Neumann KH 80 DSP Speaker Measurements: Take Two

[echopraxia]

It is the same machine, but it wasn't the same speaker. Until the same speaker is measured with both machines it is hard not to speculate about inter-specimen differences.

But it's not exactly the same machine. It's a different instance/installaction of a machine from the same product line. Ideally, this should be identical and yield perfectly consistent results vs other machine instances. But this is apparently not the case, since there is already a small list of modifications and software updates that have been applied over time to fix errors. For example, regarding the microphone diffraction/reflections issue -- why is it that Amir had to do anything at all to correct for this? Isn't this something that Klippel themselves should validate and ensure is never a problem that influences the measurements?

Consider also the point 'ctrl' made above:

In your reply you completely ignore the narrow tolerances for the frequency response on axis that Neumann specifies.
Reproduction accuracy between 100 Hz and 10 kHz:
100% of loudspeakers produced: ±0.26 dB
80% of loudspeakers produced: ±0.23 dB
50% of loudspeakers produced: ±0.17 dB

I understand why Neumann's claim here might be doubted. However, keep in mind, we do at least have some evidence that corroborates their claimed manufacturing consistency: Amir measured two independent samples of the KH80, and found they measured almost identically

Of course, we do not yet know which measurement rig here is the "most correct" one in an absolute sense, or why the measurements differ here. It's possible that Neumann is unknowingly tuning their speakers incredibly consistently to the wrong definition of 'flat'. It's also possible that there are calibration or setup validation issues with Klippel. It's also possible that Neumann has a drastically higher sample variation than they claim, though personally I doubt it.

It will be difficult to find out the truth though, without the assistance/collaboration of Neumann here (e.g. by providing their review sample to Amir).

In that sense I do think it's fair to consider the ball to be in Neumann's court, so to speak, to prove that the sample they sent to Klippel without telling Amir is not in any way a 'golden sample'. As others have said though, it's only been a day, so it seems fair to give Neumann at least a little time to respond

 

[Thomas savage]

I think these guys probably feel as we did not contact them when we found some less than stellar anomalies in our measurements they don't need to do the same for us here.

Can't blame them but it's a missed opportunity, companies that value great engineering like Neumann have a vested interest in that being shown off in reviews .

This is a place to do that , a bit of working together would be good for all concerned and the industry in general imo.

 

[Andreas007]

While a lot of what is brought is appreciated, it is not the best I have seen. Best manufacturers come to me, provide another sample, and take back what I tested for their measurements. We then have a complete circle to resolve what the problem might be. This would have been a far faster process than the many months that this interaction has taken.

Ultimately we are not helped much here with them measuring a different speaker than the pair I tested. What do I change moving forward? How do we resolve the next dispute? Surely by not taking 5 months and testing another sample elsewhere.

The approach, which I understand from their point of view, is to disprove the data here. It is not to help advance a collaboration where we arrive at a mutual solution. It leaves me half full as it is.

First, IMO this is quite a harsh comment on Neumann (maybe it's a cultural thing, again).
Before reading this post it seemed to me that everyone was really interested about causes of measurement variations. We all want to learn.
Then it started to loose track...

Second, I didn't get the impression that Neumann wants to disprove you by their response.
In fact, my impression was/is that they really care about this forum.
Complaining about something someone hasn't done or should have done from your point of view is not a positive discussion approach.

Finally, please let us all calm down. I'm quite sure Neumann is willing to cooperate with Amir in some way.
As we see, loudspeaker measurement is far more complex than many of us thought initially.
Faults, misunderstanding and misinterpretation are normal when using complex and heavily software based systems. Even from the most experienced users. I see it every day in my job (semiconductor industry).

Therefore, please do not blame each other for no obvious reason just to end up in mutual recriminations.

 

[Zvu]

...As we see, loudspeaker measurement is far more complex than many of us thought initially.]

It is complicated because you are trying to predict something, not really measure it. NFS, however advanced it may be, is made to make measurements more practical and cheaper. Their goal is to be as accurate as they can but real comparison would be quasi-anechoic measurement outdoors, free-field and compare that measurement to the one made by NFS for same loudspeaker cabinet in given conditions indoor. When you establish a reference like that (KH80 seems ideal for this job) then you can be sure your measurements are correct.

Same can be done in a large anechoic chamber (10m x 10m) but do it gated as you would do it outdoors so you don't have to deal with deficiencies of that chamber.

 

[QMuse]

But it's not exactly the same machine.

