How accurate is Klippel?

[Marin]

odzivi mjernog mikrofona su ovdje obično i varijabilni

That's true.
Also, that's where many microphone's have like peak in the response and that could explain the peak measured at Erin's.

 

[Marin]

The microphone used by a near-field scanner like a Klippel has much higher frequency resolution than 1/24th of an octave. By much higher I mean many orders of magnitude higher.

The measurement microphone also has much higher amplitude resolution than 1 dB, again by orders of magnitude.

For example, the room in which the measurement is made has sources of uncorrelated noise much larger than the mic's resolution. So the signal is smoothed by integrating the signal over a frequency range of 1/24th of an octave. This is a common practice with measurements, for example simple voltmeters have smoothing so that the needle isn't jumping all over the place while trying to measure voltage. The smoothing is a consequence of high resolution, not the actual resolution of the measurement device.

Thank you very much for explaining so I can finally have some clue..)))
So picture can talk thousand words so bear with me
Raw

Applied smoothing 1/12 octave

 

[MAB]

Thank you very much for explaining so I can finally have some clue..)))
So picture can talk thousand words so bear with me
RawView attachment 511697Applied smoothing 1/12 octave View attachment 511698

Yes.

For example, despite gating used to remove the room reflections, still have lots of reflections. Even the mic stand introduces reflections that color the measurement. One member (@Ageve) posted the impact of improving mic stand. I can't find the measurements, but his setup is pretty optimized for low-reflection. I think Amir has posted numerous times about measurement setup interactions. Some of my posts of speaker measurements have large wiggles due to mic stand. I have some posts trying to make noise measurements on relatively quiet day with no wind in my basement, and someone with a chainsaw in the distance was being picked up by the mic. I can measure my family coming down the driveway with even a cheap mic, well before I can hear them. I could probably train an algorithm to detect differences by the sound of each person's footfall, so long as other noise wasn't dominating. But I don't know much about that type of processing.

In any case, I usually don't worry too much about the noise in the measurement, or the smoothing used, since that same noise is present when I listen to music and my brain does it's own smoothing. In general look for smooth on- and off-axis frequency response, free of evidence the speaker is intrinsically resonating. The good news: when a speaker is misbehaving the resonances show up in the frequency response, the electrical impedance of the speaker, the distortion, the directivity, to name a few. These resonances (like a driver with cone-breakup, or a port that is chuffing or hooting, or crossover filters that introduce resonances between drivers, to name a few) all have characteristic fingerprints that can be deduced looking at the measurements.

People are often shocked when they see their first ungated, unsmoothed measurements. Even worse since they often lack good practices that minimize some of the gross measurement artifacts.

 

[Marin]

I have some posts trying to make noise measurements on relatively quiet day with no wind in my basement, and someone with a chainsaw in the distance was being picked up by the mic.

I'm glad that you are Okay bro)))
Texas chainsaw massacre on mic lol,
I'm still puzzleed how REW is capable measure harmonics that are below noise in my room.(5 fan's on my computer
But I stream directly from AVR through Spotify connect.
I tried Tidal and I noticed slightly more volume but no connect option,so you basically are stuck with App of my reciver or Bubble UMPM App.

 

[Marin]

Hey out of t

Yes.

For example, despite gating used to remove the room reflections, still have lots of reflections. Even the mic stand introduces reflections that color the measurement. One member (@Ageve) posted the impact of improving mic stand. I can't find the measurements, but his setup is pretty optimized for low-reflection. I think Amir has posted numerous times about measurement setup interactions. Some of my posts of speaker measurements have large wiggles due to mic stand. I have some posts trying to make noise measurements on relatively quiet day with no wind in my basement, and someone with a chainsaw in the distance was being picked up by the mic. I can measure my family coming down the driveway with even a cheap mic, well before I can hear them. I could probably train an algorithm to detect differences by the sound of each person's footfall, so long as other noise wasn't dominating. But I don't know much about that type of processing.

In any case, I usually don't worry too much about the noise in the measurement, or the smoothing used, since that same noise is present when I listen to music and my brain does it's own smoothing. In general look for smooth on- and off-axis frequency response, free of evidence the speaker is intrinsically resonating. The good news: when a speaker is misbehaving the resonances show up in the frequency response, the electrical impedance of the speaker, the distortion, the directivity, to name a few. These resonances (like a driver with cone-breakup, or a port that is chuffing or hooting, or crossover filters that introduce resonances between drivers, to name a few) all have characteristic fingerprints that can be deduced looking at the measurements.

People are often shocked when they see their first ungated, unsmoothed measurements. Even worse since they often lack good practices that minimize some of the gross measurement artifacts.

