• 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.
D

Deleted member 65

Guest
(I decided to delete my earlier post to provide more detail, didn't realize you had already responded to it, sorry for the confusion.)



Actually this is specified quite clearly in CTA-2034-A:



Note this mention: "the following method shall be used to define it whenever the manufacturer has not specified it". In other words, if the manufacturer did specify a reference axis (as Neumann did for the KH-80) then that's what should be used. It follows that measuring the KH-80 with the tweeter as the reference axis is not compliant with CTA-2034.

Great, appreciate answer!
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
"On-axis" for these speakers corresponds to 0 degrees at the floor. If they were elevated and set on the Klippel platform for measurement the response would be incorrect.

Is that so? Or maybe they should be measured at a horizontal plane at the bottom of the speaker even if they are elevated?
 

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827
Is that so? Or maybe they should be measured at a horizontal plane at the bottom of the speaker even if they are elevated?
Floor reflections are integral to the design. If you look at the pic above the on-axis plane has one half of the speaker above the floor, and the other "half" is below it. The mirror image is part of the shading function and is part of what ensures constant directivity at all heights.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Floor reflections are integral to the design. If you look at the pic above the on-axis plane has one half of the speaker above the floor, and the other "half" is below it. The mirror image is part of the shading function and is part of what ensures constant directivity at all heights.

Ahh, I see - in that case no love for Klippel there. :)

Anyway, I brought this speaker up only because it has no tweeter to acccentuate the need to use manufacturer specified reference axis whenever possible.
 
Last edited:

edechamps

Addicted to Fun and Learning
Forum Donor
Joined
Nov 21, 2018
Messages
910
Likes
3,620
Location
London, United Kingdom
Ahh, I see - in that case no love for Klippel there. :)

Actually it might still be possible to measure it using the NFS, if you post-process the data to simulate the presence of a perfectly reflective floor. In fact, isn't that what the "Predicted In-room Response" calculation does to some extent?
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Actually it might still be possible to measure it using the NFS, if you post-process the data to simulate the presence of a perfectly reflective floor. In fact, isn't that what the "Predicted In-room Response" calculation does to some extent?

This speaker is soo much different than any other so frankly I have no idea if NFS could simulate a reflective floor for it. :)
 

Absolute

Major Contributor
Forum Donor
Joined
Feb 5, 2017
Messages
1,084
Likes
2,125
For responses could be amirm is right regarding variances is not so big a deal and maybe the various scales used confuse a bit, for example below black (NFS) verse purple (anechoic) was C52 and the one that turned opinion of skeptic to approve results from NFS, now look at black (NFS) KH 80 verse red third party KH 80, of course bragging rights for the red curve is lost a bit but are we too sensitive here ...:)

View attachment 47170
Yes, this is getting a little silly. In the top range there will be some offset due to lack of calibration (at the time of the measurement) to Amir's microphone, which can explain some of the difference. The remaining differences could be because of different reference-axis, mechanical variations, different quality measurement systems etc, but in my opinion it's not important. The averaging will iron most of the difference out, and if not, there's something shit somewhere in the off-axis response.

In my opinion the separate off-axis response graphs, both horizontally and vertically, are more interesting than any simplifications of them because they will tell the whole story and will show you how to set up the speakers to maximize linearity according to preferred tonal balance.

Whether one should measure precisely on tweeter axis or wherever the manufacturer deems worthy is a question worth considering because somewhere along the line you will find big speakers that have a huge distance between drivers and so will need to be measured at a further distance and/or better point between the drivers for accurate data/predictions.

In my head I lean towards measuring between the tweeter and midrange at the center line of the baffel and tell every manufacturer that off-axis response is as important in other places than whatever angle-average they deem most pretty.

If the listening window is very smooth and the DI the same, then supreme sound will be the result. Other stuff like capacity, dispersion, distortion, CSD etc will probably take over the throne from frequency response as the most important differences for sound quality when LW and DI is above a certain quality-threshold in smoothness and linearity.

