After Neumann KH 310A Review: What to buy?

[amirm]

@amirm can you review Kii Three??

I don't have one currently to test.

 

[DSJR]

The BBC ‘types’ aren’t that bad, obviously limited bass response and just that bit ‘thrummy’ compared to more modern designs.
Properly integrate a couple of subs and they still wouldn’t be as good but listenable.
Keith

Keith, the best ones are changing fast... The best maker of this style of box who wants to develop further has cottoned on to and seemingly invested in Klippel, done away with the old ancient 'BBC' roots really apart from the trad box shape and sales are higher than ever, despite the Brexit issues which may or may not have an effect next year. Apparently, a well known designer (if you've been around as long as I have in the UK scene) from one of the first of these 'BBC Inspired' companies has just moved over to the best and most long term successful one (I'm keeping it vague for now). Will be interesting to see how these speakers develop now using decades of design experience, an objective stance and bang up to date testing systems...

 

[DSJR]

This is a general misconception. Accurate loudspeakers like those from Neumann do not alter anything - playback is guaranteed the way the producer intended it to be [unless room acoustics change the tonality, which again is no loudspeakers' fault].

In the end you could always "shape" the sound to your likings via DSP - knowing, that the loudspeakers wouldn't add coloration through excessive distortion.

You're quite right and it's tragic that speakers like these would have to be bought on spec as getting a dem would be difficult and in a wall of speakers, pretty darned useless really in my personal experience. I've owned and loved larger active monitors and miss them dearly (things we do for marriage...) but I still worry about smaller models being a bit too 'up front' in presentation when all I want to do is relax rather than 'analyse.' If my concerns are totally unfounded, then that's great and no way would I want to equalise in a 'nice tone' as that defeats the object completely. I sneakily suspect the 'Batman response shape' is deliberate in some designs though although the resurrected original LS3/5A from ancient times exaggerates this to this day and ends up thumpy, shrieky and fizzy all at once

 

[Frank Dernie]

I wouldn't say "no care given". Based on interviews I've seen with Alan Shaw, he has considered making an inert resonance free cabinet, but decided against it. He believes the "singing" cabinet is superior to a resonance free cabinet.

Nothing is resonance free.
It would have to either have zero mass or infinite stiffness.
All cabinets resonate at some frequency(s). Harbeth still follow the 1960s research that showed tuning the "cabinet talk" to a lower frequency and damping it much more than usual made it less audible.
This is a perfectly valid approach still today.
Yes making a much more rigid cabinet is easily done, but the "cabinet talk" goes up into a region where the ear is more sensitive.
Part of the difficulty in assesing whether the cabinet approach is better or worse is that the measurement systems we have are incapable of separating the sound radiating from the drivers from that from the cabinet so the data on the effect has to be calculated. I know designers who do this but their data is commercially confidential so not in the public domain.
This means informed discussion on the subject is not feasible and so there are a lot of variably technically dubious theories floating about.

 

[Frank Dernie]

A "singing" cabinet is a musical instrument!

Yes, and actually a thin walled aluminium cabinet will probably sing loudly, like a bell in fact, unless cunningly damped.

 

[q3cpma]

Nothing is resonance free.
It would have to either have zero mass or infinite stiffness.
All cabinets resonate at some frequency(s). Harbeth still follow the 1960s research that showed tuning the "cabinet talk" to a lower frequency and damping it much more than usual made it less audible.
This is a perfectly valid approach still today.
Yes making a much more rigid cabinet is easily done, but the "cabinet talk" goes up into a region where the ear is more sensitive.
Part of the difficulty in assesing whether the cabinet approach is better or worse is that the measurement systems we have are incapable of separating the sound radiating from the drivers from that from the cabinet so the data on the effect has to be calculated. I know designers who do this but their data is commercially confidential so not in the public domain.
This means informed discussion on the subject is not feasible and so there are a lot of variably technically dubious theories floating about.

