AK Referance Thread: In room response and measurement

[juniorJBL]

A discussion for members doing measurements of in room response, room modes, and speaker placement.

Index

Other measurement threads:

Speaker measurement:
Speaker Box design and Software:


More reference threads:

http://www.audiokarma.org/forums/showthread.php?t=173971

 

[Pete B]

I had mentioned in some of the other threads that I tried Praxis in free demo mode and that the RTA was working well. That was a preliminary test, I've determined that I can't use a mic calibration file and this rules it out for me:
http://www.diyaudio.com/forums/software-tools/161448-praxis-measurement-software-tips-etc.html

It does have some very powerful signal processing options in the full paid version but it is rather costly at over $500.

I'm going back to Room EQ Wizard for room RTA type measurements. It is completely free - no paid version. Preliminary testing seems good so far.

 

[aabottom]

I had mentioned in some of the other threads that I tried Praxis in free demo mode and that the RTA was working well. That was a preliminary test, I've determined that I can't use a mic calibration file and this rules it out for me:
http://www.diyaudio.com/forums/software-tools/161448-praxis-measurement-software-tips-etc.html

It does have some very powerful signal processing options in the full paid version but it is rather costly at over $500.

I'm going back to Room EQ Wizard for room RTA type measurements. It is completely free - no paid version. Preliminary testing seems good so far.

I plan on trying Room EQ Wizard.

 

[Zilch]

Courtesy member Sled108:

http://www.testaudio.com/testaudio/article_Using_Pink_Noise_and_SPL_Meter.asp

 

[RichPA]

In my limited experience, I've been interested in the differences I see in measuring response at the listening position depending on whether it's steady-state response using pink noise or impulse response transformed to the frequency domain. Anyone here have comments or links to help me understand this better?

 

[marc mc]

I would love to see real frequency response graphs from listener positions in real rooms. Anyone care post some?

marc mc

 

[Zilch]

In my limited experience, I've been interested in the differences I see in measuring response at the listening position depending on whether it's steady-state response using pink noise or impulse response transformed to the frequency domain. Anyone here have comments or links to help me understand this better?

Generally speaking, and when rooms are involved, it must be "generally," as each room is different, rooms influence the response in two major ways:

1) Above a transition region in the range of 200 - 500 Hz, depending upon room size, the response tilts downward, for two reasons:

a) the absorptive character of boundaries and objects in the room, generally (that word again) more absorptive at higher frequencies, and,

b) the interference of reflections having different path lengths at different frequencies, smaller differences having greater influence at higher frequencies.

Net result, "Optimum" in-room response with speakers having flat anechoic response looks like this:

2) Below the transition region, the room is in control of the low frequency response as a function of its dimensions, generating reinforcement lobes and cancellation nulls throughout the space, and thus requiring different strategies for both measurement (spatial averaging) and control (multiple sources).

See Toole, Sound Reproduction - Loudspeakers and Rooms at your local library or booksource.

 

[aabottom]

Pink Noise or Impulse Response

In my limited experience, I've been interested in the differences I see in measuring response at the listening position depending on whether it's steady-state response using pink noise or impulse response transformed to the frequency domain. Anyone here have comments or links to help me understand this better?

I'm not sure I understand your question, but here I go.

Measuring frequency response of a system (an audio system in a room in this case) requires instrumentation with sufficiently wide bandwidth. Pink noise and impulse response are two ways of achieving wide bandwidth. Wiki mentions both of these methods (among others) in their entry on frequency response [1].

Pink noise [2] has a frequency spectrum that falls as 1/f, where f is the frequency. In pink noise, each octave carries an equal amount of noise power. For example, 20 to 40 Hz has the same power as 40 to 80 Hz. Ideal pink noise has infinite bandwidth and therefore is impossible to achieve.

An impulse [3] transformed to the frequency domain (that is, the frequency spectrum) has a spectrum that falls as 1/f^0, the power is constant with frequency. Each constant bandwidth section carries an equal amount of power. For example, 20 to 40 Hz has the same power as 40 to 60 Hz. Again, an ideal impulse response has infinite bandwidth and therefore is impossible to achieve.

But at audio frequencies, we can get good enough. From Wiki [3], "In practical systems, it is not possible to produce a perfect impulse to serve as input for testing; therefore, a brief pulse is sometimes used as an approximation of an impulse. Provided that the pulse is short enough compared to the impulse response, the result will be close to the true, theoretical, impulse response."