Hi, sorry for going off topic, but since you said you measure using REW tool maybe you could help me with one thing.
I want to measure the near field of each driver and port but I pick up a lot of noise due to sound bouncing off surfaces and returning to the mic .
So I should somehow reduce the default time which is currently at 500 milliseconds.

I have circled the reflections that I would like to remove from the measurement?
Without measuring one by one driver separately.

 

[Maxim-ICOM]

As a short answer to the original question, Kippel states the accuracy ranges from +/-0.1dB (under "ideal circumstances") up to +/-1dB. The latter is right at the border of audibility under very ideal circumstances with steady sine waves, particularly in the presence region.

I would never sweat over 1dB differences in any gear, the most ideal home room probably easily produces +/-2-3dB variabilities.

You will be wondered, but our ability is much higher. Easy test you can do if you run mono signal trough both speakers, sit in a center where you have equal loudness in both ears, then change the volume in one channel by some fractions of dB and you find that need to shift you head few cm to get same equal loudness in both ears.

If we are talking about speaker measurements, I would say that we can hear 0.2db, but not the overall loudness but relative change in frequency response.
If you take 0.2db from highs )add serial resistor to your tweeter (0.1-0.2Ohm) you will hear better bass. But changing whole volume by 0.2db may be unnoticeable.

 

[Marin]

You will be wondered, but our ability is much higher. Easy test you can do if you run mono signal trough both speakers, sit in a center where you have equal loudness in both ears, then change the volume in one channel by some fractions of dB and you find that need to shift you head few cm to get same equal loudness in both ears.

If we are talking about speaker measurements, I would say that we can hear 0.2db, but not the overall loudness but relative change in frequency response.
If you take 0.2db from highs )add serial resistor to your tweeter (0.1-0.2Ohm) you will hear better bass. But changing whole volume by 0.2db may be unnoticeable.

Funny you say that.
I have 22 Ohms resistor in series with my tweeter.
I think it gives him good headroom, meaning he barely sweats while working.
Maybe I replace him with 15 for more air?

 

[Maxim-ICOM]

Funny you say that.
I have 22 Ohms resistor in series with my tweeter.
I think it gives him good headroom, meaning he barely sweats while working.
Maybe I replace him with 15 for more air?
View attachment 511719

Sorry, can not say anything about your crossover - you need to measure speaker so see and better to measure drivers and put them in CAD to simulate your crossover what it will give and then listen the result. Many iterations will lead to desired result

 

[pablolie]

You will be wondered, but our ability is much higher. Easy test you can do if you run mono signal trough both speakers, sit in a center where you have equal loudness in both ears, then change the volume in one channel by some fractions of dB and you find that need to shift you head few cm to get same equal loudness in both ears.

If we are talking about speaker measurements, I would say that we can hear 0.2db, but not the overall loudness but relative change in frequency response.
If you take 0.2db from highs )add serial resistor to your tweeter (0.1-0.2Ohm) you will hear better bass. But changing whole volume by 0.2db may be unnoticeable.

I'd be interested in a source for this. Everything I have ever read indicates ~1dB is at the very threshold of audibiity for humans (and that's under pretty ideal circumstances).

Perhaps if you go from total and utter silence in a perfect room you can hear the 0.2db jump when pitching in a signal in the 2-5kHz presence region, but in any other case I very, very much doubt it, and it would be a pretty limited cornercase, IMO.

And please note I am not advocating we design gear to lower standards.

 

[amirm]

What whoud you say is resolution of are ears.

The discrimination bandwidth of our hearing increases (gets worse) with frequency in a linear manner. See my article here: https://www.audiosciencereview.com/...ds/perceptual-effects-of-room-reflections.13/

As a practical matter, above transition frequencies of a few hundred hertz, you are limited more or less to 1/6th octave. This is what you should do to assess overall tonality.

In bass frequencies though, the modes can be super narrow, even going below 1 Hz! For this reason, the measured resolution I use for Klippel is about 0.7 Hz from what I recall. But this is an overkill to some extent so I usually use 1/24th octave which is pretty close to 1/20th recommendation in the CEA/CTA-2034 standard.

REW has variable smoothing for this reason which you can use but I usually don't as the task of analyzing bass response is different than tonality above transition. So I just select the bandwidth I need for the task at hand.

 

[a4eaudio]

I'd be interested in a source for this. Everything I have ever read indicates ~1dB is at the very threshold of audibiity for humans (and that's under pretty ideal circumstances)...

There is a test somewhere on the internet that allows you to test the deviations you can hear. I believe at the end there is a histogram of the test results. I could distinguish 0.5db and a small proportion of subjects can detect 0.25db. This is pretty easy to test with a Wiim Pro. Set one of the equalizer bands to +/- whatever threshold you want, (e.g., +0.75db at 2kHz), set Audio settings to mono and disconnect one of the speakers so you can listen in mono, and then just turn the EQ button in the App on and off. Try it with Pink Noise, a sine wave, and a song of your choice.

