ernestcarl
Major Contributor
I've had these two speakers for a very long time, but I've never really gone into a detailed comparison between these two side-by-side before.
This exercise is going to be divided into sections, so just skip ahead to the areas of interest to you.
Also, I'm going to be focusing on the Sceptres primarily as the Neumann are already discussed ad nauseam in ASR and elsewhere.
EQUALIZING THE SCEPTRES:
I bought a store demo pair of the S8 studio monitors at a ($200?) discount and found out that they're responses were not matching as well as one would have hoped. It's difficult to determine exactly the origin of this issue now, but my guess is that it is likely coming from driver factory variances where tolerances are simply not kept as high as with other more expensive models. Bare in mind, these drivers also seem to be a bit more complicated to build. Fortunately, the differences are not so severe that it cannot be remedied by a 'simple' A/B trace arithmetic function within REW.
The theory is this: you take two measurements from each transducer and equalize the differences between them. The traces need to be accurate enough and taken from nearfield data in order to do this. Ideally you should refer to to multiple directional curves or take an average of relevant curves from each channel. Confirm the effectiveness of the correction by measuring your speakers response in the MLP afterwards, as well as other seats/angles/distances. If you are short in time, you can also just use single on-axis traces, but this may not be as accurate.
REW may have difficulty generating the exact PEQ corrections you want so feel free making manual adjustments as needed.
Once you've ascertained that your speakers are "pair-matched", you can then go about the more complicated task of speaker EQ "correction" or optimization.
This topic can be rather convoluted (and I'm certainly no expert) so this is only going to be a simplification. You would need to refer to the directional curves of your speakers or to the specific traces found in the spinorama data. The axial response, LW, PIR etc. If you don't have that, then you'll need to generate your own directivity curves for which you will refer to during equalization.
*Seeing how the EQ works on my 'pseudo-listening window' curve:
The Sceptres are coaxial in design so it's relatively easy to obtain horizontal and vertical traces required for this exercise. I don't have a proper measurement rig and so I just simply used a small angled ruler I had already readily on hand for this purpose. Of course, you will need a calibrated measurement mic, and a stand to hold this as well.
The ff. are the directivity curves I measured with predictive corrections applied:
LEFT SPEAKER vs RIGHT SPEAKER (post pair-matching EQ and speaker 'optimizing' EQ)
Separated for clarity
LEFT SPEAKER vs RIGHT SPEAKER (post pair-matching EQ and speaker 'optimizing' EQ)
There's a bit of a suck-out around 2kHz and a small hump at 3kHz -- the latter adds to that in-your-face or forward vocal character of the monitors.
Now notice how many of the curves bunch-up closely between 5-10kHz, while I do like how the presence and brightness region of these speakers remains very well sustained off-axis, I'd just like to see that area little bit shelved down to tame the excesses I hear with sibilant and 'splashy' sounds in the HF which can get a little bit annoying over time.
Below is my most current (simplified) general speaker EQ correction to date:
Again, you have to confirm that the corrections work as intended. I used single-point as well as MMM technique at the MLP, including both the further left and right side of my listening couch.
It's just a left+right sum in the above graphic, but the per channel outcome yields very similar curve shapes just as well. Consistency at all seats!
So what the hell is the point with all this equalizing before the listening test even anyhow!? Well, I really wanted their responses to be both on par and as neutral as can be to my own ears so we can rule out any gross axial response differences as the cause of the variances in what I will be later hearing. Careful speaker EQ correction together with my pair-matching correction helps to 'optimize' the final response to better match the Neumann KH120s (applied -1dB HF shelving filter) which will prove very useful during my side-by-side mono listening session.
BTW, the correction we are using does not drastically alter the original curve. No, we don't want to apply any overly excessive correction.
SIDE-BY-SIDE LISTENING TEST
My listening space is quite small, but I did try to clear out the area so that the speakers are reasonably away from any wall boundaries.
Some very basic tech specifications:
PRESONUS SCEPTRE S8
Frequency Response (-3 dB) : 46 Hz - 20 kHz
Frequency Response (-10 dB) : 38 Hz - 23 kHz
Crossover Frequency : 2.4 kHz
LF Amplifier Power : 90W class D
HF Amplifier Power : 90W class D
Maximum Continuous SPL (with music, typical conditions?) : 105 dB (from archived 'brochure') but realistically you want to keep the ‘continuous SPL’ at 100 dB due to increasing harmonic distortion
Maximum Short-Term SPL : 110 dB (my guesstimate - but it should be frequency dependent too)
Peak SPL (transients) : 116 dB (from official manual)
LF Driver : 8" glass-reinforced paper
HF Driver : 1" horn-loaded titanium diaphragm compression driver
Horn Beamwidth Control Limits : 110° x 90° more sharply angled and no cont. transition to baffle (to bypass intermodulation distortion issue plaguing many other coax designs as it "keeps keeps HF energy off of the woofer cone.")