It's the machine with same software and different mcirophone, but that shouldn't matter once you apply microphone correction curve.

 

[ctrl]

Of course, resolving this is not made easy if Neumann doesn’t share the same sample with ASR that they gave to Klippel.

I understand why Neumann's claim here might be doubted. However, keep in mind, we do at least have some evidence that corroborates their claimed manufacturing consistency: Amir measured two independent samples of the KH80, and found they measured almost identically

The evidence available strongly indicates which measurements are most likely to be incorrect.

Already in January @napilopez had posted the following comparison (Klippel means ASR measurement)

This already indicated that something could be wrong with the ASR measurement.

Then Amir took a second sample. The second measurement also shows the strange behaviour above 4-5kHz and below 100Hz.
Not sure if the measurement conditions were identical (fitting error, resolution,...).
The second one measured pretty much the same:

If we are generous, the deviation of the two samples is +-0.5dB in the range 60-11000Hz:

This means that the first sample also worked correctly.

Now we also have the new measurements from Neumann and the confirmation of the measurement by Klippel himself.
As already mentioned, the deviations of the ASR measurements in the upper frequency range were due to Amir making measurements with the mic protection cage. Only the low frequency behavior is still a mystery.

If that were all, I would agree with Amir and others in their doubts that the measurement by Neumann and Klippel could be a "golden sample" that was corrected in the bass range.
I have smoothed the frequency responses a little bit to be able to see the deviations better.

But in the meantime we have so many ASR measurements that differ significantly in the low bass frequency range from the manufacturer's specifications or measurements in audio magazines that it makes it very likely that the problem lies with the ASR NFS.
To confirm this, we actually do not need the Neumann sample anymore.

Apart from that, the ASR measurements up to 10-12kHz now correspond very well with other measurements.

 

[infinitesymphony]

The ASR vs. outside measurement deviation graphs @ctrl created are worth a second look. Let's put them together:

 

[Vintage57]

My experience with Neumann as a customer, by email.

I have always received a response within 24 hours. I've asked questions about speaker placement, sub placement, manufacturing dates, test sheets which they would provide in the past, as well as grilles for the 420's as I have young grandchildren. In all but one instance it was Markus Wolf who replied and always helpful.

In response to the test sheets, here is Markus's reply.

We do not provide measurements results of our loudspeakers anymore since we changed our measurement equipment. Our loudspeakers still fulfill a typical deviation from loudspeaker to loudspeaker within 0.5 dB independent on the production date.


In comparison I wonder if anyone has any stories about reaching out to their manufacturer in the US and what their experience was.

 

[hmt]

@ ctrl

Yep, in the end it is a matter of ockhams razor. That something went wrong in amirms measurement is simply the best explanation as of now. I also would understand Neumann if did not have the time to dig deeper since they did not find faults in their measurements and units. Why bother with someone elses measurement errors? We know which is the outlier here and the likelyhood that amirm got a second sample that is ridden with the same issues due to sample variation is next to zero.

 

[amirm]

Let me quote the summary in this thread OP:

And from the original review:

The right story was told about the speaker and its excellent engineering. Both Neumann and its fans should have celebrated these measurements. Instead, a few of you decided to get out the pitchforks and go after me, Klippel, the phase of the moon, etc. You are welcome to continue but just keep in mind any down side will be straight at your feet.

 

[amirm]

@ ctrl

Yep, in the end it is a matter of ockhams razor. That something went wrong in amirms measurement is simply the best explanation as of now.

So we test two different speakers -- products that naturally differ from each other -- and your simplest explanation was that???

I swear some of you are just motivated to make this situation worse....

 

[Zvu]

So we test two different speakers -- products that naturally differ from each other -- and your simplest explanation was that???

I swear some of you are just motivated to make this situation worse....

Just don't pay too much attention to that. Whenever someone sets to do something big, there are always people that would like to see him fail.

And it is a big thing what you are doing here, be sure of that.

 

[hmt]

So we test two different speakers -- products that naturally differ from each other -- and your simplest explanation was that???

I swear some of you are just motivated to make this situation worse....

Yes because all the other samples (Neumann, SR and ctrl) were looking like neumanns measurement and your two samples are the only ones to look that odd. Sorry but measurement error is the most likely explaination.


And yes, I really appreciate what you are doing here but you should not take constructive criticism personal.

 

[Thomas savage]

So we test two different speakers -- products that naturally differ from each other -- and your simplest explanation was that???

