-cold toned Dynaudio models) of any description now as it's unbearably shrill and 'squeaky' where in the past younger years, it was simply annoying.

P.S. I owned NS1000's for a couple of years back in the late 80's and found them rather 'dead' up top (whatever you did with the controls) BUT, I suspect a hugely powered amp may well overcome this as the entire speaker to me was overdamped, preventing most 'period' domestic amps from fully opening them up. Mine were near-field and I ran them as suggested with mid set to 10:30 and tweeter at 11:30 positions. Setting them at '12:00' just took the bass down a tad more but didn't make them sound 'clearer.' If I knew then what I know now and used a seriously powerful professional amp with them, I may have got a better and more realistic sound..

 

[thanossapiens]

Lastly, no idea as to the age of the OP, but our (male?) ears do go off big time once over mid fifties and some like mine, far worse than others sadly. In addition to age related high frequency hearing loss, our sensitivity to potentially harsh tones can increase because of the hearing loss I gather. I cannot stand to be in the same room as Naim driven LS3/5A's (or many ice-cold toned Dynaudio models) of any description now as it's unbearably shrill and 'squeaky' where in the past younger years, it was simply annoying.

Thanks for your detailed reply. I am 25 years old can listen to about 17khz

P.S. I owned NS1000's for a couple of years back in the late 80's and found them rather 'dead' up top (whatever you did with the controls) BUT, I suspect a hugely powered amp may well overcome this as the entire speaker to me was overdamped, preventing most 'period' domestic amps from fully opening them up. Mine were near-field and I ran them as suggested with mid set to 10:30 and tweeter at 11:30 positions. Setting them at '12:00' just took the bass down a tad more but didn't make them sound 'clearer.' If I knew then what I know now and used a seriously powerful professional amp with them, I may have got a better and more realistic sound..

To me they sound bright, especially with the ncore amp I use, with older(muddier) amps it sounded more polite on the highs.
These speakers roll off pretty soon actually, so I'm surprised I feel like this towards them, and no they are not defective I measured the tweeters recently.

 

[gnarly]

The biggest cause of harshness ime/imo, is the lack of dynamic headroom (or downright average SPL compression if problem is acute.)
Matters 10-100x more than any tweeter material, or what type HF/VHF driver is used.
Applies throughout the spectrum too, mids, low, and subs.
Loss of linear output, peaks included, sound harsh.

If you hear harsh below the onset of linearity collapse, ime it's invariably simple response differences vs what we are used to.
I think then, brighter and harsh are being used as substitutes for each other..

F sound gets brighter with an equipment change...a change which is really nothing more than an EQ.....
Well, my suggestion is simply counter the equipement change EQ.....like high-shelf it down.

And then see how loud it can play before sounding harsh

 

[thanossapiens]

And then see how loud it can play before sounding harsh

Indeed most systems sound way harsher above 80-90DB imo

 

[valerianf]

"Indeed most systems sound way harsher above 80-90DB imo"
@thanossapiens , I can only disagree.
When I was young, in the 80's, I was going to a night club that had a custom sound system.
Sound level was far above what is allowed nowadays (regulation).
The large custom towers (may be 2m high by 4m large) had bullet tweeters.
Never I heard the high frequencies harsh, but for sure I lost some of my audition.

Top designs are not hash because the designer prevented it to happened.
We need to run away from harsh speakers.

 

[sigbergaudio]

One of my biggest issues when listening to different speakers is that a lot sound unbearably harsh(sibilant?) to me, even if they measure flat-ish on the treble. Sometimes its hard or impossible to EQ away for some reason, even if I know where the peak is. This made me think it might be some sort of sibilance, but looking at spectral decay plots (i dont know if these are the most relevant ones but im a newbie at this), while not perfect, a lot seem to be ok.
I noticed that the worst offenders are usually hard domes like diamond or beryllium tweeters. I'm not sure why, but I dont think its placebo since most of the time I got to hear super expensive stuff from B&W or Focal,etc I was thinking "I bet this is gonna sound amazing" and not the opposite.

Harshness is typically due to the speakers, the room acoustics or a combination of the two. Some records are inherently harsh, but they are few and far between.

