• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Neumann KH 80 DSP Monitor Review

Status
Not open for further replies.

Pio2001

Senior Member
Joined
May 15, 2018
Messages
317
Likes
507
Location
Neuville-sur-Saône, France
Hi,
Just in case, here are some measurements that I made with my JBL 305P mkII and Neumann KH-120 AW.
I took the measurements at tweeter's height, and also at mid-height between the tweeter and the woofer's axis, at a distance of 70 cm, holding the microphone by hand.
The MMM (Moving Micro Measurements) from the listening position are averages over one seat, 2 meters away, in an ordinary room, far from walls, with room eq below 600 Hz, with 0.45 RT60 speakers-to-listening-position at 500 Hz.
The listening position is located 6° above and 10° sideways the speaker's axis.

123_AcousticAxis.png


The darkened area must be ignored. I think that this peak at 350 Hz is a boundary effect of the windowing of the impulse response.

We can see that the choice of the reference axis has little effect of the frequency response, at least at a distance of 70 cm.
For the JBL, the only difference is the little peak at 9500 Hz. For the KH-120, the part from 2500 to 8000 Hz is smoother at the reference axis. Above 8000 Hz, the effect is the opposite of what we expect : highest frequencies are about 1 dB stronger on the reference axis than on the tweeter's axis. I hope I labeled my curves correctly :rolleyes:

We can also see that the result at the listening position, if we neglect the general treble attenuation, is a bit different for the Neumann, while it looks exactly like the anechoic response for the JBL from 1000 to 10000 Hz. This questions the importance of the choice for a reference axis in a non-treated room.
On the other hand, we must remember that the anechoic response is more important than the listening position response for perceived sound quality. And this correlates well with my listening impressions. The Neumann sound more realistic, on human voices, than the JBL.

Last, and this puzzles me a lot more than the reference axis question, at listening position, the oscillations of the anechoic frequency response above 5000 Hz are canceled. But if we look at the spinorama of the JBL, these oscillations exist on all curves, in all directions. It should be impossible to smooth them.
It correlates with my listening impressions : if I equalize the JBL curve given by the spinorama, it works up to 5000 Hz. Above that, if I equalize them, it sounds weird, and the measurement from the listening position shows that the result is grossly incorrect : the correction destroys the curve instead of improving it.
The same phenomenon occurs as soon as 1500 Hz if I make close-field MMM measurements in the early reflection direction or in the sound sound power directions. The permanent oscillations of the frequency response of the speaker all disappear !

Jean-Luc Ohl, a french acoustician, has noticed the same phenomenon while performing measurements in an anechoic chamber: measuring the frequency response moving the microphone around the speaker cancels some oscillations in the frequency response that, according to the spinorama data, can never be canceled.
 
Last edited:

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
Obsessed on details that doesn't mean anything in the grand scheme of things.

Yes, it doesn't mean much at all in the grand scheme of things and the KH 80 appears to be an excellent speaker, regardless of any measurement methodology issues or casual positioning and configuring in the listening test. But don't forget that this web site/forum thrives on measuring and naming new champions precisely because of such small differences. And the KH 80 was a good opportunity to have a good reference.
I learned quite a bit (again) reading all the technical comments and ignoring the unfortunate personal attacks was a minor inconvenience.

Plus, as Amir said himself in the past, this is meant to be infotainment ;)
 

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,717
Likes
2,897
Location
Finland
Well, your MMM measurements shows average of listening window response. It is perhaps the most important measurement, regarding match to audible sound. I use MMM too.

On-axis response is a different case. it is just a single one but a very crucial measurement, also the reference for off-axis responses and caclulations.
Did you notice that MZKM shows the on-axis response in his ranking. That is also the only one what Neumann shows.
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
We can see that the choice of the reference axis has little effect of the frequency response, at least at a distance of 70 cm.

Maybe. Or maybe your measurements are not accurate enough to measure the difference.
 

carlob

Addicted to Fun and Learning
Forum Donor
Joined
Aug 4, 2019
Messages
736
Likes
1,027
Location
Roma, Italy
I learned quite a bit (again) reading all the technical comments

six months from now, with a number of speakers measured, when somebody will ask for details on the measurement methodology are you going to point this person to the Neumann review thread like read the 30 and counting pages and fish for the relevant comments intermixed with other stuff and (grown?) people pulling their hairs because the speaker is not remeasured to agree with the Neumann curves +/- 0.0000000000001 dB?
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
We can see that the vertical division is 5 dB. Not looking at the tiny localized variations, the two curves may be 1 dB apart from 4000 to 10000 Hz, and 2 or 3 dB apart above 10 kHz.