If you shift the box vibration frequency up, amplitude should diminish (until it is well masked by the direct sound), as more energy is to be found in a short than long wavelength of equivalent amplitude; so at equal dissipated energy, a higher frequency will be of lower amplitude. Honestly, while you're right to be wary of "new better than old", this defense looks tainted by wishful thinking.
On the more "empirical" side, I've never seen anyone say they can hear a Genelec or Neumann cabinet, something that has obviously happened for BBC style thin walls.
https://www.audioholics.com/loudspeaker-design/detailed-look-proper-loudspeaker-cabinet-bracing is a good article on the subject

Anyway, wouldn't a spectrogram show such a resonance?

looks immaculate to me.
This is actually important because I don't think the audibility of such time domain aberrations is fully settled, cf
https://www.audiosciencereview.com/...r-30-speaker-review.11108/page-19#post-318386

 

[DSJR]

You know, thin-wall cabs (BeeBeeCee stylee) are very different in the current Harbeth (and actually, my Rogers LS5/9 implentation) to my old beloved (sentimental reasons) Spendor BC2's from 1974. the Spendors 'talk' a heck of a lot at moderate volume levels, the vibrations very easy to feel on all sides of the boxes - one reason I suspect why the lower bass was generally awful on these, although the domestic replacement SP1 mk1 is rather better here. Both the Rogers and 2007 issue Harbeths I have, barely vibrate at all at normal listening levels and maybe the resonant frequency is slightly higher too (Harbeth use thinner MDF cabs now I believe).

Frank, just reading your posts and replies has given me much thought and fresh learning as regards dealing with vibration, especially your recounting early times at Garrard in Swindon (if I have it right). My 'conditioning' was to make everything as rigid as possible where I believe you suggested dissipating said vibrational energy as sensitively as possible.

 

[Frank Dernie]

You know, thin-wall cabs (BeeBeeCee stylee) are very different in the current Harbeth (and actually, my Rogers LS5/9 implentation) to my old beloved (sentimental reasons) Spendor BC2's from 1974. the Spendors 'talk' a heck of a lot at moderate volume levels, the vibrations very easy to feel on all sides of the boxes - one reason I suspect why the lower bass was generally awful on these, although the domestic replacement SP1 mk1 is rather better here. Both the Rogers and 2007 issue Harbeths I have, barely vibrate at all at normal listening levels and maybe the resonant frequency is slightly higher too (Harbeth use thinner MDF cabs now I believe).

Frank, just reading your posts and replies has given me much thought and fresh learning as regards dealing with vibration, especially your recounting early times at Garrard in Swindon (if I have it right). My 'conditioning' was to make everything as rigid as possible where I believe you suggested dissipating said vibrational energy as sensitively as possible.

Good engineering is all about accurately assessing what is important and setting good goals before starting IME.
At the beginning of my career it was crucial since one never had resources to do everything and the most success came from this assessment, I could go on...
In the case of a record player it is what comes out of the cartridge that is important and making sure spurious signal is minimised. A good performance can be obtained quite cheaply by clever engineering, and that means understanding it is a dynamic system and not using static thinking.
On racing cars it is keeping the tyres at the optimum temperature for the maximum amount of the grip-limited time as possible, and having a suitable aero "map" of forces versus height and steering angle.

 

[Purité Audio]

Keith, the best ones are changing fast... The best maker of this style of box who wants to develop further has cottoned on to and seemingly invested in Klippel, done away with the old ancient 'BBC' roots really apart from the trad box shape and sales are higher than ever, despite the Brexit issues which may or may not have an effect next year. Apparently, a well known designer (if you've been around as long as I have in the UK scene) from one of the first of these 'BBC Inspired' companies has just moved over to the best and most long term successful one (I'm keeping it vague for now). Will be interesting to see how these speakers develop now using decades of design experience, an objective stance and bang up to date testing systems...

I don’t believe it is solely box, the colouration is most likely due to the poor off-axis performance.
Keith

 

[tuga]

I'm late!
Here is a comparison of the 0° and 30° frequency responses of the LS5/9 and the KH310.
The FR of the LS5/9 are made by Stereoplay (4-2015, measurements are strongly smoothed) and those of the KH310 are the ASR measurements, once without smoothing and below with 1/3oct smoothing, which is a bit fairer.
View attachment 97229
What is immediately noticeable is the very early sloping high frequency range of the LS5/9. I don't know how old @ichonderoga is, but a high frequency range sloping down from 9kHz should still be perceivable by older listeners, at least in comparison (compared to a loudspeaker which, for example, decreases from 15kHz).

