Neumann KH 80 DSP Speaker Measurements: Take Two

[thewas]

Indeed, as expected from a 3-way (even if a ported 2-way speaker is kind of a 3-way). But now, I really wonder how incredible the KH310 would be if it was properly ported.

The Neumann chief designer had a very strong reason of not making it ported in opposition to his other models and had told it at a seminar, but unfortunately don't remember it. Still if you listen to the KH310 it can go very loud (much much louder than a KH120 or most 2-way) as we humans perceive distortion quite differently than a "stupid" THD measurement.

 

[q3cpma]

The Neumann chief designer had a very strong reason of not making it ported in opposition to his other models and had told it at a seminar, but unfortunately don't remember it. Still if you listen to the KH310 it can go very loud (much much louder than a KH120 or most 2-way) as we humans perceive distortion quite differently than a "stupid" THD measurement.

I hope it's not a "group delay" reason, because a port would make them arguably full-range and/or with way less distorsion. Maybe it's got something to do with Neumann seemingly hating back ports and the lack of space to put them in front on this model?

 

[Vintage57]

Deleted

 

[q3cpma]

Just for the sake of clarity. Are you saying it’s not properly ported? If not. What would lead you think this?

Well, it's not ported at all. I mean that bass reflex is an incredible tool if made with all the modern tools we have (mainly computer simulation to reduce turbulence and some techniques like bending to avoid/reduce leakage).

 

[jhaider]

View attachment 68853View attachment 68854
These models being near in maximum SPL is quite coherent with the measurements. Ported vs sealed is the obvious reason for this.

That’s not what I take away from those two graphs in the band under discussion (50-100 Hz). The bottom one shows significantly higher capability. At any rate, I’m basing my argument on independent third party data, not factory data. One cannot rationally argue the Neumann bass headroom spec is accurate for KH120.

 

[ernestcarl]

Overlay of the KH120 & KH80 (light pastel lines) graphs from Klippel below with grids aligned and actual SPL offset to match. Initially, I thought the soundpower measurement would be much worse on the 120s, but it doesn't seem so obvious here. Not much of a difference to write home about really... If played at the same volume and nearfield distance, the only give away would be the extended bass (hump) of the KH120 -- that is, if both were not crossovered with a sub.

 

[Deleted member 16502]

I gather from this graph that there's not much use of adding DSP for frequency correction if the design is good in the first place. Only comfort convenience reasons (for manufacturer and user) and possible phase/room correction.

 

[TimVG]

One thing that I just noticed: If the CEA2034 calculation by the NFS extrapolates the measured nearfield data to 2m (with loudness correction to display 1m) then why the high frequency droop when taking the midpoint between tweeter and woofer as acoustical axis?

If the data were truly extrapolated to show the soundfield at 2m distance, this should not be an issue. Anyone?

 

[gr-e]

One thing that I just noticed: If the CEA2034 calculation by the NFS extrapolates the measured nearfield data to 2m (with loudness correction to display 1m) then why the high frequency droop when taking the midpoint between tweeter and woofer as acoustical axis?

View attachment 69060

If the data were truly extrapolated to show the soundfield at 2m distance, this should not be an issue. Anyone?

This is with the expansion point set between the tweeter and the woofer. For most optimal performance Klippel needs the tweeter to be the expansion point, otherwise you'll get an error like this high frequency droop.

(source)

If you want to see the actual response at acoustic center you need to look at vertical offaxis measurement a few degrees below 0.
Though, there might be a way to change the reference axis without moving the expansion center from tweeter in Klippel software.

 

[napilopez]

If you want to see the actual response at acoustic center you need to look at vertical offaxis measurement a few degrees below 0.
Though, there might be a way to change the reference axis without moving the expansion center from tweeter in Klippel software.

There is - it's what I believe @GuyLayfield means when he says "Coordinate Reference Point: Loudspeaker’s acoustical axis " in their post. You can clearly see this parameter in Klippel's demo videos for the NFS:

As you note and others have, the tweeter point for holographic expansion is not the same as the reference point, the latter which is basically where the designers choose to optimize the response for. It's not necessarily where the soundfield originates but that of course doesn't matter to the listener sitting a meter+ away =]

 

[gr-e]

Oh, right, I actually mentioned that on page 14, but forgot about it

 

[TimVG]

Thanks for clarifying.

 

[napilopez]

Oh, right, I actually mentioned that on page 14, but forgot about it

You're getting old!

 

[ernestcarl]

I gather from this graph that there's not much use of adding DSP for frequency correction if the design is good in the first place. Only comfort convenience reasons (for manufacturer and user) and possible phase/room correction.

Yep! There's nothing to correct with the speaker itself.

Your listening position in your room, though... well, you'll still need to make adjustments if DSP is available for you to use. I recently watched Neumann's demo instructional video on how to add PEQs for the KH80 using an iPad. If I'm not mistaken, there are only 8 available PEQ filters and 2 shelving filters per monitor. I think that's perfectly adequate.

To date, I'm using 8 PEQs for the left speaker and 6 for the right. Already using the speaker's -1dB HF shelfving switch in the back... And, decided to get rid of my few other mild 'cosmetic' filters which just made the graphs a teeny bit 'prettier' (hurrah!), but otherwise, not much use in reality. We could probably count the sub as a separate 'monitor' in their app -- I myself use 6 PEQs for the sub itself and some phase adjustment. But that's it. There not much else to 'fix' in my close monitoring setup.

I think there is a real risk overdoing things (deviating from what's already neutral and potentially increasing distortion -- at higher SPLs) if one had access to an infinite number of filters to play around with... that is, if you don't exactly know what you're doing.

 

[Pio2001]

If I'm not mistaken, there are only 8 available PEQ filters and 2 shelving filters per monitor. I think that's perfectly adequate.

I find it a bit limited. I currently use 17 filters, including 2 low shelves, in my room correction for Neumann KH-120.
I could get rid of several of them, but that would mean some extra work, for a less accurate result.

I've got 2 room modes below 100 Hz to deal with, 1 wide SBIR cancellation at 100 Hz, at least one SBIR peak near 200 Hz, and yet two unwanted peaks until 800 Hz. I also like to boost frequencies below 50 Hz, since I'm not using the KH-120 at full power.

That's 1 PEQ for low bass boost
5 PEQ for an accurate correction of the two room modes, that are not shaped like PEQs.
2 for the shallow 100 Hz SBIR (too wide for one PEQ only)
1 for the SBIR peak at 200 Hz
2 for the two extra peaks between 200 and 800 Hz.
The 2 low shelves are meant to choose an overall balanced target curve.

Therefore I could reduce my 17 filters to 13 without loosing accuracy, I think.
But having to use less filters than 13 would annoy me.

 

[QMuse]

I find it a bit limited. I currently use 17 filters, including 2 low shelves, in my room correction for Neumann KH-120.
I could get rid of several of them, but that would mean some extra work, for a less accurate result.

I've got 2 room modes below 100 Hz to deal with, 1 wide SBIR cancellation at 100 Hz, at least one SBIR peak near 200 Hz, and yet two unwanted peaks until 800 Hz. I also like to boost frequencies below 50 Hz, since I'm not using the KH-120 at full power.

That's 1 PEQ for low bass boost
5 PEQ for an accurate correction of the two room modes, that are not shaped like PEQs.
2 for the shallow 100 Hz SBIR (too wide for one PEQ only)
1 for the SBIR peak at 200 Hz
2 for the two extra peaks between 200 and 800 Hz.
The 2 low shelves are meant to choose an overall balanced target curve.

Therefore I could reduce my 17 filters to 13 without loosing accuracy, I think.
But having to use less filters than 13 would annoy me.

Can you post REW file with measurement of your uncorrected response?

 

[ernestcarl]

I find it a bit limited. I currently use 17 filters, including 2 low shelves, in my room correction for Neumann KH-120.
I could get rid of several of them, but that would mean some extra work, for a less accurate result.

I've got 2 room modes below 100 Hz to deal with, 1 wide SBIR cancellation at 100 Hz, at least one SBIR peak near 200 Hz, and yet two unwanted peaks until 800 Hz. I also like to boost frequencies below 50 Hz, since I'm not using the KH-120 at full power.

That's 1 PEQ for low bass boost
5 PEQ for an accurate correction of the two room modes, that are not shaped like PEQs.
2 for the shallow 100 Hz SBIR (too wide for one PEQ only)
1 for the SBIR peak at 200 Hz
2 for the two extra peaks between 200 and 800 Hz.
The 2 low shelves are meant to choose an overall balanced target curve.

Therefore I could reduce my 17 filters to 13 without loosing accuracy, I think.
But having to use less filters than 13 would annoy me.

Ah, I understand your situation, and believe you did mention your listening distance being farther away than nearfield -- maybe try adding a sub(s) to devide the workload? I tried putting my monitors nearer the corners of my smallish room to increase the distance (just to experiment by adding more 'room reflections' and for a more diffuse effect*), but that added a number of problems which in total increased the amount of EQ I needed to use. So I totally see your point in wanting more... esp. true for certain situations. This may be a pipe dream, but, by the time the new KH120DSP is released, Neumann should have gone through the effort into adding substantially more functionality and options into their control app/software and DSP'd monitors! That is my hope, at least.

*I've solved that problem by adding rears. Reducing the rears' volume by a dB or two increases front focus significantly. Increasing the rears' volume by a dB or two increases diffuse soundfield effect and feeling of spaciousness -- room sounds much bigger than it is. I use JRiver DSP presets so it just takes a second to make changes as needed/wanted.

 

[Pio2001]

Here are some measurements made some time ago.
A multipoint one (the reference)
Two MMM ones taken with furniture placed a bit differently in order to show possible variations
And a single point one in order to show group delay

 

[QMuse]

Here are some measurements made some time ago.
A multipoint one (the reference)
Two MMM ones taken with furniture placed a bit differently in order to show possible variations
And a single point one in order to show group delay

You have 27dB of peak to peak in the 30-100Hz range. That is a nightmare to correct and you should really consider finding a better position for your speakers.

Here is what you can get with 9 PEQs and 1 shelve. This is done with 0 gain filter so no attenuation is needed but filter will reduce headroom by 12.6dB. I have attached rePhase file so you can see how I did it. If this is a response of 1 speaker it has good chances to work but if you measured both speakers response than it won't.

 

[ernestcarl]

You have 27dB of peak to peak in the 30-100Hz range. That is a nightmare to correct and you should really consider finding a better position for your speakers.

Here is what you can get with 9 PEQs and 1 shelve. This is done with 0 gain filter so no attenuation is needed but filter will reduce headroom by 12.6dB. I have attached rePhase file so you can see how I did it. If this is a response of 1 speaker it has good chances to work but if you measured both speakers response than it won't.


View attachment 69513

Yep.

As the host in the Neumann control app video linked to earlier emphasized, "room acoustics first, signal processing last."

Which isn't always an easy balance to achieve...