• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Neumann KH 80 DSP Monitor Review

Pio2001

Active Member
Joined
May 15, 2018
Messages
181
Likes
238
Location
Neuville-sur-Saône, France
Applying such heavy smoothing after averaging is a little hm-hm, wouldn't you say?
That's right. I was surprised to see such strong variations on the raw curve from measurement to measurement. The systematic difference appear at the 1/3 octave level.
I wonder if it comes from the small movements of the microphone while I was holding it, or from the movements of my own body.

It is better than that assuming you aplied supplied calibration file. Although it seems not to be a common pratice around here I strongly suggest you do it. ;)
I did, of course. The mic rolls of -1 dB at 9500 Hz, -2 dB at 11000 Hz, -3 dB at 12700 Hz (90deg). This is corrected on the measurements.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
2,792
Location
Zg, Cro
That's right. I was surprised to see such strong variations on the raw curve from measurement to measurement. The systematic difference appear at the 1/3 octave level.
I wonder if it comes from the small movements of the microphone while I was holding it, or from the movements of my own body.
Or from some other reason. But my point was that you can't cure inter-measurement variations with strong smoothing as that merely covers up the issue.
 

Pio2001

Active Member
Joined
May 15, 2018
Messages
181
Likes
238
Location
Neuville-sur-Saône, France
Or from some other reason. But my point was that you can't cure inter-measurement variations with strong smoothing as that merely covers up the issue.
It doesn't come from the microphone changing its frequency response minute after minute.
It doesn't come from the speaker changing its frequency response either.
It doesn't come from the electronics.
Therefore it must come from the measurement conditions.

These conditions were a lot more constrained than any real-case listening conditions. It means that in real life, what we are going to hear when sitting in front of the speaker is even less accurate than that, because our listening conditions should suffer from the same inaccuracies at least.
 

LarsS

Addicted to Fun and Learning
Forum Donor
Joined
Mar 23, 2016
Messages
934
Likes
781
Location
Sweden, Stockholm
" You paid 50,00 USD

to Audio Science Review LLC "

Upped my forum donation supporting @amirm time & purchases and for not remeasuring the KH80.

I'm up for challenges from those who advocate a remeasure ... ;-)

" WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. "
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
2,792
Location
Zg, Cro
Sad. :mad: Try to improve your behavior.
Why sad? Judging by his posts would you say he's having fun? Do you remember any jokes he told here? As he practically lives on this forum you certainly can't say he's not taking online life too seriously, right? So tell me M8, have I said something wrong or being impolite? ;)

Btw, would you say for yourself you're also ready to be teased and not taking online life too seriosuly? :D

What else is left apart from joking? Common sense obviously has left this thread..
 
Last edited:

rodtor

Member
Forum Donor
Joined
Aug 21, 2019
Messages
72
Likes
97
Location
Toronto
Why sad, I came here to have fun. ;) Btw, would you say for yourself you're also ready to be teased and not taking online life too seriosuly? :D

What else is left apart from joking? Common sense obviously has left this thread..
There are times, in this thread, @Krunok, and in others recently, when your 'teasing' has had a distinctly nasty aspect. If this is your idea of 'fun', what does that tell you about yourself? As for the alleged absence of 'common sense' in this thread, keep in mind that you are one of its leading contributors, especially of late.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
2,792
Location
Zg, Cro
As for the alleged absence of 'common sense' in this thread, keep in mind that you are one of its leading contributors, especially of late.
When I mentioned lack of common sense I was referring to Amir's decision not to remeasure KH80 although sound arguments were provided by a decent number of competent people in favor of doing so.
 

dukanvadet

Member
Forum Donor
Joined
Oct 18, 2019
Messages
40
Likes
38
One thing i thougt about with this extremely flat on axis graph from Neumanns measurements is that i my experience its not always a great idea to try for absolute flat on axis, there is always some form of difraction effect from the box or waveguide that isnt solved with eq. When you correct on axis you could worsen something more than the original "problem" at another axis.
However there seems to be very good agreement with the off axis horizontal and listening window. I am just curious if it is really the best solution.
Not trying to make any strong claims in this post, just thinking out loud and hoping for other to add their thoughts
 

Mnyb

Active Member
Joined
Aug 14, 2019
Messages
228
Likes
247
Location
Sweden, Västerås
Amateur question here ( i’m doing the learning and fun:)

Would not flat on axis be more important in the extreme nearfield you are supposed to use these speakers ?
It seem almost headphone like ? Would the room contribute much in these circumstances for a listener ?
Acoustically it does but are not are brains able to separate out some reflections and would not this work even better in close nearfield ?
 