For example, an impulse can be approximate by a rectangle wave with a short pulse width. A rectangle wave has a frequency spectrum described by the square of the sinc function [4]. But for very short pulses the frequency spectrum is very flat over a wide range. For example, a rectangle wave with a pulse width of 2 microseconds has a frequency spectrum that is flat to within 0.1 dB from 20 Hz to 20 KHz.

Wiki is your friend.
[1] Frequency response.

[2] Pink noise, not to be confused with Pink Floyd.

[3] Impulse response. The first paragraph is not a very good description. Read further down under Mathematical considerations.

[4] Sinc function.

 

[marc mc]

I would love to see real frequency response graphs from listener positions in real rooms. Anyone care post some?

marc mc

Does anyone have one of these graphs of their speakers in their room to post?

Forget the details I would just love to see what the mic picks up from the listening position. The same way we take the 1m test.


marc mc

 

[aabottom]

Does anyone have one of these graphs of their speakers in their room to post?

Forget the details I would just love to see what the mic picks up from the listening position. The same way we take the 1m test.


marc mc

I would, but I don't have a decent mic yet. If I use the cheap mics I have, I suspect the measurement would measure the frequency response of the mics.

 

[Zilch]

Does anyone have one of these graphs of their speakers in their room to post?

Stereophile dubbed this in-room response measurement of JBL K2-S9800 "Exemplary":

Compare to the B&K "optimum" curve in #7, above.... :yes:

 

[RichPA]

Zilch and aabottom, thanks for your replies to my not-so-clear question. I'm aware of the influences of rooms on frequency response and that optimum room response won't be a a flat line. I was also aware of most of the information in the references you cited, aabottom, though it is certainly helpful to have them.

Let me try a more focused question: given the same measurement mic in the the same position, measuring the same system in the same room, are there reasons why the two measurement techniques would give different results? Or would variation between the two approaches simply be experimental error (quality of measurement equipment, small unnoticed variation in mic position, etc.)?

marc mc, here is a screen shot of the measurement of my speakers in my room, as produced by a TacT RCS 2.0. http://www.audiokarma.org/forums/album.php?albumid=525&pictureid=3036
I've since moved them to reduce the bass emphasis in the uncorrected curve (top), but the response from about 200 Hz up is very similar. I'm also using a target curve that is more similar (less mid-bass) to the "optimum" curve Zilch posted above.

Interesting thread, all.

 

[marc mc]

Thanks guys.

marc mc

 

[aabottom]

Stereophile dubbed this in-room response measurement of JBL K2-S9800 "Exemplary":

Compare to the B&K "optimum" curve in #7, above.... :yes:

There seems to resonant peaks at 28, 52, 88, and 115 Hz. These correspond to the 1st, 2nd, 3rd, and 4th harmonics of for a room 20 feet long (28, 56, 84, and 112 Hz). I don't see a unit on the vertical scale. If each horizontal line is 1 dB, then the peaks are only about 3 dB up from the mean. :thmbsp:

 

[auplater]

In room ARTA of my BG-75 Dipoles

windowed to perhaps 150 hz

 

[aabottom]

Zilch and aabottom, thanks for your replies to my not-so-clear question. I'm aware of the influences of rooms on frequency response and that optimum room response won't be a a flat line. I was also aware of most of the information in the references you cited, aabottom, though it is certainly helpful to have them.
.............
Interesting thread, all.

Your welcome. I wrote my reply for myself as much as everyone else. I learned a lot in the process. I first heard about pink noise many years ago, but I did not realise that the spectrum falls off as 1/f.

By the way, the name pink noise arises from being intermediate between white noise (1/f^0) and red noise (1/f^2), more commonly known as Brownian noise.

 

[Zilch]

windowed to perhaps 150 hz

Windowed means MLS or gated sinusoidal, and thus the room is NOT in the plot, because the reflections have been removed, no?

 

[auplater]

reflections with a path length greater than 5 feet or so, yes

 

[Zilch]

Yes, I'm suggesting that's why the plot does not slope downward toward the high frequencies.

If you turn off the windowing, or switch to RTA, it probably will, measured from the same location, even.... :thmbsp:

 

[auplater]

in-room response

Actually, I forced that situation while adjusting the crossover to the planars.

As tested, the sound in room was bright indeed, so I padded down the ATC planar and raised its xover point to 6Khz rather than running concurrently with the RD75's, which run free from ~600 Hz up after 2 pole high pass.

Learned the hard way, not to demo them w/o confirming their condition, as I inadvertantly took them to a DIY session in Lexington,KY Nov '08 in the "brite" condition, rather than re-setting them to a more "mellow" sound...:tears::thumbsdn:

If I can find some quiet time and setup, I'll run them again as suggested.

John L.