I haven't heard the NS1000s in many years, but if I remember correctly they are a bit lean (too little bass). This can end up sounding harsh due to the tonal imbalance.

Perhaps most recordings, even "audiophile" staples like Dire Straits are kinda bad and it shows more on good speakers?
I'd love to hear your thoughts if you have any, thanks

No, most recordings are quite okay, and I've yet to encounter a harsh Dire Straits recording. Most are good to very good. Speakers that "reveal bad recordings" have a problem (often several).

There are lots of (often expensive) speakers that claim to be good that sound harsh.
There are no good speakers that sound harsh.

 

[sigbergaudio]

Something else I have thought of is that maybe these tweeters can handle power a lot easier than their woofers, so at higher volumes there is elevated treble? I have no idea if that's a thing generally though.
On that note I also want to ask if "trails" in spectral decay plots imply sibilant sound or something else.

The tonality (energy balance) of the speaker itself doesn't change with volume unless one or more drivers go into compression or distort. What can however happen in a room that isn't properly dampened is that you get excess reflections as the volume is turned up. This is typically experienced as harsh/hard midrange and treble.

 

[Sokel]

The tonality (energy balance) of the speaker itself doesn't change with volume unless one or more drivers go into compression or distort. What can however happen in a room that isn't properly dampened is that you get excess reflections as the volume is turned up. This is typically experienced as harsh/hard midrange and treble.

Same goes with amps that are considered enough for max levels but NOT peaks which can be 20db above so they often clip.
First thing an amp must have is clip indicator.

 

[fpitas]

First thing an amp must have is clip indicator.

Or just size it appropriately. Hard to have too much power.

 

[Thomas_A]

I think harsh should be defined since it could mean any error between 1-10 kHz. IMO harshness can be related both to peaking in the 3-4 kHz region and sibiliant distorsion higher up in frequency. It can be both in the recording and in the speakers. Perhaps the room also if you have some odd shapes or very lively. That said I almost never percieve harshness from my speakers, something that others also say about them. ”Polite and kind sounding speakers, never aggressive” ( or ”snälla” in Swedish).

 

[fpitas]

I think harsh should be defined since it could mean any error between 1-10 kHz

Heck, in developing my crossover I found that excess 800Hz sounds harsh.

 

[mhardy6647]

Heck, in developing my crossover I found that excess 800Hz sounds harsh.

In all seriousness (realizing that this might be a rabbit hole that this thread doesn't need) -- some of us, @fpitas included, live in a loudspeaker world where the interactions of a compression driver, its acoustic transformer

Spoiler

horn -- or "waveguide"

, the transition from driver to said transformer (the "throat"), and the air in the room are critical to the wavefront that emerges from that transformer. All such devices have a cutoff frequency, and behavior in the vicinity of that cutoff can get... interesting.

In other words, depending on the system, that harshness could arise at 800 Hz, or higher (and maybe even lower). Since we're squarely in the range of fundamentals (not to mention the human voice -- to which evolution has tuned our auditory system in toto), it can all matter (quite a bit) in the perceived quality of reproduction.

I hope it's clear that I don't mean to speak for @fpitas, but I suspect he might agree.

 

[fpitas]

In all seriousness (realizing that this might be a rabbit hole that this thread doesn't need) -- some of us, @fpitas included, live in a loudspeaker world where the interactions of a compression driver, its acoustic transformer

Spoiler

horn -- or "waveguide"

, the transition from driver to said transformer (the "throat"), and the air in the room are critical to the wavefront that emerges from that transformer. All such devices have a cutoff frequency, and behavior in the vicinity of that cutoff can get... interesting.

In other words, depending on the system, that harshness could arise at 800 Hz, or higher (and maybe even lower). Since we're squarely in the range of fundamentals (not to mention the human voice -- to which evolution has tuned our auditory system in toto), it can all matter (quite a bit) in the perceived quality of reproduction.

I hope it's clear that I don't mean to speak for @fpitas, but I suspect he might agree.

Horns are skill-level 5. Agreed.

 

[Sokel]

peaking in the 3-4 kHz region and sibiliant distorsion

That's not only harsh,that's painful,or to be precise can cause pain (round headache) specially when combined with thin sounded midbass.
Has nothing to do with taste,it's physical.