- Joined
- Sep 16, 2019
- Messages
- 1,202
- Likes
- 2,672
It was referred to as a 'studio monitor' in the 2nd webinar along with the picture and data. I would be very surprised if it wasn't the KH120
-
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 Speaker Measurements: Take Two
- Thread starter amirm
- Start date
It was referred to as a 'studio monitor' in the 2nd webinar along with the picture and data. I would be very surprised if it wasn't the KH120
I agree, just wondered if I'd missed something and it'd been explicitly mentioned
.Anyway, I personally wouldn't take that spin much to heart given its purpose in the presentation(i.e. "here's a spinorama!" rather than "here's how the KH120 performs!"). No guarantee the speaker was set to neutral settings, might be taken off axis or referenced to an improper distance(though I've assumed the NFS references the 2034 plot to 2m by default), and maybe even without the mic calibration loaded (rising treble, especially in the sound power, strikes me as suspect).
DI curves should be representative though.
Last edited:
- Joined
- Sep 16, 2019
- Messages
- 1,202
- Likes
- 2,672
I agree, just wondered if I'd missed something and it'd been explicitly mentioned
Anyway, I personally wouldn't take that spin much to heart given its purpose in the presentation(i.e. "here's a spinorama!" rather than "here's how the KH120 performs!"). No guarantee the speaker was set to neutral settings, might be taken off axis or referenced to an improper distance(though I've assumed the NFS references the 2034 plot to 2m by default), and maybe even without the mic calibration loaded (rising treble, especially in the sound power, strikes me as suspect).
DI curves should be representative though.
True. We don't know all the details. In any case the measured result was pretty darn good, at least in my opinion?
Hello, Guy Layfield and Markus Wolff from Neumann Studio Monitor development here. We’ve received a lot of requests to contribute to this discussion, and in particular to comment on the differences between our published KH 80 measurements, and measurements made on the Klippel Near Field Scanner for this forum. The measurements from the Near Field Scanner shown here have a 2 dB dip below 200 Hz, and strong peaks and dips from 2 kHz upwards.
We therefore measured a KH 80 and then sent it to Klippel. The below plots compare measurements done by Neumann, and by Klippel with their Near Field Scanner.
Neumann’s measurement set up:
Klippel’s measurement set up:
The below graph compares the frequency responses from the 2 measurement methods. There are a couple of differences – the NFS measures up to 0.6 dB higher than the free field measurement from 200 – 500 Hz. The reason for this difference is unknown. The NFS measurement has a peak at 15 kHz and a dip at 20 kHz. The calibration chart for the NFS measurement microphone has a peak and dip at the same frequencies although of smaller magnitude (see below plot, 15 kHz +0.5 dB, 20 kHz -1.5 dB). The NFS measurement does not include this microphone calibration curve which partially explains the difference. Beyond that, the reason for the mismatch is unknown.
Despite these 2 differences, the 2 measurements are very close, both lie within +/- 0.8 dB of the target 80 dB curve up to 13.5 kHz. As the unit was measured on different systems in different conditions on different days, we have to allow for a degree of measurement uncertainty.
NFS measurement microphone free field response shown in blue:
The close agreement between Neumann’s and Klippel’s measurements does not match the measurements shown in this forum. I think it was already mentioned in the thread that the errors at higher frequencies were caused by the microphone protection cage and the microphone’s own frequency response. The 2 dB dip below 200 Hz is not seen in the NFS measurements done at Klippel.
One possible error is that Audio Science Review’s CEA2034 calculation was done with the virtual measurement mic at 0.3 m (unless I’ve misunderstood the graph). This is then a near field measurement with a different frequency response compared to a measurement at 1.4 m (closer to far field). The effect is shown in the below graph, but this still doesn’t fully explain the earlier measurements on this forum. In any case, we are satisfied that our low frequency measurements (<200 Hz) are accurate due to the good fit between our own and Klippel’s measurements.
The other measurements from Klippel are shown below:
The above Isobar plots show Neumann’s directivity measurements on the left (Horizontal then Vertical) and the Klippel NFS results on the right. The horizontal measurements match very well. The vertical measurement from the Klippel NFS is slightly narrower than Neumann’s own measurements. One possible reason is that the omnidirectional measurement microphone becomes increasingly directional at higher frequencies. In horizontal measurement this is not a problem as the microphone rotates around the DUT, so is always on-axis. For vertical measurements the microphone moves up and down in a plane parallel to the floor, so the further away the microphone is from the 0° position, the more off-axis the microphone is, relative to the DUT, and so the more the microphone’s directivity affects the measurements.