(Im pretty sure Klippel has something similar for testing your sensitivity to distortion, but that isn't the one I'm referring to).

 

[a4eaudio]

Hey out of t

Hi, sorry for going off topic, but since you said you measure using REW tool maybe you could help me with one thing.
I want to measure the near field of each driver and port but I pick up a lot of noise due to sound bouncing off surfaces and returning to the mic .
So I should somehow reduce the default time which is currently at 500 milliseconds.View attachment 511708I have circled the reflections that I would like to remove from the measurement?
Without measuring one by one driver separately.

Those are not "reflections" that you would remove from the measurement. When you set the "window" or "gate" for the impulse response, you set this based on the "first reflection". The first reflection is the closest surface, which is usually the floor or ceiling - and if you have set too long of a window you will see a wave-like behaviour in the measurements - these represent reflections of the sound in the room and thus not an "anechoic" (i.e., reflection free) measurement of the speaker. Buy once you have set the "window" correctly, your measurements above are just the actual driver and port resonances. The resonance you have circled is about 22 dB lower than the main output of 112 dB at 50 Hz which is not particularly bad.

 

[pablolie]

There is a test somewhere on the internet..

OK. That is not ever going to be anything I shall (and nobody should) take as gospel. Chance for a remotely convincing argument sadly passed....

 

[a4eaudio]

OK. That is not ever going to be anything I shall (and nobody should) take as gospel. Chance for a remotely convincing argument sadly passed....

My intention wasn't for anyone to accept that as the answer, but rather if you (or anyone else) were interested enough in knowing the answer, to search the internet and find it.
I have used it and know the answer for my own ears. Also, if you are interested, just do the test yourself with a Wiim or other EQ.

 

[a4eaudio]

From Everest and Pohlmann (2009)

 

[a4eaudio]

My intention wasn't for anyone to accept that as the answer, but rather if you (or anyone else) were interested enough in knowing the answer, to search the internet and find it.
....

It is the first "Challenge" here...https://www.audiocheck.net/blindtests_index.php

 

[MAB]

I'm glad that you are Okay bro)))
Texas chainsaw massacre on mic lol,
I'm still puzzleed how REW is capable measure harmonics that are below noise in my room.(5 fan's on my computer
But I stream directly from AVR through Spotify connect.
I tried Tidal and I noticed slightly more volume but no connect option,so you basically are stuck with App of my reciver or Bubble UMPM App.

The practice of making valid measurements in the presence of noise and distortion is a deep topic. For harmonic distortion measurements, resonances show up as harmonics of the fundamental. Electronics don't go and make up fake signals, or leave part of the signal out as we read in the trade magazines and marketing material. They produce distortion correlated to the input signal, and noise. If you feed a known input signal into your electronics, you can measure the harmonics and the background noise. In this case a 4 Volt 1kHz tone is used to evaluate a preamp.

The harmonic distortion peaks are easy to pick out and separate from the mains noise and the background. I could repeat this test at any frequency I want. If I used 3.14 kHz we would see the distortion at 6.28 kHz, 9.42 kHz, etc. 1 kHz is used for the dashboard since it puts the HD in the most sensitive part of human hearing. Note the floor of my measurement rig connected to this preamp is about -150 dB. If it was worse (around -105dB or higher) none of the HD would be visible since they would be buried in the noise floor of the contraption.

I hope this answers your question.

 

[MAB]

Hey out of t

Hi, sorry for going off topic, but since you said you measure using REW tool maybe you could help me with one thing.
I want to measure the near field of each driver and port but I pick up a lot of noise due to sound bouncing off surfaces and returning to the mic .
So I should somehow reduce the default time which is currently at 500 milliseconds.View attachment 511708I have circled the reflections that I would like to remove from the measurement?
Without measuring one by one driver separately.

VituixCAD has a good primer on making near- and far-field driver measurements with REW

The measurement trace you circled looks like a port. If so, no surprise it is resonating since many ports make out of band noises. If not maybe you are getting drivers bleeding. Close mic distances about 5 cm may help. Disconnecting other drivers while measuring one single unit is also useful. Bi-wire posts can be used to isolate drivers, in fact it's one of the few valid uses for bi-wire posts!

 

[Ellebob]

Another factor with discrepancy of measurements even if using the same exact speaker is environmental. Temperature and humidity can alter a speaker and microphone's performance. This could cause some variations.

 

[amirm]

Another difference between my measurements and Erin is near 20 kHz. I always see this discrepancy. I have cross checked my testing with likes of Nuemann and Genelec so trust my response more. So don't assume that is sample to sample variation. It is likely a microphone calibration issue.