NEUMANN KH120
Frequency Response (-3 dB) : 52 Hz - 21 kHz
Frequency Response (-10 dB) : 46 Hz - 24 kHz
Crossover Frequency : 2.0 kHz
LF Amplifier Power : 80W peak class AB
HF Amplifier Power : 80W peak class AB
Maximum Continuous SPL (with music, typical conditions) : ~95 dB (my guesstimate OR approximation from their tech specs)
Bass capability: Max. SPL half space 3% THD 1m : 104.8 dB (official tech spec)
Peak SPL (transients) : 111.1 dB (very much frequency dependent, of course)
LF Driver : 5.25" sandwich cone composite material
HF Driver : 1" titanium fabric dome
Waveguide Dispersion : not so sure, but it should be wider due to very smooth waveguide seamlessly transitioning to the rest of baffle
I use the ff. manual channel mixing/routing settings to create a mono signal and switching between monitors with JRiver:
As for my own physical positioning: I was mainly seated just over a meter to a meter and a half away mid-point both monitors. I stood up as well for extended amount of time, and listened at extreme opposite angles too (both seated and standing/walking around a bunch).
I spent well over six to seven hours listening -- by the end of the long session, I was fatigued as hell and had to take a break!
The following are my notes and conclusion:
Presonus Sceptre S8
Neumann KH120
CONCLUSION AFTER SIDE-BY-SIDE LISTENING SESSION:
It's ironic, but I actually feel a little bit sad because of how "good"/"realistic" and... surprisingly bright and extended HF transients sound on the (BTW, cheaper) S8 monitors. Having just a very flat and smooth frequency response and impressive spinorama graphs unfortunately is not enough reason for me to choose another speaker when it's finally time to upgrade or replace these. As I, and many others have frequently remarked, there are always, ALWAYS other considerations... that is, if you're the meticulous type when it comes to these esoteric speaker design matters. There are a number of times where I may have unfairly judged the Sceptres due to their obvious deficiency in frequency response evenness here and there. Not to mention the driver to driver consistency (frequency response matching) seems nowhere near as good it should be -- even cheaper speakers like the JBL LSRs or Mackie MR series (the few samples of speakers I've measured) seem to perform better in that regard, at least.
However, don't be so alarmed at the lobing behaviour... it's actually much more benign than you might think. This is greatly offset by the superior performance for off-axis listening vs a conventional speaker even with completely zero horizontal lobing -- though most of these conventional designs are also not too great in their vertical coherency anyway. The unevenness in the frequency response and (possible) pair-matching inconsistency can also be improved upon greatly if one is willing to make the effort -- that is, if you're comfortable working with a measurement microphone, REW, and with DSP correction. Overall, despite the listed disadvantages, I think it may still well be worth effort to try/audition these out. People like to move -- C'mon! We even shift our bodies all of the time when sitting on the same chair or couch! Don't you ever pause for a moment to stretch your bodies, arms, and legs? The advantages of this point source behaviour in a speaker is just so obvious, esp. in nearfield listening where you might still find yourself needing to move around a lot, despite one's supposedly fixed listening position.
I've asked the question before in ASR if there is a point in applying further phase correction via something like rePhase (FIR designer, or the like) and convolution to an already time optimized speaker design -- maybe there are further refinements that can be made? Well, perhaps there's always something that can be improved on -- as the few time-domain obsessed guys here have so confidently asserted. But unlike basic modification of the frequency response, I'm somewhat more skeptical about altering what already came out of the box in what essentially are supposedly "finished systems" -- even more-so in the time-domain given I don't exactly have the tools and extensive expertise or know-how the original speaker and DSP designer(s) have. Okay, maybe I'm just a worry-wart; maybe as long as it does not deviate so much from the original response it'll be fine? Hmmmn... I dunno ... although, I would rather prefer not have any more added processing delay time if I can help it -- so there's that consideration.
It looks like the S8 is currently $100 cheaper than the KH120s in many online retail stores... but despite it what you get is a bigger 8" woofer with better dynamic-range and SPL peak capacity. Which means these monitors can handle very "LOUD" when called for -- no LEDs flashing alarmingly bright red -- Warning! Warning! Back Down the Volume Now! -- like with the smaller Neumann KH120 each time limiters are activated... I've said this before, and I'll say it again, it's a great "value" alternative to much more expensive model brands of speakers, esp. when utilized home for theater use (don't forget the off-axis advantage) -- IF! Again, IF, you're not specifically requiring "far-field" performant speakers. Hmmmn... I gather, in most use-case-scenarios, they're still going to be much more suited to small or medium-sized rooms anyhow.
@watchnerd mentioned a problem with translation... I think part of this can be explained by the fact that most speakers in the market are simply not phase-linear (mids to treble) -- and esp. at all directions! These do sound very different from most of the conventional speakers we often hear out there in the open, there's no doubt about it. But you have to hear it for yourself in person using test tracks that particularly highlight these monitors' design strengths. You also need to compare it directly side-by-side to a conventional state-of-the-art two-way or three-way design to find out what are the details it can resolve more clearly that more conventional designs simply cannot -- most obvious of which are very fast, very dense transients in the HF. It almost makes you wonder if there's something actually broken or wrong with the Neumann KH120s. It's a bit of an annoyance because I also own and use the KH120s almost everyday for the past, what? Almost ten years, now?