We therefore measured a KH 80 and then sent it to Klippel. The below plots compare measurements done by Neumann, and by Klippel with their Near Field Scanner.
Neumann’s measurement set up:
- Free field measurement (anechoic chamber)
- Input: swept sine wave, -20 dBu, 8x averaging
- Sample rate: 96 kHz
- Impulse windowed in the time domain to reduce noise and remove distortion artefacts
- Frequency resolution: 17th order FFT (giving 65536 frequency bins 0 Hz to fs/2)
- No smoothing
- Measured at 1.4 m distance, level corrected by 3 dB to give sensitivity at 1 m
- Low frequency correction curve (for frequencies below room’s anechoic cut-off) generated by ground plane method
- High frequency correction curve (from measurement microphone response)
Klippel’s measurement set up:
- Near Field Scanner
- Input: swept sine wave, -20 dBu
- 48 kHz sample rate
- Frequency resolution: 12 points per octave
- No microphone correction applied
- Coordinate Reference Point: Loudspeaker’s acoustical axis
The below graph compares the frequency responses from the 2 measurement methods. There are a couple of differences – the NFS measures up to 0.6 dB higher than the free field measurement from 200 – 500 Hz. The reason for this difference is unknown. The NFS measurement has a peak at 15 kHz and a dip at 20 kHz. The calibration chart for the NFS measurement microphone has a peak and dip at the same frequencies although of smaller magnitude (see below plot, 15 kHz +0.5 dB, 20 kHz -1.5 dB). The NFS measurement does not include this microphone calibration curve which partially explains the difference. Beyond that, the reason for the mismatch is unknown.
Despite these 2 differences, the 2 measurements are very close, both lie within +/- 0.8 dB of the target 80 dB curve up to 13.5 kHz. As the unit was measured on different systems in different conditions on different days, we have to allow for a degree of measurement uncertainty.
NFS measurement microphone free field response shown in blue:
The close agreement between Neumann’s and Klippel’s measurements does not match the measurements shown in this forum. I think it was already mentioned in the thread that the errors at higher frequencies were caused by the microphone protection cage and the microphone’s own frequency response. The 2 dB dip below 200 Hz is not seen in the NFS measurements done at Klippel.
One possible error is that Audio Science Review’s CEA2034 calculation was done with the virtual measurement mic at 0.3 m (unless I’ve misunderstood the graph). This is then a near field measurement with a different frequency response compared to a measurement at 1.4 m (closer to far field). The effect is shown in the below graph, but this still doesn’t fully explain the earlier measurements on this forum. In any case, we are satisfied that our low frequency measurements (<200 Hz) are accurate due to the good fit between our own and Klippel’s measurements.
The other measurements from Klippel are shown below:
The above Isobar plots show Neumann’s directivity measurements on the left (Horizontal then Vertical) and the Klippel NFS results on the right. The horizontal measurements match very well. The vertical measurement from the Klippel NFS is slightly narrower than Neumann’s own measurements. One possible reason is that the omnidirectional measurement microphone becomes increasingly directional at higher frequencies. In horizontal measurement this is not a problem as the microphone rotates around the DUT, so is always on-axis. For vertical measurements the microphone moves up and down in a plane parallel to the floor, so the further away the microphone is from the 0° position, the more off-axis the microphone is, relative to the DUT, and so the more the microphone’s directivity affects the measurements.
We were also asked about measurements of the KH 120 in the NFS, so we repeated the process. Measurement setup is as above for the KH 80. There is a slight low frequency difference between measurements done in free field and in the Klippel Near Field Scanner, the cause of this difference is unclear. The high frequency difference is similar to that seen on the KH 80. Part of this is from the NFS microphone itself, but there appears to also be some unknown fixed offset between the 2 measurement set ups above 13 kHz.
The other measurements from Klippel are shown below:
The other measurements from Klippel are shown below:
Thanks a lot for coming here to discuss these measurements. Unless it was found somewhere in this thread, is that 0.3 m distance also the cause of the "wrong" acoustic center?
We should consider a New forum category for manufacturers feedback. Considering we have a good start here.