Actually, the Audax 35mm dome tweeter reaches 20kHz relatively easily - so the stereoplay measurement of the LS5/9 is a mystery or deliberate crossover tuning by GrahamAudio.
Source: audax-speaker.de
View attachment 97235

All LS5/9 measurements I've seen show the roll-off above 10kHz:

 

[tuga]

If you shift the box vibration frequency up, amplitude should diminish (until it is well masked by the direct sound), as more energy is to be found in a short than long wavelength of equivalent amplitude; so at equal dissipated energy, a higher frequency will be of lower amplitude. Honestly, while you're right to be wary of "new better than old", this defense looks tainted by wishful thinking.
On the more "empirical" side, I've never seen anyone say they can hear a Genelec or Neumann cabinet, something that has obviously happened for BBC style thin walls.
https://www.audioholics.com/loudspeaker-design/detailed-look-proper-loudspeaker-cabinet-bracing is a good article on the subject

Anyway, wouldn't a spectrogram show such a resonance?

looks immaculate to me.
This is actually important because I don't think the audibility of such time domain aberrations is fully settled, cf
https://www.audiosciencereview.com/...r-30-speaker-review.11108/page-19#post-318386

I think it's possible that cabinet resonances might be audible but only at high SPLs.

Both the Super HL5+ and the M40.1 are large as standmounts go but produce very negligible harmonic distortion below 90dB:

The Kii Three for comparison:

Many people believe that BBC cabinets are designed to resonate audibly although the research paper will attest to the contrary.
Perhaps they believe that they hear the cabinets on sighted listening of what because they've read?

And isn't harmonic distortion thought to be less audible as frequency gets lower?

Amir didn't mention audible cabintet resonances:

Someone asked me if I had yet listened to the speaker. I have now. I hooked it up to my main system which means no ability to switch between speakers as I have setup in near-field setting at my workstation. Playing my reference quality track (read: not necessarily good for speaker evaluation) the sound is very good. It certainly belies the plain look of the speakers. SPL handling is also quite good. I can turn up as loud as I can tolerate (500 to 1000 watt on tap from amplifier) and I can't hear any distortion. And this is in a very large space with a single speaker player.

The highs seem to be a bit limited and the bass a bit exaggerated. The latter is of course determined by the placement of the speaker and my seating position.

The real test will be a comparison against another speaker which I cannot do at the moment. So as casual observations go, this is an enjoyable speaker.

 

[ctrl]

All LS5/9 measurements I've seen show the roll-off above 10kHz

Whether the roll-off is at 9 or 10kHz is not really important, more important is that practically all measurements show that at 15kHz the sound pressure is reduced by 6-8dB.

I'm usually not a nitpicker when it comes to the super high frequency range, insisting that the full sound level must be maintained up to 20kHz, but 6-8dB sound pressure drop at 15kHz is a bit severe (and should be audible especially with cymbals).

 

[tuga]

Whether the roll-off is at 9 or 10kHz is not really important, more important is that practically all measurements show that at 15kHz the sound pressure is reduced by 6-8dB.

I'm usually not a nitpicker when it comes to the super high frequency range, insisting that the full sound level must be maintained up to 20kHz, but 6-8dB sound pressure drop at 15kHz is a bit severe (and should be audible especially with cymbals).

I wasn't correcting you regarding the exact frequency; I could have easily said 8, or 9.5k...

I've owned speakers, actually designed by the same engineer as the GA LS5/9s, which also rolled-off the top anechoically (they used the Scan Speak D2010) and in my experience the top octave will affect mostly the sense of "air" but not so much the timbre of cymbals.

 

[Frank Dernie]

Many people believe that BBC cabinets are designed to resonate audibly although the research paper will attest to the contrary.
Perhaps they believe that they hear the cabinets on sighted listening of what because they've read?

The BBC cabinets are designed using the same engineering approach as loudspeaker drive units, or at least using the sort of materials most are made from.
The breakup of cone is damped so it is a broad peak that isn't noticeable. The BBC cabinet is approached the same way, the resonance is put where the ear is less sensitive and heavily damped so it won't be noticed.
The cabinet of a BBC type speaker is no more likely to add colour than a speaker using doped paper coned drivers, for example.