*I mentioned a specific test track sample with lots of dense transients and this is the one: Foc by Rodrigo y Gabriela
JRiver's audio analysis had showed the ff. report:
Confirmation with another VST analyzer:
This is before any DSP and volume level processing
More subsequent graphs, and measurements to follow.
In the meantime, the ff. are some informational details provided by Presonus for the Sceptres (which I've toned down quite a bit -- the language was full of marketing hype and filler words).
There's an interview with the designer, Dave Gunness in Nathan Lively's podcast. He doesn't talk about the Sceptres which are being produced by Presonus specifically, but I found it extremely interesting. It allowed me to better my own EQ methodology using some of his ideas -- but more importantly, his own remarks about the "brightening" effect after phase linearization (at least of his own coax speaker designs) was indeed very much confirmed in my own recent listening test above. It's also where I got the idea of adding Jennifer Warnes' song Somewhere, Somebody to my critical speaker listening tracks list. Not that I should know how a Fender bass guitar actually sounds like in real life, LOL!
Here's the YT media link in case someone is interested:
WARNING: It's about an hour-long, so if you'd rather prefer to skip that one, the ff. is a transcript of some things he had said which I copied on my phone (simplified, clarified and abbreviated).
----------------------------
Nathan asks him how it's possible to have "higher gain before feedback" about his speaker designs (which are mostly used for live sound applications).
Where Dave mentions a couple of important points:
FF. discussion more about how they "voice" their speakers:
Primarily it allows us to make all the speakers sound the same. Killing all the resonances makes the voicing of speakers much easier. Typically you are trying to overcome resonances in one speaker that doesn’t exist in another, and it’s futile to try to make those two sound the same if one has a resonance the other one does not have.
You can’t really make them sound the same. You are just mitigating problems in each of them until they both sound good. But if you switch between them, you are still going to hear differences. TQ allows us to have everything in live sound to sound very, very similar. The smallest speaker and the biggest speaker we make have very similar low frequency extension — so they actually have the same frequency response. If you are far enough away from the bigger model, they sound about the same. It’s just that one gets 14dB(?) louder than the other. They have similar character so that it’s a very seamless transition between those speakers. Having an entire line-up of speakers that sound like they’re voiced by the same person saves a tremendous amount of time during commissioning... You, of course, mitigate bass problems caused by placement and the room. But what you are left with 500Hz and up doesn’t need to be mitigated with... in the end the system has a much more consistent character as you work around the room. Listeners are less conscious as they walk around the room. They’re not gonna look up and say, “I’m gonna walk up over there because I don’t like the sound of the speaker where I’m standing by.”
The voicing is just flat through the mids. With displaced systems where they are not coaxial, it’s usually beneficial to have a dip where they have the crossover because the HP on the HF and the LP on the LF are both resonant — and they’re both sitting right next to crossovers with all kinds of resonances. These resonances call attention to themselves. If you have a 15kHz xo and you dial it flat, you are going to hear a 15kHz character that does not belong there — you are hearing the resonant character. What typically happens is eventually there’s a dip at crossover because when people are EQing by ear they are coming back to that 15kHz by ear and pulling that back down — so evetually you have a dip at xo. With coaxial systems you don’t have that issue because you don’t have the change in directivity as you’re going through the crossover. We also use extremely low Q crossovers where we have sometimes up to two octaves of overlap between the low and high frequencies in order to stabilize the polar pattern between the entire range. As a result, even without the FIR filtering it is very hard to pick out where the xo is. When somebody asks “where’s the xo?” It’s spread over two octaves... With displaced speakers you have to try to get out of one driver and into the other one as quickly as possible because if they’re on at the same time they screw up the directionality.
That is the first thing that anyone would notice — just flat throughout the xo. It sounds very different than what we are used to listening — speakers with displaced drivers xo at 15kHz or so which sounds normal. But when the artifacts associated with that go away: “Wow! This very different!”
The other aspect of the housecurve is the low end and the top end. For the low end, we do a very gradual beginning to roll-off and don’t turn and go down hard until maybe 4 or 5 dB. So it looks a little bit like a butterworth cascaded with a 6dB. That’s the curve that we found sounding most natural and 'even' in terms of notes. If you have a speaker that starts to go away at 80Hz, and you use this particular shape, the notes on the bass guitar will all have about the same weight — the melody will still be there. Whereas if you put a butterworth filter on it where there is a sharp corner, all the notes are going to sound like that corner — all the notes will have too much 80Hz and they all sound kind of the same. When you go to the 70Hz note, it’s weak. When you go back to the 80Hz note it’s strong... so the bass-line wonders around. It’s the technical response, but the reason is aesthetic. This is what makes all bass notes sound like they have equal weight.
At the top end, it will generally roll-off very, very slightly from about 1kHz up — not quite a dB per octave. At 10kHz it may be 2-3dB less than it was at 2kHz. Partly it makes the speaker sound warm and powerful. One of the cardinal rules of EQ is you don’t want a peak to be louder than something below it in frequency. If you have a 102dB peak at 2kHz but it’s running a 100dB from 500Hz to 1kHz, that 2kHz is going to stand out. If the overall level were 2dB lower so that peak matches 500Hz to 1kHz, that 2kHz area will not call attention to itself as much. This gradual decline from 1kHz up is what we find more forgiving of reflections in the space. It keeps everything warm and full.