I'm not sure what you mean by "causing" the wrong acoustic center. However, one possible hypothesis is that, if @amirm's CEA2034 charts are referenced at the wrong distance, and at the same time the wrong reference axis is used, then these two issues could conspire with each other to produce an on-axis frequency response that looks correct in the treble because the evaluation point is in front of and pointed towards the tweeter, but is wrong in the bass because the evaluation point is too far from the woofer relative to the tweeter.
That would be an amazing "perfect storm" if that was the case though, and it seems highly unlikely one would get an overall response that's as flat as @amirm obtained if such fundamental issues were at play.
Also, my understanding is that Klippel's software generates CEA2034 charts from far-field data by default, and you'd have to go out of your way to make it generate this chart from near-field data (if that's even possible), so it seems unlikely @amirm would do that by accident.
I think the point that may require clearing is this one:
"One possible error is that Audio Science Review’s CEA2034 calculation was done with the virtual measurement mic at 0.3 m (unless I’ve misunderstood the graph). This is then a near field measurement with a different frequency response compared to a measurement at 1.4 m (closer to far field). The effect is shown in the below graph, but this still doesn’t fully explain the earlier measurements on this forum. In any case, we are satisfied that our low frequency measurements (<200 Hz) are accurate due to the good fit between our own and Klippel’s measurements."
Thank you for the comprehensive tests done to compare and validate your own measurements. I, and I think I also speak for many users here, admire Neumann's transparency. Your commitment to evidence-based loudspeaker engineering is some of the best I've seen in this industry.
Is it possible that setting the virtual mic too close to the speaker affects the driver summation (or the calculation of it), leading to the observed differences?
Vindermere
Member
- Joined
- Mar 12, 2020
- Messages
- 40
- Likes
- 31
impressive performance of the small KH120
ernestcarl
Major Contributor
Well, it is a surprise that detailed measurements for the KH120 are posted here. Didn’t expect to see them at all. But here they are...
This clearly shows that the manufacturer's measurements are very reliable, contrary to what some have claimed here.
Since other sources do not show the frequency response increase from 8kHz on, I would attribute this as an error to the NFS. This could be caused by poorly calibrated measuring microphones, which still have problems even with the calibration file.
Here is a comparison of the measurements of Klippel, Neumann and ASR (pay attention to the scaling):
Due to the errors made when measuring the speakers here in the forum, there were considerable deviations. But Amir has now corrected these errors (The measuring microphone should perhaps be calibrated again or measured with a completely different measuring microphone - if possible - to clarify whether the measuring microphone itself is the cause of the deviation).
If we compare the measurements of Neumann and ASR with the Klippel measurement and leave out the range above 8kHz (until the cause of the deviation is finally clarified), we see how reliable the manufacturer measurement is - Deviation from normalized Klippel measurement:
With a little smoothing (1/24 oct) the deviations would be around +-0.5dB (40-12000Hz) - a better match can hardly be achieved.
In comparison, the ASR measurement - deviation from the normalized Klippel measurement (ignore curves below 50Hz):
Here again for comparison the measurements of ASR, Neumann and Sound&Recording:
The slightly falling frequency response above 8kHz in the Sound&Recording measurement is probably due to the measurement method.
Sound&Recording measures via "ground plane" method which leads to deviations in high frequencies even with minimal alignment errors - see here in the thread. Or a point between tweeter and bass-midrange driver was taken as reference point - which would also explain the slightly falling curve.
Last edited:
It's great to see Neumann contributing and helping clear up any discrepancies. It's wonderful how this forum is helping really bring out the trustworthy manufacturers.
I think between the two threads and many speaker measurements since the KH80, most of the discrepancies had been figured out. These were small ones in the grand scheme of things, and the KH80 was early on in Amir's testing process. The bass is the only thing that remained (and remains) a bit of a mystery.
At this point I didn't care much about the discrepancies, but by testing the same speaker in both your own setup and klippels, it goes a long way in showing Neumann's commitment to accuracy and transparency.
Also, if I may toot my own horn a bit, I thought I'd reiterate just how close you can get to "the real thing" using DIY methods, in my case measuring the speaker on top of a plastic turntable placed on my kitchen island:
Note that the new neumann/klippel measurements seem to be without the early reflections fix. So above I've included both the ER as calculated by the NFS and as by explained by harman. Also note my spin was generated with 15-degree interval data rather than the 10-degrees mandated by the spin, and incomplete rear hemisphere data(interpolated with VituixCAD). In this case, that's still close enough.