Whether or not Amir is able to produce the absolutely prettiest graph possible for any given speaker is not even remotely relevant, imo. The advanced graphs will not lie or tell a different story anyway. Let's focus on the bigger picture :)
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
Hi,

@amirm
I really value your great work!

I'm excited about independent loudspeaker measurements, but as others pointed out, don't hurry things up.

It makes a difference to measure a loudspeaker outside the reference axis specified by manufacturer, as we all can see in Amir's measurements.
Neumann is one of a few manufacturer I would trust to release "realistic" measurements. Furthermore the measurements of Anselm Goertz in Sound&Recording are the real deal ;-)

To make the point how much difference the measurement outside the reference axis causes, I traced the SPL of Amirm, Neumann, Sound&Recording and combined the curves in one diagram.
KH80_diff_measurements_1.jpg


With a more detailed scaling we get (one dB scaling):
KH80_diff_measurements_2.jpg


If we normalize the other measurements to Amirs, we can easily recognize the "huge" difference caused by off reference axis measurement and the "chassis protection" of the KH 80 regulate sound power on low frequencies (one dB scaling, 1/3oct smoothing):
UPDATE: Below 50Hz the diagram doesn't make sense.
KH80_diff_norm_2.jpg



So, I would also recommend to repeat the measurement with mic on reference axis and maximal 90dB@1m sound pressure.
 
Last edited:

BYRTT

Addicted to Fun and Learning
Forum Donor
Joined
Nov 2, 2018
Messages
956
Likes
2,452
Location
Denmark (Jutland)
:)... part two of post #314 or the other way around using virtual EQ in VituixCAD from the 72 steps amirm shared to get Neumann KH 80 as smooth as possible on axis plus another EQ transfer function to same curve as JBL 305P Mark ii was scanned, think graphs is interesting comparison to JBL 305P Mark ii in post #314.

Neumann KH 80 out of box as scanned:
Neuman_KH_80_Power+DI.png



Neumann KH 80 EQed smooth on axis filter function included:
Neuman_KH_80_Power+DI_smooth.png



Neumann KH 80 transfer EQ to JBL 305P Mark ii out of box response filter function included:
Neuman_KH_80_Power+DI_transfer_to_JBL305.png
 

DJBonoBobo

Major Contributor
Joined
Jan 21, 2020
Messages
1,360
Likes
2,851
Location
any germ
Thanks to everyone who investigated possible explanations for the differences - very interesting!
I just want to comment why it seems so important to do a proper re-measurement: More than most other loudspeaker-reviews this one is about trust. Neumann is one of the most "transparent" companies who are providing loads of measurements on their own. If someone "independent" provides measurements with obvious differences, this automatically rises the question of whom to trust. Trust is very hard to earn and easy to lose, and i guess trust in the brand is a major selling point for Neumann.
So i think this is not really about how good the KH80 are or what difference it makes for the listener but whom to trust with measurements.
 

BYRTT

Addicted to Fun and Learning
Forum Donor
Joined
Nov 2, 2018
Messages
956
Likes
2,452
Location
Denmark (Jutland)
Did you with this 3rd curve applied on-axis filter to make JBl to have the same curve as KH80? :D

Yes :p because in comparison to the other real thing think they interesting.

It seems that both Di graphs are identical on all 3 charts. It would be interesting to see how they woud look with filters applied.

The two DI graphs seems frozen dictated by pattern of the 72 hor/ver txt files, aplying EQ filters the other curves is a puzzle of compromise, for this exercise "In-room response" curve same as amirm's "Estimated In-Room Response" was mirror EQed flat as a pancake and can be seen in the smooth orange graph, rest of compromise was done using few PEQ and shelves, btw can share VituixCad project file should anyone like it, it could be fantastic practical visual feature if ones in room acoustic enviroment was scanned and the polar graph plus spinorama graphs of VituixCad was coupled up to ones DSP engine.

True. But repating the measurements with lower level and using "official" axis, as @edechamps suggested, can't hurt.