And also, when you have very coherent high frequencies, they sound louder. People are used to listening to speakers that are nominally flat, but with displaced drivers and without TQ. When you flatten the phase it gets brighter — supressing that just a little bit makes it sound a little bit more like what we are used to... the high frequencies are clearer and more articulate not because they’re louder.
When we are tuning a speaker, there is a very definite house sound we are going for and we know how to describe the departure from that sound.
We don’t EQ by only the axial response, we EQ by looking at the whole family of off-axis curves including the axial.
If there is a bit of a rise near the pattern edge of a given frequency, there needs to be a dip so that you de-emphasize that peak in the response away from the primary axis. It makes it go a lot faster to be able to to look at the response in every direction on the computer screen while you are developing the filters. And then when you turn it on and listen to it you start out a lot closer (to your end goal) than we’re used to in the old days.
We typically do about ten curves in various places of the pattern. The way our proprietary software works is when I tweak an EQ, all ten curves morph at the same time. If I’m going after an off-axis peak to try and lower it, I can see the effect it is having in the on-axis at the same time so you don’t (unwittingly) strike a compromise between those two curves.
For people developing they’re own FIR filters, I will definitely look into multiple directions. When you’re listening there is a tempetation to sit dead straight to a speaker and that is fine if what you are making is a home hi-fi speaker or a studio monitor. But if what you are making is something that is going to address an audience, all the audience members are equally important. You have to walk through the entire pattern and you may have to compromise a little bit with the axial response in order to mitigate a problem near the pattern edge. It’s not how good you can make it in front of house, it’s how 'even' you can make the response everywhere.
This exercise is going to be divided into sections, so just skip ahead to the areas of interest to you.
Also, I'm going to be focusing on the Sceptres primarily as the Neumann are already discussed ad nauseam in ASR and elsewhere.
EQUALIZING THE SCEPTRES:
I bought a store demo pair of the S8 studio monitors at a ($200?) discount and found out that they're responses were not matching as well as one would have hoped. It's difficult to determine exactly the origin of this issue now, but my guess is that it is likely coming from driver factory variances where tolerances are simply not kept as high as with other more expensive models. Bare in mind, these drivers also seem to be a bit more complicated to build. Fortunately, the differences are not so severe that it cannot be remedied by a 'simple' A/B trace arithmetic function within REW.
The theory is this: you take two measurements from each transducer and equalize the differences between them. The traces need to be accurate enough and taken from nearfield data in order to do this. Ideally you should refer to to multiple directional curves or take an average of relevant curves from each channel. Confirm the effectiveness of the correction by measuring your speakers response in the MLP afterwards, as well as other seats/angles/distances. If you are short in time, you can also just use single on-axis traces, but this may not be as accurate.
REW may have difficulty generating the exact PEQ corrections you want so feel free making manual adjustments as needed.
Once you've ascertained that your speakers are "pair-matched", you can then go about the more complicated task of speaker EQ "correction" or optimization.
This topic can be rather convoluted (and I'm certainly no expert) so this is only going to be a simplification. You would need to refer to the directional curves of your speakers or to the specific traces found in the spinorama data. The axial response, LW, PIR etc. If you don't have that, then you'll need to generate your own directivity curves for which you will refer to during equalization.
*Seeing how the EQ works on my 'pseudo-listening window' curve:
The Sceptres are coaxial in design so it's relatively easy to obtain horizontal and vertical traces required for this exercise. I don't have a proper measurement rig and so I just simply used a small angled ruler I had already readily on hand for this purpose. Of course, you will need a calibrated measurement mic, and a stand to hold this as well.
The ff. are the directivity curves I measured with predictive corrections applied:
LEFT SPEAKER vs RIGHT SPEAKER (post pair-matching EQ and speaker 'optimizing' EQ)
Separated for clarity
LEFT SPEAKER vs RIGHT SPEAKER (post pair-matching EQ and speaker 'optimizing' EQ)
There's a bit of a suck-out around 2kHz and a small hump at 3kHz -- the latter adds to that in-your-face or forward vocal character of the monitors.
Now notice how many of the curves bunch-up closely between 5-10kHz, while I do like how the presence and brightness region of these speakers remains very well sustained off-axis, I'd just like to see that area little bit shelved down to tame the excesses I hear with sibilant and 'splashy' sounds in the HF which can get a little bit annoying over time.
Below is my most current (simplified) general speaker EQ correction to date:
Again, you have to confirm that the corrections work as intended. I used single-point as well as MMM technique at the MLP, including both the further left and right side of my listening couch.
It's just a left+right sum in the above graphic, but the per channel outcome yields very similar curve shapes just as well. Consistency at all seats!
So what the hell is the point with all this equalizing before the listening test even anyhow!? Well, I really wanted their responses to be both on par and as neutral as can be to my own ears so we can rule out any gross axial response differences as the cause of the variances in what I will be later hearing. Careful speaker EQ correction together with my pair-matching correction helps to 'optimize' the final response to better match the Neumann KH120s (applied -1dB HF shelving filter) which will prove very useful during my side-by-side mono listening session.
