Question about Measurement volume level vs Measurement results...

[welldun]

Good day all, this is going to seem somewhat basic, but I'm going to ask it anyway.

I'm wondering what the "perfect world" Frequency Response SHOULD be based on the volume (SPL) that is used prior to the actual measurement sweep in REW.

For example:
If prior to the measurement sweep I use pink noise in the Generator and use the SPL Meter to set the AVR volume to a level to say 75dB, and then I run the sweep (all things remaining the same), in a perfect world, should the Frequency Response from that measurement result in a flat response of 75dB across the sweep range (provided the speaker had the capability to reproduce all of the measured frequencies at 75dB)?

If that is in theory what should happen, then would it follow that the real world results could be interpreted to mean that any part of the frequency response that measures above 75dB is simply that same 75dB signal being boosted by the effects of the room?
And if so, then using that same line of reasoning, it would follow that any points along the graph that fall below 75dB are also being caused by the room?

The reason I ask is because if in the above example the 75dB line is where the response should having been, it makes the dips/nulls seem less severe when measured from that 75dB point instead of comparing them to the bumps/peaks that rise above the 75dB line in other areas of the graph.

Using the attached graph, if the Generator was used to set the output SPL to 75dB for the sweep, would it be safe to assume that everything from around 9Hz to about 122Hz is room gain? So that the dip at around 72Hz is not really a null but just less room gain at that point? If so the that would make the response seem better the what that null between 30Hz and 60Hz would otherwise suggest.

 

[skid00]

A 'perfect', imaginary speaker, that produced the exact same loudness at every frequency, *** in an anechoic chamber ***, would exhibit peaks and nulls at varying frequencies, in any room. Those peaks and nulls would vary in frequency as objects moved around the room (chairs, people, etc).

Your measurements indicate that you might want to reduce the room gain around 20 Hz, 65 Hz, etc.

Before you do that, you need to know that moving the mic even 1/4 inch/centimeter will significantly change your graph. Research the 'moving microphone method' for the easiest way to get better bass response at your listening position.

 

[welldun]

A 'perfect', imaginary speaker, that produced the exact same loudness at every frequency, *** in an anechoic chamber ***, would exhibit peaks and nulls at varying frequencies, in any room. Those peaks and nulls would vary in frequency as objects moved around the room (chairs, people, etc).

Your measurements indicate that you might want to reduce the room gain around 20 Hz, 65 Hz, etc.

Before you do that, you need to know that moving the mic even 1/4 inch/centimeter will significantly change your graph. Research the 'moving microphone method' for the easiest way to get better bass response at your listening position.

Thanks for the reply and suggestions, much appreciated.
The actual measurement results for my setup are much better than the graph that I posted. I mainly shared that graph to add a visual aid to my question.
I understand that the room (its size, items in it, mic position, speaker placement, etc) will cause the variations that are present on that graph, the main question I have is about the expected SPL level for the FR in a perfect scenario.

I'm trying to figure out if the 75dB SPL that was set using the pink noise generated via REW should also result in a 75dB sweep if all settings that lead to the 75dB pink noise SPL remain the same. In other words, in a perfect world, when using REW, should the two SPLs be the same?

Here is what lies behind that question...
In the past, when I've seen an FR graph like this one being discussed in the forums, many of the comments seem to suggest that a "better" result would have had the FR measuring flat across the 89dB point (of this graph), which makes the gap/null between 29Hz and 60Hz seem like a problem.
However, if REW really intended that measurement to be flat across the 75dB point (based on the 75dB pink noise SPL that was used to set the measurement volume), then technically everything above 75dB is room gain, which could be cut in order to correct/flatten the FR. And if the "expected" FR of the sweep was indeed 75dB, then this would make the null between 29Hz and 60Hz on this graph seem more accurate since its closer to the "expected" 75dB.

In essence, not knowing what the "expected" SPL level for the FR is supposed to be, could potentially lead to two different ways of interpreting the same graph.

 

[sm52]

Before starting the measurement, REW suggests doing a sound pressure calibration. If you have a separate calibrated sound pressure meter. The readings on it should match the readings when calibrating in REW. At what pressure to adjust the graph, 75 dB or 85 dB, it does not matter much. I would use 85 dB. It matters how the sound pressure at individual frequencies correlates with each other.

 

[AudiocRaver]

I'm wondering what the "perfect world" Frequency Response SHOULD be based on the volume (SPL) that is used prior to the actual measurement sweep in REW.

The "ideal" frequency response curve you are referring to sounds like what we call a "target curve." You ask what it SHOULD be, like there is a universally agreed upon ideal, but there is not. "Flat" is considered best for a recording studio control room, mixing room, or mastering room, with lots of acoustical absorption designed in to help control reflections. In a home listening room or home theater there will usually be some amount of absorptive treatment but probably a lot less than in the recording studio control room, Why?

• It costs money and takes up space and limits decorating approaches
• The room purpose and the way we listen is different - listening and enjoying rather than working at a creative process
• Absorption is a lot easier at higher frequencies (HF) than at lower frequencies (LF)
• Other acoustical principles
• Personal preference

The target curve most would choose for listening room or home theater will usually end at a point lower at higher frequencies (HF) than at lower frequencies (LF). How much? Where are the bendy parts of the curve? Later...

For example:
If prior to the measurement sweep I use pink noise in the Generator and use the SPL Meter to set the AVR volume to a level to say 75dB, and then I run the sweep (all things remaining the same), in a perfect world, should the Frequency Response from that measurement result in a flat response of 75dB across the sweep range (provided the speaker had the capability to reproduce all of the measured frequencies at 75dB)?