So? You would need to repeat those sweeps at least 10 times to check inter-variation of your measurements.
 

BYRTT

Addicted to Fun and Learning
Forum Donor
Joined
Nov 2, 2018
Messages
956
Likes
2,452
Location
Denmark (Jutland)
Tryed tons of stuff in VituixCAD to simulate a bit lower axis, for example made dual transducers based the same step data in the 72 txt files and crossed them textbook LR 8th order @1.8kHz, tried small uS of delay here and there and also renamed all the txt files for verticals so their steps was offset 10º, nothing seem to work right to smooth on axis or XO dip other than dial on some real EQ where for example listening window is very sensitive to any changes.

Gave up simulate more logic clues for axis deviation but now i had it crossed repaired the XO dip add +1,5dB PEQ @1.8kHz to LF passband and +0,75dB PEQ @1.8kHz to HF passband to simulate microphone axis sat too high and also used PEQ to repair low end for what is percieved a compression error, now case for a real rescan seems closed can share how it looks when also interpolated using VAR smoothing into REW.

Warning nice looking KH 80 below is modelled and interpolated :oops: where rest of them is raw true data output from NFS:
PANTHER_Race_7.png
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Done ;)
10 times at tweeter's height, 10 times at acoustic center's heigh. The vertical scale is 1 dB per division, which means that it is strongly zoomed.

View attachment 47458

Same thing with 1/3 octave smoothing:

View attachment 47459

As KH-120AW is specced to have ±1 dB in the 100Hz-10kHz range resolution of 1dB is actually not enough to confirm the spec.
 

porsche917k

New Member
Joined
Jan 25, 2020
Messages
1
Likes
5
I think these are fabolous little speakers in many ways, but that little woofer can't really handle much SPL even below 1kHz without significant rise in distortion and not even KH750 can save them because it would need to be crossed much higher then 80-100Hz.

I did quick measurements through ADI-2 DAC at +1, +7, +13 and +19. My speakers are set to 94 dB SPL output level and -15 input gain which should equal to 80, 86, 92 and 98 dB @1m. Please ignore absolute SPL levels shown and lower frequency response because i did not bother at all with calibrating rew or positioning both speakers and mic, just wanted to show how distortion rises with volume.

neumann_kh80_01dBu.png

neumann_kh80_07dBu.png

neumann_kh80_13dBu.png

neumann_kh80_17dBu.png


So if you're fine with low to moderate listening volumes as i (enjoying them event at +1, almost never going over +7, very rarely hitting +13 and +19 being just too much for my room and listening distance anyway), i think they are hard to beat considering price, quality at lower volumes, auto on/off (no need to turn then on/off manually which i wouldn't stand peronally) and build in dsp (again no need to worry about the source being a computer if you want to set some software dsp and simply enjoying them through whatever you're currently using). But if high listening volumes is what you're after, or don't care about comfort of use they provide, then it's probably better to look elsewhere even if considering buying a sub.

Something like KH 120 DSP would probably solve everything though...
 

dukanvadet

Member
Forum Donor
Joined
Oct 18, 2019
Messages
72
Likes
114
Tryed tons of stuff in VituixCAD to simulate a bit lower axis, for example made dual transducers based the same step data in the 72 txt files and crossed them textbook LR 8th order @1.8kHz, tried small uS of delay here and there and also renamed all the txt files for verticals so their steps was offset 10º, nothing seem to work right to smooth on axis or XO dip other than dial on some real EQ where for example listening window is very sensitive to any changes.

Gave up simulate more logic clues for axis deviation but now i had it crossed repaired the XO dip add +1,5dB PEQ @1.8kHz to LF passband and +0,75dB PEQ @1.8kHz to HF passband to simulate microphone axis sat too high and also used PEQ to repair low end for what is percieved a compression error, now case for a real rescan seems closed can share how it looks when also interpolated using VAR smoothing into REW.

Warning nice looking KH 80 below is modelled and interpolated :oops: where rest of them is raw true data output from NFS:
View attachment 47457
I'm not sure what you are telling us here? That EQ can change the frequency response on axis to look like neumanns graph? You could do the same thing with the yamaha hs5s.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
I think these are fabolous little speakers in many ways, but that little woofer can't really handle much SPL even below 1kHz without significant rise in distortion and not even KH750 can save them because it would need to be crossed much higher then 80-100Hz.

I did quick measurements through ADI-2 DAC at +1, +7, +13 and +19. My speakers are set to 94 dB SPL output level and -15 input gain which should equal to 80, 86, 92 and 98 dB @1m. Please ignore absolute SPL levels shown and lower frequency response because i did not bother at all with calibrating rew or positioning both speakers and mic, just wanted to show how distortion rises with volume.