BTW, the correction we are using does not drastically alter the original curve. No, we don't want to apply any overly excessive correction.
SIDE-BY-SIDE LISTENING TEST
My listening space is quite small, but I did try to clear out the area so that the speakers are reasonably away from any wall boundaries.
Some very basic tech specifications:
PRESONUS SCEPTRE S8
Frequency Response (-3 dB) : 46 Hz - 20 kHz
Frequency Response (-10 dB) : 38 Hz - 23 kHz
Crossover Frequency : 2.4 kHz
LF Amplifier Power : 90W class D
HF Amplifier Power : 90W class D
Maximum Continuous SPL (with music, typical conditions?) : 105 dB (from archived 'brochure') but realistically you want to keep the ‘continuous SPL’ at 100 dB due to increasing harmonic distortion
Maximum Short-Term SPL : 110 dB (my guesstimate - but it should be frequency dependent too)
Peak SPL (transients) : 116 dB (from official manual)
LF Driver : 8" glass-reinforced paper
HF Driver : 1" horn-loaded titanium diaphragm compression driver
Horn Beamwidth Control Limits : 110° x 90° more sharply angled and no cont. transition to baffle (to bypass intermodulation distortion issue plaguing many other coax designs as it "keeps keeps HF energy off of the woofer cone.")
NEUMANN KH120
Frequency Response (-3 dB) : 52 Hz - 21 kHz
Frequency Response (-10 dB) : 46 Hz - 24 kHz
Crossover Frequency : 2.0 kHz
LF Amplifier Power : 80W peak class AB
HF Amplifier Power : 80W peak class AB
Maximum Continuous SPL (with music, typical conditions) : ~95 dB (my guesstimate OR approximation from their tech specs)
Bass capability: Max. SPL half space 3% THD 1m : 104.8 dB (official tech spec)
Peak SPL (transients) : 111.1 dB (very much frequency dependent, of course)
LF Driver : 5.25" sandwich cone composite material
HF Driver : 1" titanium fabric dome
Waveguide Dispersion : not so sure, but it should be wider due to very smooth waveguide seamlessly transitioning to the rest of baffle
I use the ff. manual channel mixing/routing settings to create a mono signal and switching between monitors with JRiver:
As for my own physical positioning: I was mainly seated just over a meter to a meter and a half away mid-point both monitors. I stood up as well for extended amount of time, and listened at extreme opposite angles too (both seated and standing/walking around a bunch).
I spent well over six to seven hours listening -- by the end of the long session, I was fatigued as hell and had to take a break!
The following are my notes and conclusion:
Presonus Sceptre S8
- Speaker EQ which attenuates HF from original sound may not impress at first
- Very clear & great clarity, although also less harsh & quite restrained in character
- Can listen to it much longer
- Can listen to it at a higher SPL
- Vocals still sound forward after EQ, BUT also somewhat "rounder" or "warmer"
- Highs, mids, and lows sound more cohesive which is very obvious when bobbing & swaying head and body up, down, left, right -- change in sound or any kind of vocal displacement is far less obvious/disturbing
- FASTER, clearer transients, very obvious with extreme flamenco guitar track tested (also has a brightening effect)
- Quick sliding hand movements across guitar strings sound more real -- to me this is significant! LOL
- Cymbals, bells, and tamborine-like and other 'ringing' brass, string & percussion acoustic instruments sound nicer and are more pleasing
- HF transient peaks do not sound squashed, compressed, or cut-off short
- The excellent transient performance makes music sound "alive"
- Overall quality sounds "rounder" / "softer" / less aggressive -- only partially explained by HF shelving and dips in response
- Much preferred for everyday casual listening & for "finished" tracks
- 'Some' deliberately added electronic distortion effects are more obvious to hear with these monitors
- Port 'chuffing'? or more like cabinet resonance/distortion is audible with extreme sub bass-y electronic tracks -- but only at extreme volume levels where the KH120's can't even reach
Neumann KH120
- Tonality for voices is preferred with these
- Clearer it seems, or at least more deliberately obvious accentuation of details
- Esp. high treble frequencies which is also much more extended
- Easier to separate elements of a mix
- HF extension and treble detail will very much highlight harshness (EXCEPT for harshness that comes from really fast transients which this monitor isn't able to resolve nearly as well)
- More aggressive sounding (i.e. less "soft" / "restrained") in many of the tracks listened to
- My ears tire out quicker/sooner with these monitors
- YET, the monitors sometimes glosses over/misses deliberately added distortion effects in electronic tracks
- Nevertheless, I probably will choose this first for critical listening / or when doing detailed dissection of sound tracks -- IF I'm not particularly fatigued/tired as it sounds perceptually clearer to me to more of the tracks being listened to anyway
- Sitting exaggeratedly straight up or slouching down one can notice sound displacement between HF & mid drivers.