Sancus

Addicted to Fun and Learning
Forum Donor
Joined
Nov 30, 2018
Messages
606
Likes
1,212
It seems fairly clear that this speaker is both very flat on axis AND has excellent directivity, so I'm not sure there would be anything to be gained by making it less flat on axis. Usually that is done in response to directivity issues, but this speaker already beats basically everything I've ever seen measured except for Genelec's (much more expensive) coaxials.
 

Thomas_A

Senior Member
Forum Donor
Joined
Jun 20, 2019
Messages
443
Likes
247
It seems fairly clear that this speaker is both very flat on axis AND has excellent directivity, so I'm not sure there would be anything to be gained by making it less flat on axis. Usually that is done in response to directivity issues, but this speaker already beats basically everything I've ever seen measured except for Genelec's (much more expensive) coaxials.
Depens on the design philosophy, I don't aim for a flat on axis response, but adjustments in the +/- 1.5 dB range based on the stereo system errors to get a more neutral timbre of the center phantom image.
 

Ilkless

Addicted to Fun and Learning
Joined
Jan 26, 2019
Messages
603
Likes
932
Location
Singapore
One thing i thougt about with this extremely flat on axis graph from Neumanns measurements is that i my experience its not always a great idea to try for absolute flat on axis, there is always some form of difraction effect from the box or waveguide that isnt solved with eq. When you correct on axis you could worsen something more than the original "problem" at another axis.
However there seems to be very good agreement with the off axis horizontal and listening window. I am just curious if it is really the best solution.
Not trying to make any strong claims in this post, just thinking out loud and hoping for other to add their thoughts
The enclosure is an optimised low-diffraction design using moulded polycarbonate, so that is far from the issue. We are not talking thin ply boxes with with right angles.
 

hardisj

Senior Member
Joined
Jul 18, 2019
Messages
456
Likes
1,429
From someone who tested (using a Klippel for quite some time) and posted this data for everyone to see on my now-defunct website medleysmusings.com, here are my takes:
1) When you're starting off, it's easy to be enraptured with the new toy. It's easy to take things personally. Years away and experience showed me that in my early days, those who I viewed as being combative were (mostly) trying to help and/or curious of certain "inaccuracies" themselves.
2) No matter what is said to/from this is what I am seeing: a bit of an ego because you have the Klippel... and the Klippel doesn't lie. I was that same person for quite some time. I fully admit it. It took me a few months to get over myself. I believe this whole ordeal is no different... I think cooler heads will prevail and time will lead to the same result that I ultimately came up with in #1 above.
3) The purpose of the NFS is to offer the user automation. That's where the cost really lies. Heck, just a standard loudpseaker turntable and CLIO costs upwards of $5k. Numerous software (as well as Klippel) also offer the ability to conduct intelligent nearfield/farfield splicing based on the DUTs effective diameter. But the big benefit of the NFS to me is the wholly automated way of performing the measurement and obtaining the results that would otherwise take hours of manually adjusting. I know. I did this for years myself. I would legitimately spend a good 8 hours to test a single drive unit: from build up, set up, record, analyze, report findings and tear down. The ability to automate the testing to a high degree of accuracy and repeatability is ... well, a god send, to be blunt.
4) With 3 above, and with the desire of this site to be so focused on objective data, I don't understand why re-testing the DUT at the manufacturer specified listening axis isn't performed. First, that's how the 2034 spec says to do it. Second, it's honestly just a no-brainer. Thirdly, if the test is called in to question legitimately then the thread should be updated with the new test results. Or at the very least a huge disclaimer with the points made here should be posted. Still, I've learned that people see graphs and run. They don't take time to read discussion the majority of the time. And unfortunately, as we will see in the future if this thing continues to happen, the internet will be filled with incorrect assumptions about data captured in a non-idealized way. Pages and pages of arguments could have been saved. But, if I'm being honest, I think this has more to do with #2 above.
5) Blitz testing should be avoided at first. I get it. It's a new toy. You want to test a bunch of stuff. But that leads to errors as we are seeing discussed here. At least in the early stages it should be: Perform a test. Post the data. Wait for some feedback. Learn from both it and the viewership's feedback. Implement necessary changes (some of which I believe have been implemented, like the SPL adjustment). And then go forward. Throwing a bunch of data out without taking time to consider what it means leads us to where we are now. I've already seen others on another forum now question the accuracy of this site's testing thanks to this very thread. Think about that for a second. One mis-step and one acknowledgement of said mis-step will now lead to less trust. That's a shame. Truly.