View attachment 47453
View attachment 47454
View attachment 47455
View attachment 47456

So if you're fine with low to moderate listening volumes as i (enjoying them event at +1, almost never going over +7, very rarely hitting +13 and +19 being just too much for my room and listening distance anyway), i think they are hard to beat considering price, quality at lower volumes, auto on/off (no need to turn then on/off manually which i wouldn't stand peronally) and build in dsp (again no need to worry about the source being a computer if you want to set some software dsp and simply enjoying them through whatever you're currently using). But if high listening volumes is what you're after, or don't care about comfort of use they provide, then it's probably better to look elsewhere even if considering buying a sub.

Something like KH 120 DSP would probably solve everything though...

Your main issue is not distortion but non linear frequency response below 500Hz.
 

Pio2001

Senior Member
Joined
May 15, 2018
Messages
317
Likes
507
Location
Neuville-sur-Saône, France
Done ;)
10 times at tweeter's height, 10 times at acoustic center's heigh.

The prettier graph is the one at tweeter's height :rolleyes:
Maybe 70 cm is too close to the speaker for the two drivers to merge properly.

As KH-120AW is specced to have ±1 dB in the 100Hz-10kHz range resolution of 1dB is actually not enough to confirm the spec.

With 1/3 octave smoothing, the sets of 10 measurements span +/- 0.15 dB each below 10 kHz, +/- 0.2 dB above 10 kHz.
It won't confirm the specs, though, as the Umik-1 microphone is probably not better than +/- 1 dB itself, plus the fact that the measurements might look different at various distances from the speaker.

However, it is enough to measure the variations between the two references axis. The maximum deviation below 10 kHz occurs at 4000 Hz, where the two sets of 1/3 octave measurements difference peaks at 0.4 dB (0.6 dB maximum if we include the inter-measurement variation).
Above 10 kHz, the maximum 1/3 octave deviation reaches 0.6 dB at 13 kHz (0.9 dB maximum, taking inter-measurement variation into account).

But this is only valid at 70 cm in front of the speaker.
 

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,717
Likes
2,897
Location
Finland
six months from now, with a number of speakers measured, when somebody will ask for details on the measurement methodology are you going to point this person to the Neumann review thread like read the 30 and counting pages and fish for the relevant comments intermixed with other stuff and (grown?) people pulling their hairs because the speaker is not remeasured to agree with the Neumann curves +/- 0.0000000000001 dB?

You obviously don't understand, where the real problem and discussion lies. Not about matching factory response, but the reason for that! We suspect some issues in how Amir installs the DUT and sets up reference 0-axis. If his practice is not as per instructions of Klippel, and varies with each speaker, we loose the validity and comparability of all measurements!

Remember what is the name of this website!
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
With 1/3 octave smoothing, the sets of 10 measurements span +/- 0.15 dB each below 10 kHz, +/- 0.2 dB above 10 kHz.

Applying such heavy smoothing after averaging is a little hm-hm, wouldn't you say?

It won't confirm the specs, though, as the Umik-1 microphone is probably not better than +/- 1 dB itself, plus the fact that the measurements might look different at various distances from the speaker.

It is better than that assuming you applied supplied calibration file. Although it seems not to be a common pratice around here I strongly suggest you do it. ;)
 
Last edited:

carlob

Addicted to Fun and Learning
Forum Donor
Joined
Aug 4, 2019
Messages
736
Likes
1,027
Location
Roma, Italy
At the end of the day this thread is good for one reason, it's a kind of brains trust of the most stubborn people I've seen in my life and you get to ignore them all!
 

BYRTT

Addicted to Fun and Learning
Forum Donor
Joined
Nov 2, 2018
Messages
956
Likes
2,452
Location
Denmark (Jutland)
I'm not sure what you are telling us here? That EQ can change the frequency response on axis to look like neumanns graph? You could do the same thing with the yamaha hs5s.

You right, sorry story got long in a try to manipulate one of the four critic subjects point in space for measurement axis.
 

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,736
Likes
16,159
At the end of the day this thread is good for one reason, it's a kind of brains trust of the most stubborn people I've seen in my life and you get to ignore them all!
In German language a joke is quite known, I will try to translate it:

A car driver hears an announcement on the radio: "Attention! Attention! There is one ghost driver on the A9 motorway coming towards you. Please drive slowly on the right lane and don't overtake."
The driver: "What do you mean with one? I see dozens of them, dozens!"
 
Status
Not open for further replies.
Top Bottom