- Bobbing & swaying head and body up, down, left, right -- more apparent change in sound, vocal displacement or some type of shadowing effect can be heard -- actually I've had this problem with them for a very long time as I use a sit-stand desk and I change positions heights a lot. For example, wearing shoes changes the the sound from what I have been used to listening from the several past days, and I find this disturbing enough to finally take-off what I'm wearing or make re-adjustments to the height of my speaker stands. BTW, minute changes to the horizontal angle placement makes bigger apparent changes to the sound (again, my past experience). This is magnified with my sometimes very nearfield desk setup use-case scenario (I can pull both monitors and keyboard closer or nearer as necessary).
- Sounds somewhat "smaller", maybe even just a little less coherent (again, shifting of location source or sound character from driver displacement is magnified with any gross physical movement, esp. the closer you are from the speakers)
- SLOWER transients, very obvious with extreme flamenco guitar track tested (which also has a darkening/dulling/de-brightening effect)
- HF transient peaks sound as though they're being cut-off short or muted -- sadly, this effect is not subtle when switching monitors
- And so it glosses over excessive harshness from fast transients in certain tracks -- BUT, of which can really only be heard clearly on the S8s (believe me, I wish the that observation was not so obviously the case since the frequency response of the KH120 is so incredibly flat and smooth.
- Quite impressive and punchy bass for its size! I think many people will be happy and plenty satisfied even without the presence of a sub.
- Port 'chuffing'? or distortion audible with extreme sub bass-y electronic tracks -- I think it relies too much in it's natural acoustic roll-off which can make it sound more boomy/distorted -- if not sometimes artificially extended than the S8 with its very steep 10th-order roll-off by comparison. I do feel like this may hype the bass quite a bit more than the S8s
CONCLUSION AFTER SIDE-BY-SIDE LISTENING SESSION:
It's ironic, but I actually feel a little bit sad because of how "good"/"realistic" and... surprisingly bright and extended HF transients sound on the (BTW, cheaper) S8 monitors. Having just a very flat and smooth frequency response and impressive spinorama graphs unfortunately is not enough reason for me to choose another speaker when it's finally time to upgrade or replace these. As I, and many others have frequently remarked, there are always, ALWAYS other considerations... that is, if you're the meticulous type when it comes to these esoteric speaker design matters. There are a number of times where I may have unfairly judged the Sceptres due to their obvious deficiency in frequency response evenness here and there. Not to mention the driver to driver consistency (frequency response matching) seems nowhere near as good it should be -- even cheaper speakers like the JBL LSRs or Mackie MR series (the few samples of speakers I've measured) seem to perform better in that regard, at least.
However, don't be so alarmed at the lobing behaviour... it's actually much more benign than you might think. This is greatly offset by the superior performance for off-axis listening vs a conventional speaker even with completely zero horizontal lobing -- though most of these conventional designs are also not too great in their vertical coherency anyway. The unevenness in the frequency response and (possible) pair-matching inconsistency can also be improved upon greatly if one is willing to make the effort -- that is, if you're comfortable working with a measurement microphone, REW, and with DSP correction. Overall, despite the listed disadvantages, I think it may still well be worth effort to try/audition these out. People like to move -- C'mon! We even shift our bodies all of the time when sitting on the same chair or couch! Don't you ever pause for a moment to stretch your bodies, arms, and legs? The advantages of this point source behaviour in a speaker is just so obvious, esp. in nearfield listening where you might still find yourself needing to move around a lot, despite one's supposedly fixed listening position.
I've asked the question before in ASR if there is a point in applying further phase correction via something like rePhase (FIR designer, or the like) and convolution to an already time optimized speaker design -- maybe there are further refinements that can be made? Well, perhaps there's always something that can be improved on -- as the few time-domain obsessed guys here have so confidently asserted. But unlike basic modification of the frequency response, I'm somewhat more skeptical about altering what already came out of the box in what essentially are supposedly "finished systems" -- even more-so in the time-domain given I don't exactly have the tools and extensive expertise or know-how the original speaker and DSP designer(s) have. Okay, maybe I'm just a worry-wart; maybe as long as it does not deviate so much from the original response it'll be fine? Hmmmn... I dunno ... although, I would rather prefer not have any more added processing delay time if I can help it -- so there's that consideration.
It looks like the S8 is currently $100 cheaper than the KH120s in many online retail stores... but despite it what you get is a bigger 8" woofer with better dynamic-range and SPL peak capacity. Which means these monitors can handle very "LOUD" when called for -- no LEDs flashing alarmingly bright red -- Warning! Warning! Back Down the Volume Now! -- like with the smaller Neumann KH120 each time limiters are activated... I've said this before, and I'll say it again, it's a great "value" alternative to much more expensive model brands of speakers, esp. when utilized home for theater use (don't forget the off-axis advantage) -- IF! Again, IF, you're not specifically requiring "far-field" performant speakers. Hmmmn... I gather, in most use-case-scenarios, they're still going to be much more suited to small or medium-sized rooms anyhow.