Argue with me if you want. I can point to pages and pages of me going through the EXACT same thing when I first started testing with the Klippel (LSI/LPM, TRF/DIS to be more concise). I've been in the exact same shoes as Amir. And I am the only other person I know of who has used the Klippel to provide data for all. So, I can speak from literal experience on this matter. I made the same mistakes... from letting my ego get in the way by thinking that if I were the owner of said machine then how could anyone question my expertise to not taking time to step back and review what I'm doing. I realize the NFS was a costly undertaking. I know this full well. I hope he can take the time to step back and see that my reply here comes from a place of hope. Hope that he will remember to breathe and evaluate not only his community's concern but also evaluate what his goal is: to provide a lot of data... or provide the most accurate data he can. And I hope you all who seem to be going at this in a spiteful way will try to step back yourselves and ask how you can help the community (not just yourself or not just Amir) learn from this event.

- Erin
 
Last edited:

dukanvadet

Member
Forum Donor
Joined
Oct 18, 2019
Messages
40
Likes
38
The enclosure is an optimised low-diffraction design using moulded polycarbonate, so that is far from the issue. We are not talking thin ply boxes with with right angles.
Sure, this is a very competent design but i dont think taht problem exclusively apply to thin plywood boxes with straight edges, one example is earl geddes speakers. Generally very good directivity and large roundovers on all edges but the 0° axis (not the listening axis) is a little ragged. According to him that is a concious compromise to get the best sound where you listen. The speakers are toed in 20 or 18° wich will give you a wider sweet "spot" due to time intensity trading and balances out the a little bright 0° axis when you are at that axis from one of the speakers with the weaker off axis from the other. If you would EQ the 0° axis perfectly flat all other would be crappy
 
Last edited:

hardisj

Senior Member
Joined
Jul 18, 2019
Messages
456
Likes
1,429
Also, I'll add this...

I did testing from about 2010-2017. My daughter's health declined around 2017 and I made the obvious choice to stop testing. But even before that final decision came, I grew aggravated with people almost expecting me to test things for them. It initially went from "hey, Erin, if I send you xx driver, will you test it" to "I'm sending you this driver to test and I need it back in a week".

I hope you guys remember that what Amir is doing is a service. Whether or not he's gaining money from it... he's not charging you all to see the data. So, once this thing finally gets on track just remember that he's not your puppet. ASK. BE POLITE. BE APPRECIATIVE. Don't take for granted what he's providing. Because I can tell you from my experience, it will get old fast and he may just stop it altogether.

I'm not saying you have to kiss his butt. But don't get to the place where you feel entitled.

I am about to start back up testing and I already am dreading the day I start getting those kind of messages.