Right can't hurt but also there seems tons of work in a long row, would love see it myself especially because NFS looks print fantastic high resolution at low end knee but still think the variations is to live with for the bigger view.
 
Last edited:

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,109
Likes
8,424
Location
NYC
(Please see the following as a philosophical discussion about future measurements:). At this point I've said everything there is to say about the Neumann measurements specifically, but it'd be nice to make things as good as possible for future measurements and not have discussions about remeasuring in the first place).

On the reference axis: I brought this up earlier in the thread, but since we're already 18 pages deep in less than 3 days...

As @edechamps points out, CTA-2034-A says measurements should be referred to the manufacturer axis unless not specified. Granted, we're not following everything in the document to a tee, but this seems like no-brainer. Although I do understand from Klippel's documents that it's faster doing the expansion from the tweeter axis.

Its effects on KH80 specifically aside, I think it's really important to use the manufacturer- reference axis going forward, because sometimes meaningful, potentially audible differences do emerge. The KH80 is one of the best-controlled vertically-aligned speakers out there, so issues with missing the vertical axis by a few degrees are not major. Others are not so lucky, and audibility aside, it can really screw up people's perception of a speaker's performance if the presented on-axis graph does not reflect the intended reference axis.

Here's is the Buchardt S400 at its reference axis (between woofer and waveguide) in white and just 10 degrees above the reference axis, which I estimate is roughly close to pointing at the tweeter. In other words, basically the exact same situation as the KH80 EDIT: Brainfart. I forgot the S400 is an upside-down configuration meant to be with the woofer above the tweeter. Nonetheless, errors show up both ways. Here's 10 degrees above the reference axis (aimed at the woofer)

S400 Reference Axis.png


And 10 degrees below the reference (at the tweeter):

S400 Ref and D10.png
Barring a major issue with my measurements in that region - which I doubt is the case - that is a massive difference in the upper mids. It's okay on the KH80 because it doesn't change that much in the first place, but if I saw the blue or puple curve as the on-axis graph for the Buchardt, I'd be concerned. Granted, the listening window would smooth things out a bit, but it would still be significantly worse than the listening window centered around the proper reference axis.

This is a great speaker on the whole and these listening window vertical issues are significantly diminished and balance out in a typical living room setup. But you nevertheless want to get the vertical height right to maximize optimal listening position, especially if you listen from unusually close.

Another example: here's the Focal Chora 806. Being 10 degrees below the reference point (which in this case is the tweeter, as normal) causes a major shift:

Chora Reference.png


Also remember, we do not just hear an even balance of direct sound and reflected sound. Which one we hear more of varies with frequency. The higher in frequency, the more the direct sound matters. In addition, the lower the relative volume of reflections vs direct sound, the more the direct sound matters (obviously).

Per Toole's book:

Snag_45ce0de3.png


In the nearfield or a heavily treated room, it is fair to assume the distribution of perception looks more like the bottom cinema hart, as reflections are less audible. In my experience, small changes in vertical position are occasionally audible in the farfield, and sometimes very audible in the nearfield. Changes in vertical position are most likely to manifest in the crossover frequencies and above, where we begin to be more sensitive to the direct sound.
 
Last edited:

HammerSandwich

Major Contributor
Forum Donor
Joined
Nov 22, 2018
Messages
1,137
Likes
1,497
It seems that both Di graphs are identical on all 3 charts. It would be interesting to see how they woud look with filters applied.
You've noticed the fundamental lesson of modern speaker design: you cannot EQ directivity. Because the DI shows only the differences between on- & off-axis sound, changing the input signal makes no difference. (Confusing? Consider polar maps normalized so the on-axis response appears dead flat.) Use all the fancy FIR you want, but you need to get the acoustics correct, too.

There is an exception to this: the massively complicated approach. You can control an array's directivity by EQing each driver individually. Note that this is more of an acoustical change than it is EQ per se. See the big Beolab for one example. Or the CBT shading above.