I swear some of you are just motivated to make this situation worse....

People on the internet are often on their own side , it's not about making the situation worse they won't be thinking like that , they just won't be thinking about anything but themselves and being right or casting doubt to appear clever etc .

It's the nature of the beast.

 

[Juhazi]

I still don't know which of these measuements shows "real or right" spl below 200Hz! Amir's and Klippel's NFS give different result, that should be investigated in the first place! Are they using same real mic distance/radius?

Other methods of measuring will for sure give different results - ground plane/summed nearfield/huge anechoic chamber.

Room response is the final outcome anyway, and it depends on tens of other factors which vary with different types of bass systems. It is too easy to let subwoofer(s) take care of that, large hifi speakers should be capable to reach 30Hz in-room without help from modes, but that is too much asked from KH80 or similar small nearfield monitors. I am questioning floorstander measurements more than these shoeboxes!

 

[edechamps]

a few of you decided to get out the pitchforks and go after me, Klippel, the phase of the moon, etc.

People here want to understand the limits of the various measurement setups used (yours and others). Yes, they are obsessing over small differences. It's a healthy obsession, though. These differences should definitely not be there (they are not down to random measurement uncertainty as @ctrl clearly demonstrated). Understanding and fixing these differences would make everyone better off and would clarify potentially lingering doubt around bass response in other measurements. I am surprised that you seem to be annoyed by this effort, given that you gladly investigated and fixed similar issues (e.g. treble ripple due to reflections off the microphone cage) in the past.

So we test two different speakers -- products that naturally differ from each other -- and your simplest explanation was that???

Before @GuyLayfield posted his data, a popular hypothesis was that the bass discrepancy was somehow related to the way the NFS works. But now that hypothesis has been refuted. We now know for sure that the NFS, or at least the NFS that Klippel used, can produce a measurement that is completely consistent with other methods in the specific case of the KH 80, at least in the bass range. (Which is great news in and of itself!)

With that in mind, let's assume that your KH 80 measurements are accurate and reflect the true response of the samples you tested. That would mean that both samples you tested have the exact same deviation in their bass response that does not show up in any other sample than anyone else tested. Including Klippel themselves using another NFS.

That seems… unlikely. The only way that could possibly make sense to me is if both your samples suffered from an identical production issue. But a production issue causing such a consistent, perfectly repeatable problem would be unusual. I suppose it could happen if Neumann's factory calibration system was itself miscalibrated and used the wrong target perhaps? I do agree that it would be great if measurements were compared on the same sample to rule that out.

I can understand that some people such as @hmt now favour the hypothesis of a problem with your measurement setup (your particular NFS, or your procedure, or something else). They simply updated their credences in response to new data. Which is pretty much the essence of the scientific method, really.

Other methods of measuring will for sure give different results - ground plane/summed nearfield/huge anechoic chamber.

Nope. As has been discussed many times in this thread, multiple measurement methods from different people all agree on what the bass response of this speaker looks like. Including NFS-based methods, and even across multiple samples. Except Amir's of course, hence this whole debate.

 

[hardisj]

Guys, a simple (and proper) ground plane measurement will tell us all we need to know regarding the bass of this sample unit. This is the easiest part of measuring a speaker. Whoever is in possession of the speaker at this point can conduct an on-axis GP measurement in minutes and report the data here.

All you need is a mic and REW. Don’t need levels. Don’t need accuracy above 400hz. Just need shape of the bass. A large backyard or parking lot will suffice.

 

[Mawclaw]

 

[spacevector]

Guys, a simple (and proper) ground plane measurement will tell us all we need to know regarding the bass of this sample unit. This is the easiest part of measuring a speaker. Whoever is in possession of the speaker at this point can conduct an on-axis GP measurement in minutes and report the data here.

All you need is a mic and REW. Don’t need levels. Don’t need accuracy above 400hz. Just need shape of the bass. A large backyard or parking lot will suffice.

Is this really as simple as finding a large open area, setting the speaker on the ground, setting the mic 1m from the speaker on the ground and running a sweep using REW?

 

[ernestcarl]

Room response is the final outcome anyway ... It is too easy to let subwoofer(s) take care of that ... I am questioning floorstander measurements more than these shoeboxes!

I don't know about others, but foresee using KH80s with a sub -- so a dB or two dip in the bass for me personally would be largely inconsequential for small monitors such as these. As has been observed, though, other people's mileage vary.