@watchnerd mentioned a problem with translation... I think part of this can be explained by the fact that most speakers in the market are simply not phase-linear (mids to treble) -- and esp. at all directions! These do sound very different from most of the conventional speakers we often hear out there in the open, there's no doubt about it. But you have to hear it for yourself in person using test tracks that particularly highlight these monitors' design strengths. You also need to compare it directly side-by-side to a conventional state-of-the-art two-way or three-way design to find out what are the details it can resolve more clearly that more conventional designs simply cannot -- most obvious of which are very fast, very dense transients in the HF. It almost makes you wonder if there's something actually broken or wrong with the Neumann KH120s. It's a bit of an annoyance because I also own and use the KH120s almost everyday for the past, what? Almost ten years, now?
*I mentioned a specific test track sample with lots of dense transients and this is the one: Foc by Rodrigo y Gabriela
JRiver's audio analysis had showed the ff. report:
Confirmation with another VST analyzer:
This is before any DSP and volume level processing
More subsequent graphs, and measurements to follow.
In the meantime, the ff. are some informational details provided by Presonus for the Sceptres (which I've toned down quite a bit -- the language was full of marketing hype and filler words).
- The human ear hears most sounds as point sources.
- Even when we trick our brains into perceiving a stereo soundstage, each ear most easily “assembles” the stereo effect from its own point source (think headphones).
- Benefts of coaxials are related to radiation characteristics of a point source and a consistent acoustic center.
- Response behavior is symmetric in both the horizontal and vertical axes.
- Frequency response at a given angle in relation to the axis is “mirrored” at the same angle in the opposite direction.
- Conventional two- and three-way speaker systems can’t achieve this. High and low frequencies originate from different physical points. Bass and treble arrive at the ear at different times — and from slightly different axes. This subtle “smearing” cannot be corrected with digital signal processing.
- Multi-way monitors suffer from the changing relationship between the listener and the speaker elements and cannot possibly have an optimal impulse response over a wide area.
- Crossover transition can theoretically be made inaudible.
- The frequency point where highs and lows cross over can result in an audible dip (or peak) in the response of the monitor.
- A coaxial multi-way speaker is a point source, it has the potential to completely hide the crossover point. Note that we said potential. It’s no slam dunk. And this brings us to why all studio monitors aren’t coaxial:
- Both transducers nestled together on the same axis generate signifcant problems. A horn right in the middle of the low frequency speaker cone can cause diffraction and distortion. Rearward sound from the horn can bounce off the woofer cone and arrive at listener’s ears too late, smearing the stereo image.
- The proximity of horn and woofer can cause the two outputs to experience audible intermodulation distortion.
- Sceptre starts out with a transducer that
- a) solves many basic coaxial problems through its physical confguation and
- b) is also designed so that DSP can be used to correct other issues.
- Resulting in smooth off-axis response.
- The high-frequency horn contributes to frequency pattern control, and keeps HF energy off of the woofer cone.
- The woofer’s larger radiating surface works with the HF horn to improve the bottom operating-range directional control.
- TQ uses:
- Infnite Impulse Response (IIR) highpass, lowpass, and parametric flters, plus delay
- Finite Impulse Response (FIR) flters implement more detailed frequency response adjustments; and precise temporal (time domain) filters
- DSP can eliminate physical horn reflections induced resonances when they happen consistently. Knowing in advance how the loudspeaker will respond to a particular signal, it is possible to calculate a new signal that not only avoids exciting natural resonances, but also actively kills these resonances before they become audible.
There's an interview with the designer, Dave Gunness in Nathan Lively's podcast. He doesn't talk about the Sceptres which are being produced by Presonus specifically, but I found it extremely interesting. It allowed me to better my own EQ methodology using some of his ideas -- but more importantly, his own remarks about the "brightening" effect after phase linearization (at least of his own coax speaker designs) was indeed very much confirmed in my own recent listening test above. It's also where I got the idea of adding Jennifer Warnes' song Somewhere, Somebody to my critical speaker listening tracks list. Not that I should know how a Fender bass guitar actually sounds like in real life, LOL!
Here's the YT media link in case someone is interested:
WARNING: It's about an hour-long, so if you'd rather prefer to skip that one, the ff. is a transcript of some things he had said which I copied on my phone (simplified, clarified and abbreviated).
----------------------------
Nathan asks him how it's possible to have "higher gain before feedback" about his speaker designs (which are mostly used for live sound applications).
Where Dave mentions a couple of important points:
- The response is more consistent over the entire coverage pattern.
- Having a really flat phase response helps with stability.
- Wobbles in phase correspond to resonances.
- Resonances encourage feedback.
- If your’re phase is flat and not ripply, that means you are essentially resonance-free.
- It’s a combination of TQ (proprietary FIR based EQ tech) which flattens phase and the coaxial approach which gives you a very consistent response and it’s noticeably more stable with regard to feedback
FF. discussion more about how they "voice" their speakers:
Primarily it allows us to make all the speakers sound the same. Killing all the resonances makes the voicing of speakers much easier. Typically you are trying to overcome resonances in one speaker that doesn’t exist in another, and it’s futile to try to make those two sound the same if one has a resonance the other one does not have.