 

[q3cpma]

I think it's possible that cabinet resonances might be audible but only at high SPLs.

Both the Super HL5+ and the M40.1 are large as standmounts go but produce very negligible harmonic distortion below 90dB:

The Kii Three for comparison:

Many people believe that BBC cabinets are designed to resonate audibly although the research paper will attest to the contrary.
Perhaps they believe that they hear the cabinets on sighted listening of what because they've read?

And isn't harmonic distortion thought to be less audible as frequency gets lower?

Amir didn't mention audible cabintet resonances:

I'll be honest, I don't trust the measurements from Stereoplay very much as they're very small and smoothed; do you have any information on their setup/methods, by the way? But that's really not to go against your point, as I agree that cabinet resonances aren't very audible in general. Still, who knows if those THD spikes around 300 Hz for the SHL5+ and 220 Hz for the 40.2 (Stereophile shows the panel mess being at 100-200 Hz for the M40.1) aren't due to that? I mean, these are a bit strange, as their proprietary woofer is supposedly a selling point.
Anyway, my point isn't that they're bad on a sound quality level because of this, but that the engineering argument is fallacious and very possibly marketing in disguise; in the 21th century, of course.

And there's of course the unease of having a very messed-up time domain result without enough research to completely disregard it.

 

[tuga]

I'll be honest, I don't trust the measurements from Stereoplay very much as they're very small and smoothed; do you have any information on their setup/methods, by the way? But that's really not to go against your point, as I agree that cabinet resonances aren't very audible in general. Still, who knows if those THD spikes around 300 Hz for the SHL5+ and 220 Hz for the 40.2 (Stereophile shows the panel mess being at 100-200 Hz for the M40.1) aren't due to that? I mean, these are a bit strange, as their proprietary woofer is supposedly a selling point.
Anyway, my point isn't that they're bad on a sound quality level because of this, but that the engineering argument is fallacious and very possibly marketing in disguise; in the 21th century, of course.

And there's of course the unease of having a very messed-up time domain result without enough research to completely disregard it.

The engineering that you refer to has remained unchanged for 43 years, and the design brief was for easily available materials, robustness and low cost.
I agree that they're no longer good value for money but these modern iterations perform really well if you take the age of the design into consideration.

Factors in the design of loudspeaker cabinets
H.D. Harwood, R. Matthews - January 1977


The mechanical properties of timber, wood products and other materials potentially suitable for the construction of loudspeaker cabinets have been measured and details of the results are given. Various commercially available damping materials have also been assessed and their relative efficiencies are listed. A new method of test for the cabinets of completed loudspeakers has been devised and a tentative performance specification has been produced.

(...)

11. Conclusions
Extensive measurements have been made on the elastic modulus of wood and wood products over the whole of the audio band and show a previously unreported dependance on frequency. This information considerably extends that which was previously available and shows that plain timber is an unreliable material because of its variability and its inherent anisotropy. Manufactured products such as ply-wood and chipboard are more consistent and nearly isotropic and are therefore potentially suitable for use in loudspeaker cabinets. Of the two, plywood is preferred because it is manufactured in a greater range of thicknesses and has less resonant modes for a given configuration than chipboard. Although plywood appears to be the most suitable material at present, glass reinforced plastics should not be ignored as they may become competitive on a cost basis, their elastic properties are suitable and they can be moulded into any desired shape.
Damping needs to be applied to the inner surfaces of the panels of a loudspeaker cabinet; this report gives quantitative information about many damping materials. Common roofing felt which is easily available and is manu-factured to satisfy a British Standard Specification seems to rank highly in the list of suitable materials. A self adhesive material such as Bostik is also suitable and cuts down the manufacturing costs. Mutacell with a mild steel backing is good but probably too expensive when labour costs are included.
A provisional specification has been proposed for a simple test method which enables the radiation from panels to be compared with that from the loudspeaker unit. Provisional figures are quoted (Fig. 12) which show the permissible relative levels over the audio frequency band.

http://downloads.bbc.co.uk/rd/pubs/reports/1977-03.pdf


I wonder if the THD in Soundstage's measurements can be linked to Amir's measurements (unfortunately no THD plots of the M30 were produced by ASR)?