Of course, the quasi anechoic resolution is nearly useless for detecting small, high Q resonances in the lower midrange, but there are ways around this if needed.
EDIT: If I'd chosen a higher splice point for the bass it'd be even closer. Despite being different samples with completely different measurement setups, agreement on all curves is basically within +/-1dB up to 12kHZ or so. Curiously, I even have the little dip at 550hz too, and bass contour matches perfectly, despite being an imperfect nearfield summation.
I think between the two threads and many speaker measurements since the KH80, most of the discrepancies had been figured out. These were small ones in the grand scheme of things, and the KH80 was early on in Amir's testing process. The bass is the only thing that remained (and remains) a bit of a mystery.
At this point I didn't care much about the discrepancies, but by testing the same speaker in both your own setup and klippels, it goes a long way in showing Neumann's commitment to accuracy and transparency.
Also, if I may toot my own horn a bit, I thought I'd reiterate just how close you can get to "the real thing" using DIY methods, in my case measuring the speaker on top of a plastic turntable placed on my kitchen island:
Note that the new neumann/klippel measurements seem to be without the early reflections fix. So above I've included both the ER as calculated by the NFS and as by explained by harman. Also note my spin was generated with 15-degree interval data rather than the 10-degrees mandated by the spin, and incomplete rear hemisphere data(interpolated with VituixCAD). In this case, that's still close enough.
Of course, the quasi anechoic resolution is nearly useless for detecting small, high Q resonances in the lower midrange, but there are ways around this if needed.
EDIT: If I'd chosen a higher splice point for the bass it'd be even closer. Despite being different samples with completely different measurement setups, agreement on all curves is basically within +/-1dB up to 12kHZ or so. Curiously, I even have the little dip at 550hz too, and bass contour matches perfectly, despite being an imperfect nearfield summation.
Last edited:
And, also, thank you for showing us that two very well thought out measurement processes can also still result in deviation. It makes the 'fight' easier for backyard testers like myself to not stress over a small degree of variance (so long as care and effort is taken to ensure as high a degree of accuracy as possible).
OP
- Thread Starter
- #438
Thank you very much for investigating the measurements and joining the forum to share them. Much appreciated.
I wish you had contacted me though and shared the same sample for testing. With me testing a different speakers than yours, and measurement parameters varying from Klippel (I use higher resolution for example), we still have some unknowns that could have been worked out.
For now, the big difference I see is that your measurements are at acoustic center and mine are relative to tweeter axis. I suspect there is diffraction when measured at tweeter axis that shows up in my and Klippel's measurements that don't show up in yours.
Anyway, if you are willing, I would be very interested in measuring the same speaker sample.
Your transparency in providing detailed measurements both on the Neumann website and here is unrivaled, as far as I know. Thank you so much for doing this.
Also, I’ve never seen a better manufacturer response to a review than this: It doesn’t get much better than producing new rigorously measured data!
I have no doubt I am going to be a lifelong Neumann fan, and will continue to recommend the KH120 to everyone I know as perhaps the best speaker I’ve heard in its price range
Last edited:
Yes, this seems extremely important. I was under the impression that the Klippel software doesn’t have virtual mic distance as an adjustable parameter to the CEA2034 computations. I was also under the impression that CEA2034 requires an exact microphone distance of 2m (virtual or real being irrelevant here) with +6db added to scale SPL for easy comparability with 1m SPL specs.I think the point that may require clearing is this one:
"One possible error is that Audio Science Review’s CEA2034 calculation was done with the virtual measurement mic at 0.3 m (unless I’ve misunderstood the graph). This is then a near field measurement with a different frequency response compared to a measurement at 1.4 m (closer to far field). The effect is shown in the below graph, but this still doesn’t fully explain the earlier measurements on this forum. In any case, we are satisfied that our low frequency measurements (<200 Hz) are accurate due to the good fit between our own and Klippel’s measurements."
If the various CEA2034 charts we’ve seen (around here and elsewhere) have varying measurement parameters like this, doesn’t that throw the whole notion of directly comparing different speaker’s spins and preference scores into question?
Fortunately, even if so, I believe it should be possible (in theory at least) to recompute CEA2034 charts using consistent virtual mic distances — since the whole point of the Klippel data is that it enables such things without re-measuring.
Last edited:
Similar threads