You can’t really make them sound the same. You are just mitigating problems in each of them until they both sound good. But if you switch between them, you are still going to hear differences. TQ allows us to have everything in live sound to sound very, very similar. The smallest speaker and the biggest speaker we make have very similar low frequency extension — so they actually have the same frequency response. If you are far enough away from the bigger model, they sound about the same. It’s just that one gets 14dB(?) louder than the other. They have similar character so that it’s a very seamless transition between those speakers. Having an entire line-up of speakers that sound like they’re voiced by the same person saves a tremendous amount of time during commissioning... You, of course, mitigate bass problems caused by placement and the room. But what you are left with 500Hz and up doesn’t need to be mitigated with... in the end the system has a much more consistent character as you work around the room. Listeners are less conscious as they walk around the room. They’re not gonna look up and say, “I’m gonna walk up over there because I don’t like the sound of the speaker where I’m standing by.”
The voicing is just flat through the mids. With displaced systems where they are not coaxial, it’s usually beneficial to have a dip where they have the crossover because the HP on the HF and the LP on the LF are both resonant — and they’re both sitting right next to crossovers with all kinds of resonances. These resonances call attention to themselves. If you have a 15kHz xo and you dial it flat, you are going to hear a 15kHz character that does not belong there — you are hearing the resonant character. What typically happens is eventually there’s a dip at crossover because when people are EQing by ear they are coming back to that 15kHz by ear and pulling that back down — so evetually you have a dip at xo. With coaxial systems you don’t have that issue because you don’t have the change in directivity as you’re going through the crossover. We also use extremely low Q crossovers where we have sometimes up to two octaves of overlap between the low and high frequencies in order to stabilize the polar pattern between the entire range. As a result, even without the FIR filtering it is very hard to pick out where the xo is. When somebody asks “where’s the xo?” It’s spread over two octaves... With displaced speakers you have to try to get out of one driver and into the other one as quickly as possible because if they’re on at the same time they screw up the directionality.
That is the first thing that anyone would notice — just flat throughout the xo. It sounds very different than what we are used to listening — speakers with displaced drivers xo at 15kHz or so which sounds normal. But when the artifacts associated with that go away: “Wow! This very different!”
The other aspect of the housecurve is the low end and the top end. For the low end, we do a very gradual beginning to roll-off and don’t turn and go down hard until maybe 4 or 5 dB. So it looks a little bit like a butterworth cascaded with a 6dB. That’s the curve that we found sounding most natural and 'even' in terms of notes. If you have a speaker that starts to go away at 80Hz, and you use this particular shape, the notes on the bass guitar will all have about the same weight — the melody will still be there. Whereas if you put a butterworth filter on it where there is a sharp corner, all the notes are going to sound like that corner — all the notes will have too much 80Hz and they all sound kind of the same. When you go to the 70Hz note, it’s weak. When you go back to the 80Hz note it’s strong... so the bass-line wonders around. It’s the technical response, but the reason is aesthetic. This is what makes all bass notes sound like they have equal weight.
At the top end, it will generally roll-off very, very slightly from about 1kHz up — not quite a dB per octave. At 10kHz it may be 2-3dB less than it was at 2kHz. Partly it makes the speaker sound warm and powerful. One of the cardinal rules of EQ is you don’t want a peak to be louder than something below it in frequency. If you have a 102dB peak at 2kHz but it’s running a 100dB from 500Hz to 1kHz, that 2kHz is going to stand out. If the overall level were 2dB lower so that peak matches 500Hz to 1kHz, that 2kHz area will not call attention to itself as much. This gradual decline from 1kHz up is what we find more forgiving of reflections in the space. It keeps everything warm and full.
And also, when you have very coherent high frequencies, they sound louder. People are used to listening to speakers that are nominally flat, but with displaced drivers and without TQ. When you flatten the phase it gets brighter — supressing that just a little bit makes it sound a little bit more like what we are used to... the high frequencies are clearer and more articulate not because they’re louder.
When we are tuning a speaker, there is a very definite house sound we are going for and we know how to describe the departure from that sound.
We don’t EQ by only the axial response, we EQ by looking at the whole family of off-axis curves including the axial.
If there is a bit of a rise near the pattern edge of a given frequency, there needs to be a dip so that you de-emphasize that peak in the response away from the primary axis. It makes it go a lot faster to be able to to look at the response in every direction on the computer screen while you are developing the filters. And then when you turn it on and listen to it you start out a lot closer (to your end goal) than we’re used to in the old days.
We typically do about ten curves in various places of the pattern. The way our proprietary software works is when I tweak an EQ, all ten curves morph at the same time. If I’m going after an off-axis peak to try and lower it, I can see the effect it is having in the on-axis at the same time so you don’t (unwittingly) strike a compromise between those two curves.
For people developing they’re own FIR filters, I will definitely look into multiple directions. When you’re listening there is a tempetation to sit dead straight to a speaker and that is fine if what you are making is a home hi-fi speaker or a studio monitor. But if what you are making is something that is going to address an audience, all the audience members are equally important. You have to walk through the entire pattern and you may have to compromise a little bit with the axial response in order to mitigate a problem near the pattern edge. It’s not how good you can make it in front of house, it’s how 'even' you can make the response everywhere.
Attachments
Last edited: