• 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 Speaker Measurements: Take Two

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,027
Likes
1,298
Location
Finland
Instructions for IR gating in REW https://www.roomeqwizard.com/
- photo editor used was Photoscape http://www.photoscape.org/ps/main/index.php both excellent freeware!

ER18DXT two-way measurement at 1500mm distance, speaker on a stand in large room, pillows on floor to suppress floor reflection. Microphone was UMIK-1 with correlation. First reflection at 2,5ms, second at 5ms. Tukey 0,25ms windowing used. Notice we adjust Right window after the signal peak.

First, IR right window gating set to 2,5ms to mask the first reflection. Everything below 1kHz is obscure
er18 instr 500ms to 25ms 124-vert.jpg


But when we set IR gating to 5ms, we see lower freq and the first 2,5ms reflection's effect

er18 instr 500ms to 5ms 124-vert.jpg
 
Last edited:

LeftCoastTim

Active Member
Forum Donor
Joined
Apr 15, 2019
Messages
191
Likes
389
Here it is. Speaker's acoustic axis is 4.5 ft off the ground, middle of the room.

#2 3ft window.png
3ft gated.png
 

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,039
Likes
2,984
Location
NYC
Here it is. Speaker's acoustic axis is 4.5 ft off the ground, middle of the room.

View attachment 51397 View attachment 51398
Thanks! That looks better. There's still something interfering with the data at 3.5ms, but closer to what I expect. What's your reference axis? NVM you said it's the reference axis. The dip above 15K is interesting. My uncalibrated umik does that but not the CSL one.
 
Last edited:

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,039
Likes
2,984
Location
NYC
Sorry, not trynig to be a pain - but if I saw that blip at 3.5ms I assume something is interfering with the data. Anything that might've caused a reflection? Or perhaps some errant sound during the sweep? For reference, here's my impulse response vs the same measurement in the frequency domain both gated and in-room at 1m.

There's only so much you can get a clean emasurement in room but assuming a clear path from the speaker to the mic you can get a pretty clean measurement. I'm lucky to have a pretty open patch in my apartment where im able to do measurements though.

KH80 IR.png

KH80 gated vs in room.png
 

Attachments

LeftCoastTim

Active Member
Forum Donor
Joined
Apr 15, 2019
Messages
191
Likes
389
Sorry, not trynig to be a pain - but if I saw that blip at 3.5ms I assume something is interfering with the data. Anything that might've caused a reflection? Or perhaps some errant sound during the sweep? For reference, here's my impulse response vs the same measurement in the frequency domain both gated and in-room at 1m.

There's only so much you can get a clean emasurement in room but assuming a clear path from the speaker to the mic you can get a pretty clean measurement. I'm lucky to have a pretty open patch in my apartment where im able to do measurements though.
Yeah, that 3.5ms blip is always there.

Maybe I got two bad units that measure the same? :)
 

LeftCoastTim

Active Member
Forum Donor
Joined
Apr 15, 2019
Messages
191
Likes
389
Nah, that doesn't look like a speaker issue - it's too far from the initial impulse. More likely a small reflection from your measurement setup. A mystery!
I was thinking more of the frequency response, unlike your KH80.
 

Sancus

Addicted to Fun and Learning
Forum Donor
Joined
Nov 30, 2018
Messages
617
Likes
1,222
FWIW, when I receive the rest of my K&M speaker stand components(sigh, EU -> Canada mail delivery...) I'm happy to attempt another independent gated measurement. Don't feel like attempting it with stacked boxes right now. :p
 

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,039
Likes
2,984
Location
NYC
I was thinking more of the frequency response, unlike your KH80.
The frequency response is the impulse response. :) Same data presented differently. If there's a blip in the impulse response, it shows up in the frequency response =] a blip so far after the initial impulse points to a reflection or other artefact not related to the speaker. I just can't tell you what it is without seeing your home and setup unfortunately.

What does the FR look like if you set the window to 3.3ms(or right before that blip?) Also set the right window to rectangular or tukey 0.1 for a bit higher resolution
 
Last edited:

LeftCoastTim

Active Member
Forum Donor
Joined
Apr 15, 2019
Messages
191
Likes
389
The frequency response is the impulse response. :) Same data presented differently. If there's a blip in the impulse response, it shows up in the frequency response =] a blip so far after the initial impulse points to a reflection or other artefact not related to the speaker. I just can't tell you what it is without seeing your home and setup unfortunately.

What does the FR look like if you set the window to 3.3ms(or right before that blip?) Also set the right window to rectangular or tukey 0.1 for a bit higher resolution
Here you go:
Untitled1.png
Untitled.png
 

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,039
Likes
2,984
Location
NYC
That's more like it! 3m might be cleaner too(3.3ms might've cut it a bit too close), but other than the HF droop which I don't have an explanation for, that looks pretty close to what I'd expect.+/- 1dB within the region with the most resolution (should have 20 points per octave starting around 6K with a 3ms gate).

The small dip at 3K is probably still a bit of a measurement artefact as it's not in amir's measurement either.

Anyway, thanks again. Feel free to mess around with the impulse response and see what results you get. For future reference, Using speaker and microphone stands with minimal footprint helps for future measurements you might do too (the 3.5ms blip might be from your microphone stand, even)

We know that the old Neumann measurements(and all the other ones) had some high frequency ripples that Amir corrected in the Genelec's measurements. Your measurements tell me largely what i expected - the big HF ripples are not really there and the speaker is quite flat. Amir seems to have solved just about all of the Klippel's quirks now, which you data helps confirm. And now you know how to take basic gated measurements for future speakers =]

Curious as to what causes that HF droop . It might be something in your signal chain since it's present in both speakers, but it's also in amir's so who knows, maybe this is a potential weakness present in some KH80s and you happened to get two of them.

I think i can finally stop being persnickety about these measurements. Curious to see what Neumann says once they respond to this thread, though we already know the cause for almost all the anomalies. And of course, almost none of this is very audible in the grand scheme, but we've learned a lot about the Klippel by scrutinizing the data.
 
Last edited:

ctrl

Active Member
Joined
Jan 24, 2020
Messages
236
Likes
797
Location
DE
Curious as to what causes that HF droop . It might be something in your signal chain since it's present in both speakers, but it's also in amir's so who knows, maybe this is a potential weakness present in some KH80s and you happened to get two of them.
Something similar I would also suspect.

- It could be the low-pass filter of the audio interface, if it starts early and the measurement is not done in two-channel mode, i.e. it is a single-channel measurement.
You could use a higher sample rate to see if something changes - e.g. try 96kHz instead of 48kHz.

- The used gate is set a bit narrow. You could set 0 or -0.1 in the menu "Left Window (ms)" in REW and see if something changes.
 
Joined
Feb 24, 2020
Messages
11
Likes
13
Dear Amirm,
As you have have appear to have solved some of the HF systematic errors in the measurements and got some sensible results from the Genelec speaker you measured, can you measure the KH 80 again to correct the record? At the same time could I suggest you try lowering the level a bit as the bass response is probably "suffering" from the protection system activating. It is easy the forget this is a very small box with a very small driver in it and SPL is finite ;-)
Cheers,
Rincewind
 

Sancus

Addicted to Fun and Learning
Forum Donor
Joined
Nov 30, 2018
Messages
617
Likes
1,222
Dear Amirm,
As you have have appear to have solved some of the HF systematic errors in the measurements and got some sensible results from the Genelec speaker you measured, can you measure the KH 80 again to correct the record? At the same time could I suggest you try lowering the level a bit as the bass response is probably "suffering" from the protection system activating. It is easy the forget this is a very small box with a very small driver in it and SPL is finite ;-)
If you read the first post thoroughly, you'll see the second measurement was performed at much lower levels and so it established that the small bass dip between 60-100hz is not due to compression. It is present(to a lesser extent) in the S&R measurements somewhere in this thread also. So that was already done.

I doubt Amir still has the KH80 to re-measure again, but the 5-10khz comb filtering doesn't materially affect the verdict that the KH80 is very good(within SPL specs), and it's a known factor, so it doesn't seem worth replacing a new speaker review with a third re-measurement of this known-excellent speaker.

It's just obsessing over graphs at this point. Which I like as much as anyone else, but the KH80 probably shouldn't be re-visited until at least a couple hundred other speakers have been measured. Perhaps as an anniversary thing since the KH80 drama was a forum event in the early stages of NFS reviews. :)

Also, Neumann is allegedly performing their own re-testing and may respond at some point or another.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
26,063
Likes
58,566
Location
Seattle Area
Thread Starter #376
As noted, the samples have gone to the owners. Interesting that we never heard from Neumann. They could have helped clarify the discrepancies given the fact that they could work very easily with Klippel on this.
 
Top Bottom