That is the kind of measurement one might find in a recording studio, what we call a flat frequency response.

If that is in theory what should happen, then would it follow that the real world results could be interpreted to mean that any part of the frequency response that measures above 75dB is simply that same 75dB signal being boosted by the effects of the room?
And if so, then using that same line of reasoning, it would follow that any points along the graph that fall below 75dB are also being caused by the room?

Correct.

The reason I ask is because if in the above example the 75dB line is where the response should having been, it makes the dips/nulls seem less severe when measured from that 75dB point instead of comparing them to the bumps/peaks that rise above the 75dB line in other areas of the graph.

Using the attached graph, if the Generator was used to set the output SPL to 75dB for the sweep, would it be safe to assume that everything from around 9Hz to about 122Hz is room gain? So that the dip at around 72Hz is not really a null but just less room gain at that point? If so the that would make the response seem better the what that null between 30Hz and 60Hz would otherwise suggest.

The term "Room gain" is the increase in low frequencies when a speaker is placed close to a wall or in a corner, so that term could apply up to about 30 Hz. The peaks around 60 and 80 Hz could be a result of reflections that constructively combine with direct speaker-to-mic signal at the measurement point to cause a measurement peak, but it is hard to say without knowing more.

View attachment 50710

 

[AudiocRaver]

Here is what lies behind that question...
In the past, when I've seen an FR graph like this one being discussed in the forums, many of the comments seem to suggest that a "better" result would have had the FR measuring flat across the 89dB point (of this graph), which makes the gap/null between 29Hz and 60Hz seem like a problem.
However, if REW really intended that measurement to be flat across the 75dB point (based on the 75dB pink noise SPL that was used to set the measurement volume), then technically everything above 75dB is room gain, which could be cut in order to correct/flatten the FR. And if the "expected" FR of the sweep was indeed 75dB, then this would make the null between 29Hz and 60Hz on this graph seem more accurate since its closer to the "expected" 75dB.

In essence, not knowing what the "expected" SPL level for the FR is supposed to be, could potentially lead to two different ways of interpreting the same graph.

Yes, there is a bit of a "chicken/egg" problem, unfortunately, unless you have a USB mic with calibration sensitivity supplied. What is the ideal? If the calibration is done with an analog-output mic with audio interface, the calibration is done with an SPL meter and is affected by the very room you are hoping to measure. If that calibration is done using wideband pink noise, you will usually get results close enough so as not to have to split theoretical hairs.

 

[welldun]

Thank you all for taking time to reply to this thread. Its always nice to get feedback (no pun intended).

I fear I may have overcomplicated the discussion by adding the graph, only because it took the focus off of the main question (which I guess I didn't state clearly).
I used terms like room gain loosely, just to point out that I expect the room to have some effect on the actual measurement results. Which is why I also used the phrase "in a perfect world" just to say that if we removed the room, would the resulting SPL be the same?

At the end of the day, my question is just technical question about the signal strength used to get the SPL measurement from REW.


IF after using the Generator in REW to set the desired output level, we then run a measurement sweep, is the measurement sweep signal being sent at the same signal strength as the pink noise?

 

[FargateOne]

'moving microphone method' fo

Since I always do that when I want to EQ my system, my OCD levels are much more manageable; the FR is less scary and I sleep better! In other word, doing so makes things easier.

 

[John Mulcahy]

IF after using the Generator in REW to set the desired output level, we then run a measurement sweep, is the measurement sweep signal being sent at the same signal strength as the pink noise?

The rms levels of the signals are whatever you set them to. If you use pink noise to set an SPL you should set the sweep level on the Measure dialog to the same level you set for the pink noise signal.

 

[welldun]

The rms levels of the signals are whatever you set them to. If you use pink noise to set an SPL you should set the sweep level on the Measure dialog to the same level you set for the pink noise signal.

Thanks John.

I was curious because sometimes I see people asking what they should use as a point of reference (SPL wise) from which to make corrections when reviewing their own FR graph (as if to say that they don't know what SPL range the FR should have followed).
When I see those question, in my mind, I'm thinking the reference point should be whatever the sweep SPL was set to.

With that said, I'm still wondering about which parts of the FR on a graph should be considered a null?


If we use the graph that I provided earlier as an example, and say that the sweep signal used was set for 75dB.

Should that area in the graph between 29Hz and 60Hz be considered a null despite the fact that it measures above the 75dB level that the sweep was set to?

 

[skid00]

You don't need to worry about the SPL your speakers are producing, when EQing them. Other than clipping the amp/dac, frying the speakers, or having a sweep at a level barely above room noise, it shouldn't matter.

You just want your sweep measurement to follow the 'room curve' you want. For example, I'm running this curve at the moment:

And the result:

The result doesn't follow my desired room curve because I'm EQing a curve measured 12 feet away from the speaker (with -lots- of room/furniture interaction), but this last curve was measured at 45 inches away.

 

[AudiocRaver]

With that said, I'm still wondering about which parts of the FR on a graph should be considered a null?

If we use the graph that I provided earlier as an example, and say that the sweep signal used was set for 75dB.

Should that area in the graph between 29Hz and 60Hz be considered a null despite the fact that it measures above the 75dB level that the sweep was set to?

Turn off smoothing temporarily (go to 1/48th octave smoothing), and look for SHARP notches.