- Erin
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
8,776
Likes
11,183
From someone who tested (using a Klippel for quite some time) and posted this data for everyone to see on my now-defunct website medleysmusings.com, here are my takes:
1) When you're starting off, it's easy to be enraptured with the new toy. It's easy to take things personally. Years away and experience showed me that in my early days, those who I viewed as being combative were (mostly) trying to help and/or curious of certain "inaccuracies" themselves.
2) No matter what is said to/from this is what I am seeing: a bit of an ego because you have the Klippel... and the Klippel doesn't lie. I was that same person for quite some time. I fully admit it. It took me a few months to get over myself. I believe this whole ordeal is no different... I think cooler heads will prevail and time will lead to the same result that I ultimately came up with in #1 above.
3) The purpose of the NFS is to offer the user automation. That's where the cost really lies. Heck, just a standard loudpseaker turntable and CLIO costs upwards of $5k. Numerous software (as well as Klippel) also offer the ability to conduct intelligent nearfield/farfield splicing based on the DUTs effective diameter. But the big benefit of the NFS to me is the wholly automated way of performing the measurement and obtaining the results that would otherwise take hours of manually adjusting. I know. I did this for years myself. I would legitimately spend a good 8 hours to test a single drive unit: from build up, set up, record, analyze, report findings and tear down. The ability to automate the testing to a high degree of accuracy and repeatability is ... well, a god send, to be blunt.
4) With 3 above, and with the desire of this site to be so focused on objective data, I don't understand why re-testing the DUT at the manufacturer specified listening axis isn't performed. First, that's how the 2034 spec says to do it. Second, it's honestly just a no-brainer. Thirdly, if the test is called in to question legitimately then the thread should be updated with the new test results. Or at the very least a huge disclaimer with the points made here should be posted. Still, I've learned that people see graphs and run. They don't take time to read discussion the majority of the time. And unfortunately, as we will see in the future if this thing continues to happen, the internet will be filled with incorrect assumptions about data captured in a non-idealized way. Pages and pages of arguments could have been saved. But, if I'm being honest, I think this has more to do with #2 above.
5) Blitz testing should be avoided at first. I get it. It's a new toy. You want to test a bunch of stuff. But that leads to errors as we are seeing discussed here. At least in the early stages it should be: Perform a test. Post the data. Wait for some feedback. Learn from both it and the viewership's feedback. Implement necessary changes (some of which I believe have been implemented, like the SPL adjustment). And then go forward. Throwing a bunch of data out without taking time to consider what it means leads us to where we are now. I've already seen others on another forum now question the accuracy of this site's testing thanks to this very thread. Think about that for a second. One mis-step and one acknowledgement of said mis-step will now lead to less trust. That's a shame. Truly.

Argue with me if you want. I can point to pages and pages of me going through the EXACT same thing when I first started testing with the Klippel (LSI/LPM, TRF/DIS to be more concise). I've been in the exact same shoes as Amir. And I am the only other person I know of who has used the Klippel to provide data for all. So, I can speak from literal experience on this matter. I made the same mistakes... from letting my ego get in the way by thinking that if I were the owner of said machine then how could anyone question my expertise to not taking time to step back and review what I'm doing. I realize the NFS was a costly undertaking. I know this full well. I hope he can take the time to step back and see that my reply here comes from a place of hope. Hope that he will remember to breathe and evaluate not only his community's concern but also evaluate what his goal is: to provide a lot of data... or provide the most accurate data he can. And I hope you all who seem to be going at this in a spiteful way will try to step back yourselves and ask how you can help the community (not just yourself or not just Amir) learn from this event.

- Erin
This plus 1000.

There is a good deal of simple wisdom in this post. And obviously it was gained from personal experience.

The biggest issue of all in Amir not doing a retest is an immediate erosion of trust by a large number of people. I'm not saying I distrust him, but there are plenty who will. Once that has happened, most especially at the very beginning of the testing history, it will take a couple dozen more speakers measured without any glitches to get most of that trust back. And some of that trust will never be fully regained.

IMO, benefits of doing a re-test far, far outweighs any negatives or time involved in the second round of testing.

I shake my head in the resistance to this one retest. Especially as a retest was immediately planned on another speaker due to results being a bit off from expectaions (speaking of the Kali). Why the hard line against a retest of the KH80? Humor the few dozen or few hundred of us this one time.
 

Ilkless

Addicted to Fun and Learning
Joined
Jan 26, 2019
Messages
603
Likes
932
Location
Singapore
Sure, this is a very competent design but i dont think taht problem exclusively apply to thin plywood boxes with straight edges, one example is earl geddes speakers. Generally very good directivity and large roundovers on all edges but the 0° axis (not the listening axis) is a little ragged. According to him that is a concious compromise to get the best sound where you listen. The speakers are toed in 20 or 18° wich will give you a wider sweet "spot" due to time intensity trading and balances out the a little bright 0° axis when you are at that axis from one of the speakers with the weaker off axis from the other. If you would EQ the 0° axis perfectly flat all other would be crappy
I am familiar with the Geddes oblate spheroid. The 0° dip is a consequence of how low diffraction the waveguide is. It is mathematically optimised for low diffraction especially in the throat (prioritising that over SPL/compression driver loading), giving the happy problem of such a coherent wavefront at the mouth that there is a minimal amount of reflection at the mouth that is a single high-Q dip at ONLY 0° horizontal. The listening window is excellent - Stanford's 3D3A measured one example.

Geddes had a post about it on DIYAudio that I would dig up were I not traveling right now.
 
Top Bottom