Neumann KH120 - tweeter issue or normal unit difference?

[ernestcarl]

You' re talking about the KH120? How did you figure that out?

KH120, yes. Quasi-anechoic measurements. The on-axis do not look too different, but it’s the off-ax curves moreso — totally expected given the aforementioned offset plastic lens which acts to improve the diffusion of the highest frequencies.

 

[HarmonicTHD]

When I measure both speakers at the same position the mic was directing to the tweeter in a distance of ~10 cm. I move the mic slightly (~2 cm) from measurement to measurement to rule out diffraction issues. That was relatively uncritical.

I edited my post above. Sorry. Was a stupid idea. The acoustic axis is not coincidental with the tweeter. But see above. My fault.

 

[Lilith]

KH120, yes. Quasi-anechoic measurements. The on-axis do not look too different, but it’s the off-ax curves moreso — totally expected given the aforementioned offset plastic lens which acts to improve the diffusion of the highest frequencies.

I was looking for pictures where this can be seen, but didn't find anything.

 

[LTig]

That's the measurement with periodic pink noise and moving the mic at the listening position for both speakers. SPL was around ~78 dB(C) at listening position.

Nothing to worry about, I'd say. I wonder much more about the difference below 50 Hz - is SNR too low in this region?

Anyway if you could fix the deep null at 100 Hz the improvement in SQ should be much higher compared to replacing the tweeter ...

PS: I think you used the wrong mode in REW's RTA window (Spectrum) since the curve is falling - it should be flat. Using pink noise you must use RTA 1/48 octave.

 

[Lilith]

Nothing to worry about, I'd say. I wonder much more about the difference below 50 Hz - is SNR too low in this region?

Anyway if you could fix the deep null at 100 Hz the improvement in SQ should be much higher compared to replacing the tweeter ...

PS: I think you used the wrong mode in REW's RTA window (Spectrum) since the curve is falling - it should be flat. Using pink noise you must use RTA 1/48 octave.

Difference from below 50Hz comes from handling the mic.

 

[ernestcarl]

I was looking for pictures where this can be seen, but didn't find anything.

It’s the plastic sticker behind the protective tweeter grill. The deviation in yours look more like stemming from something in the driver unit itself. If the response difference is consistent with volume (i.e. doesn’t get worse with higher SPL sweeps) then manual “speaker EQ” akin to those two PEQs applied may be enough to mitigate the difference on- and off-axis. You are anyways using DRC which already applies considerably much more EQ.

*Here’s a description of the term used: https://www.sweetwater.com/insync/acoustic-lens/

 

[wwenze]

My interpretation of your data + others' data is that this is manufacturing variation within acceptable limits. Which one is more correct? That would be philosophical. This kind of thing can even be a loose screw, or even a mechanical or electrical part.

Without knowing the root cause or which is more correct I won't bother with EQ. Your head positioning between the two speakers would cause bigger effects anyway.

 

[Lilith]

My interpretation of your data + others' data is that this is manufacturing variation within acceptable limits. Which one is more correct? That would be philosophical. This kind of thing can even be a loose screw, or even a mechanical or electrical part.

Without knowing the root cause or which is more correct I won't bother with EQ. Your head positioning between the two speakers would cause bigger effects anyway.

That's my impression too. Maybe the next speaker will have a deviation, but in the other direction.

 

[Lilith]

Nothing to worry about, I'd say. I wonder much more about the difference below 50 Hz - is SNR too low in this region?

Anyway if you could fix the deep null at 100 Hz the improvement in SQ should be much higher compared to replacing the tweeter ...

PS: I think you used the wrong mode in REW's RTA window (Spectrum) since the curve is falling - it should be flat. Using pink noise you must use RTA 1/48 octave.

I think the 100 Hz null is ceiling bounce. Seems my 10 cm Basotect panel doesn't help here. I have it across the whole room width at listening position. When I move back the frequency shifts to higher frequencies. When I get closer to the speaker the hole gets filled. I don't hear any difference when moving closer to the speaker, even when playing a kick at 100 Hz. So it seems to look more serious than it is.

 

[Lilith]

I ordered a second pair. Price went up btw., but still get the old price. Meanwhile I'm waiting for the email by Sennheiser.

 

[wwenze]

I think the 100 Hz null is ceiling bounce. Seems my 10 cm Basotect panel doesn't help here. I have it across the whole room width at listening position. When I move back the frequency shifts to higher frequencies. When I get closer to the speaker the hole gets filled. I don't hear any difference when moving closer to the speaker, even when playing a kick at 100 Hz. So it seems to look more serious than it is.

I also notice my ear does not agree entirely with my mic measurement of the room

That's why I finish the tuning by ear. After all it's my ear listening to the music not the mic.

 

[Lilith]

No reply from Sennheiser yet. I don't think they will answer. New speakers are not in stock at the moment, so I'm waiting for another pair to double check.

 

[Lilith]

I have some update: Today the second pair of KH 120 arrived and the last 3 hours I measured all of them at the same position. Slight variations in the mic position are possible of course and I slighty varied the mic between measurements. Volume was around 84 dB at 1 m distance and I measured at ~10 cm in front of the tweeter and at 1 m distance.

Fig 1 shows measurements at the tweeter directly. The green curve is my old right speaker that showed that peak at 15.5 kHz, the blue is my old left speaker that didn't show that peak or at least less pronounced. Green and black curves are the new monitors (green = new 1, black = new 2). The "new 2" monitor is much more (!) noisy when connected to the interface compared to all three other monitors and that's why I will not keep it. Maybe a grounding issue? I tend to keep green and red, cause they both show that resonance with a similar amplitude and I think that's normal as three out of four speaker show it. What do you think? I also attached the mdat files below in case some wants to have a look.


Fig 1

Fig. 2: at 1 m distance:

The old left speaker has a small peak in the distortion measurement at ca. 470 Hz (see Fig 3) which is not present in all others (e.g. monitor new 1, Fig 4)
Fig 3:

Fig 4

300622_K120.mdat - Shared with pCloud

Keep all your files safe, access them on any device you own and share with just the right people. Create a free pCloud account!

e1.pcloud.link

 

[Lilith]

Is the difference in the distortion plot at 40-50 Hz relevant? By ear I hear port noise at that range for all speakers testet.