So I guessed, implicitly we’ll then end up with two categories.
@edechamps gave the best answer, but I'll note that we already have 2 categories. Explicitly. Speakers with a recommended axis and those without. This situation was created by the manufacturers, not ASR.

But I expect that this axis issue can be handled in post. Even if the NFS does not include the feature, it spits out enough data to compute the best direct-sound axis later. Right?
 

Absolute

Major Contributor
Forum Donor
Joined
Feb 5, 2017
Messages
1,084
Likes
2,125
I'm mathematically and computer retarded, can anyone use the supplied graphs from the Neumann to change the calculation of the listening window to use the - 10 degree vertical angle as the reference? If @napilopez is right to be concerned, we should see a significant difference in the listening window average.
@BYRTT? You seem to be a magician with that virtual EQ program and graphs :)
 

Pio2001

Senior Member
Joined
May 15, 2018
Messages
317
Likes
507
Location
Neuville-sur-Saône, France
Actually this is specified quite clearly in CTA-2034-A:

Note this mention: "the following method shall be used to define it whenever the manufacturer has not specified it". In other words, if the manufacturer did specify a reference axis (as Neumann did for the KH-80) then that's what should be used. It follows that measuring the KH-80 with the tweeter as the reference axis is not compliant with CTA-2034.

Thanks for the clarification.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
You've noticed the fundamental lesson of modern speaker design: you cannot EQ directivity. Because the DI shows only the differences between on- & off-axis sound, changing the input signal makes no difference. (Confusing? Consider polar maps normalized so the on-axis response appears dead flat.) Use all the fancy FIR you want, but you need to get the acoustics correct, too.

LOL
Thank you for a lesson in elementary school math but I'm having a flew and my body temp was 39C when I wrote that post so I was not quite myself. ;)

Early reflection DI = Listening window - Early reflections

After EQ filter of X dB:
Early refflection DI = Listening window+X - (Early reflections +X)= Listening window -Early reflections
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,389
Location
Seattle Area
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,389
Location
Seattle Area
Don't you want to know what your system is truly capable of?
I already know what it is capable of. It produced the flattest response of any speaker I have tested with this sample. That is what the manufacturer promotes, and that is what the results are. There is just no way anyone should try to focus so minutely on speaker measurements. The spin data for example sums many graphs as to get rid of such nuances.

If you all want to dig in, you are welcome but please don't involve me. I have spent more time dealing with protests on the measurements than remotely measuring speakers. It is not fun and promotes a strategy that is just wrong with it comes to evaluating speakers based on measurements.

If you want to dig in, how about asking the manufacturer what "interpolated" means on their graph? What samples were used and whether customer units are tested to comply with that spec and if not, they accept returns. Ask them what the frequency resolution is. What lab and technique was used to make those measurements.

Collectively, you all are taking what started as a fun hobby to measure speakers into a miserable back and forth over things that don't matter. You are welcome to keep going but know the costs as you do so....
 
D

Deleted member 65

Guest
I already know what it is capable of. It produced the flattest response of any speaker I have tested with this sample. That is what the manufacturer promotes, and that is what the results are. There is just no way anyone should try to focus so minutely on speaker measurements. The spin data for example sums many graphs as to get rid of such nuances.

If you all want to dig in, you are welcome but please don't involve me. I have spent more time dealing with protests on the measurements than remotely measuring speakers. It is not fun and promotes a strategy that is just wrong with it comes to evaluating speakers based on measurements.

If you want to dig in, how about asking the manufacturer what "interpolated" means on their graph? What samples were used and whether customer units are tested to comply with that spec and if not, they accept returns. Ask them what the frequency resolution is. What lab and technique was used to make those measurements.

Collectively, you all are taking what started as a fun hobby to measure speakers into a miserable back and forth over things that don't matter. You are welcome to keep going but know the costs as you do so....

You’re doing a great job @amirm, don’t get distracted. I’m sure we’re (most of us) behind and supporting what you’re accomplishing.
Just keep walking the ASR path!
/Cheers
Lars
 
Status
Not open for further replies.
Top Bottom