THD+N @ 90dB, 50Hz - 10kHz (measured @ 2m)

THD+N @ 95dB, 50Hz - 10kHz (measured @ 2m)

http://www.soundstagenetwork.com/measurements/speakers/harbeth_30_domestic/

 

[richard12511]

Nothing is resonance free.
It would have to either have zero mass or infinite stiffness.
All cabinets resonate at some frequency(s). Harbeth still follow the 1960s research that showed tuning the "cabinet talk" to a lower frequency and damping it much more than usual made it less audible.
This is a perfectly valid approach still today.
Yes making a much more rigid cabinet is easily done, but the "cabinet talk" goes up into a region where the ear is more sensitive.
Part of the difficulty in assesing whether the cabinet approach is better or worse is that the measurement systems we have are incapable of separating the sound radiating from the drivers from that from the cabinet so the data on the effect has to be calculated. I know designers who do this but their data is commercially confidential so not in the public domain.
This means informed discussion on the subject is not feasible and so there are a lot of variably technically dubious theories floating about.

I think you might be missing my point a bit. I was replying to a post claiming that the Harbeths had been designed with "no care" for cabinet resonance. I was merely pointing out that this is untrue(though @q3cpma did raise an interesting question in response). Harbeth does care about resonance, they're just using a different technique(based on different research) to try and solve it.

As for your points, I do agree with some of your opinions, though I do have slightly different opinions myself:

I agree that nothing is resonance free, but I don't think that makes aiming for it as a goal a bad thing.
I disagree that it's not feasible to discuss the validity of the differing approaches. I think it's fairly clear that the BBC thin wall approach is an inferior approach to speaker design at this point. Possibly one reason why many once SOTA British designs seem to be falling behind their German, Finnish, Swiss, and even American contemporaries. Another reason might be that a lot of them also seem to be behind the curve with respect to waveguide tech. KEF is a wonderful exception, made in the Maidstone, UK. I'm sure the BBC research was great for it's time, but that was a long time ago, and I feel that it's been superseded by the Harman/Genelec/KEF/Neumann research.

 

[ichonderoga]

omg I should have known what I started *heh* but immensely grateful at the insights this thread is giving me. I've heard different models of Grahams and Harbeths - on several occasions over the last five years. At tradeshows, stereoshops and at home - and it's now that I realize why EQ'ing these is necessary if I don't want to miss the upper ranges.

 

[DSJR]

I don’t believe it is solely box, the colouration is most likely due to the poor off-axis performance.
Keith

We don't fully know what changes - if any - have been made in recent designs though. Does a reproduced voice sound like a voice, or a mechanical sounding noise - I exaggerate, but it's amazing how many established audiophool speaker confections can't get this one most important thing right and dispersion be damned here

 

[DSJR]

The classic 5/9 was domestically reviewed in HiFi News at the time and back then, the lower midband and tweeter level were broadly level, showing a pronounced dip in the upper hundred Hertz region. The M30 turned this 'smiley response' into a more gentle down-tilt as measured here and elsewhere and the 30 served to flatten this somewhat, which it does. The Graham 5/9 as tested in HiFi World using a low res plot system, showed the response of these is more like the 30 as tested here with the tweeter set slightly lower perhaps than the Rogers originals, the 3khz peak showing a little more and I feel from some experience that this will be heard either as a tubby mid bass-lower midrange, or a slightly laid back tone but with a stark 'treble' register. I have jpegs gathered from the HUG on a few other original 5/9 samples tested in various facilities and they match pretty well the overall response shape - approx. 3 - 4db response suckout from 500 to 3khz approx, a peak at 11khz or so followed by the smooth but pronounced death dive of the tweeter. It looks as if the re-manufactured tweeter is slightly smoother though than the originals used, but mauybe setting them slightly lower in level helps this illusion?

As for the post 13kHz rolling away, this I believe is a factor of the tweeter used (a once high power model I gather and durable in pro use). Remember, FM radio is brick-walled in the UK at 15khz anyway (not sure about our dire DAB system but it can't be any better), so no need for a wideband tweeter for basic monitoring in editing suites as these were used for.