Graphs from REW

[Mitchibo]

That is an example of how I learned to look at resonances. Peaks and dips are all resonances, when they can be attributed to room dimensions. The predicted room mode for a 20' wall is 56 hz. (1131/20= 56.55) Doors and windows allow flex in boundaries which can lower the actual resonant frequency (mode) from predicted. Your measurements indicate that resonance is actually most active at 45 hz. That is the center of the peak, which would be the frequency used in creating a DSP Filter.

Next, you have to measure the bandwidth, Q. That is the width of a resonance, measured in Hz. That resonance appears to be 11 hz wide, so Q= 11

Amplitude is how many dB above or below the relative (average dB across operating range) sound the offending frequency is playing. You would apply however many dB is necessary to bring that resonant (mode) frequency back to the relative level.

Post #41 and #99 show steady improvement, but without knowing what inputs you've made, any other suggestions would be guesses.

The other part I mentioned is I do not know what resolution you are measuring in. For bass, it really has to be 1/24 octave with phase, with and without room correction. to ensure you capture the exact amplitude and Q of said resonance. Lower res can mask offending frequencies. So that 45 hz resonance, could be higher in amplitude, or have a different bandwidth than is indicated in that measurement. Being sure of that, saves time on the guessing of amplitude and Q.

Yet another part with multiple subs is ensuring they are all playing at the exact same time. I've only setup a few multi sub rooms, but I've never NOT needed to adjust the delays, even with identical subs with identical cable lengths. And this is where I am in my learning: I do not know how to confirm subs that are not playing together in measurements.

For that, I listen. I use music with strong punchy drum playing (Rush) and really strap in for critical listening. Turning my head side to side (intramural cross correlation - your ears are not symmetrical, which means your own physiology can interfere with how and when you hear certain frequencies) and moving between seats I listen for subs that can be easily located (bass is omni directional, you should not be able to locate the source of the bass, if you can, it's too loud) and try to achieve the best balance at all seats. The delay of each sub should be set so that when a drum should sound snappy, it does. It's immediate, and does not rumble or resonate.

*Disclaimer: in my experience, the Q being measured in single Hz only seems to apply to frequencies within the subs operating range. Minidsp can be used between preamp and separate amp for passive speakers as well. Subwoofers have circuits to protect them from being over driven, where passive speakers do not. Anyone applying my advice about treating subwoofer frequencies to higher frequencies should be aware that they need to be certain of what they are doing as their passive speakers will not be protected from being over driven if the amplitude of any frequency above the subwoofer, is increased.

Thanks for slowing down for me. I am beginning to understand what you are saying here. Thanks for sharing.

 

[carole5]

May I ask which tool do you use to create the graphs?

 

[PENG]

I am curious to know if there are more than a few, e.g., 3-5, 5-10, >10? on this forum are using REW, Umik-1 mic or similar. I have plotted graphs that show the differences (only what REW can measure) between preamp, amps, using the same speakers and wonder if others are interested.

 

[Pogre]

May I ask which tool do you use to create the graphs?

REW and a Umik-1 microphone.

 

[Mitchibo]

Provide all the details on any inputs you've made, the dimensions of your room, placement of subs and their relation to the LP. Would be useful to know distance from walls to mic, as that distance will measure as a dip in the frequency response. The wavelength matching distance will cancel out (1131fps/12 feet= 94 hz, for example) Looks like you're using the correct resolution. With your stated goal, I would include measurements up to 1khz so we can confirm the effectiveness of DSP inputs to the subs, with how well higher order reflections are eliminated.

I made a comment in my above post about setting delay. If you find something else that works for you, please share!

So I'm no the sharpest tack, but if I understand what you are saying (and I've read about this in the mini and REW instructions but it seemed complicated) you are creating a file to use in a DSP by making physical room measurements and via SoS/d you are predicting where the peaks and troughs will be before you take mic measurements. Even if this done with degree of accuracy one will still load it in the DSP run a measurement thereby, hopefully, find a positive results,right? I now have better grasp of Q and how obviously important this is too.

 

[TheWarrior]

So I'm no the sharpest tack, but if I understand what you are saying (and I've read about this in the mini and REW instructions but it seemed complicated) you are creating a file to use in a DSP by making physical room measurements and via SoS/d you are predicting where the peaks and troughs will be before you take mic measurements. Even if this done with degree of accuracy one will still load it in the DSP run a measurement thereby, hopefully, find a positive results,right? I now have better grasp of Q and how obviously important this is too.

The DSP is allowing you to modify the signal before amplification, to specific frequencies. Predicting room modes makes looking at actual measurements a lot easier, in addition to understanding how the room influenced the acoustic measurement. You are using the DSP to match the resonance, with opposite amplitude and equal bandwidth(Q), to cancel it out. NOT to bring all subwoofer frequencies up to the relative level, as many seem to be doing. You'll get more bass for sure, but forget about accuracy.

There seems to be a great misconception that humans have linear hearing, like a microphone. Arbitrarily adding DSP filters to make a smooth line is to assume that's what you'll hear. And then after the subwoofer crossover, you're right back to peaks and dips. The heavily boosted subwoofer frequencies seem to keep having a steep roll off at the LFE crossover. Oops.

 

[carole5]

REW and a Umik-1 microphone.

Thanks for the suggestions. Btw, is it difficult to use these tools? I did some research and found this list of tools. So... I am not sure which one will be the best for me because I do not have any previous experience in these tools. Maybe you can suggest what will be better for me?

 

[Pogre]

Thanks for the suggestions. Btw, is it difficult to use these tools? I did some research and found this list of tools. So... I am not sure which one will be the best for me because I do not have any previous experience in these tools. Maybe you can suggest what will be better for me?

Oh boy, I've never used any of that. What exactly are your goals with making a chart? The graphs we've been making are generated by using the mic to measure the frequency response in room. You play a test sweep and the program (REW) generates the graph based on what the Umik mic picked up from the sweep.

*Edit: Heres a link that walks you through the process if you're curious.

 

[carole5]

Oh boy, I've never used any of that. What exactly are your goals with making a chart? The graphs we've been making are generated by using the mic to measure the frequency response in room. You play a test sweep and the program (REW) generates the graph based on what the Umik mic picked up from the sweep.

*Edit: Heres a link that walks you through the process if you're curious.

I just want to practice, nothing special. In overall, I like to develop new skills in myself. I will try with this one which you have suggested and with some others to see what do I like more. Thanks for your help

 

[everettT]

Rew with Dayton UMM6 and sometimes Earthworks M23 (tascam us2x2 pre) with win7 I5 8g laptop using HDMI out.

 

[carole5]

Thank you for info!

 

[agarwalro]

You guys are correcting the FR to flatten it. This is only the first step.

Warrior has a valid point about notching out room modes peaks. Left unchecked, the energy corresponding with peaks will not dissipate (high RT60 time). This results in the "lose" and "muddy" feeling. Conversely, no amount of boost supplied to a room mode corresponding with a null will pull up that band.

Thankfully, REW offers a very simple way to evaluate the room mode interactions. Check out the Waterfall Plot option therein.

More on generating and reading REW Waterfall Plots:
https://www.roomeqwizard.com/help/help_en-GB/html/graph_waterfall.html
http://www.gikacoustics.com/understanding-decay-times/

 

[TheWarrior]

You guys are correcting the FR to flatten it. This is only the first step.

Warrior has a valid point about notching out room modes peaks. Left unchecked, the energy corresponding with peaks will not dissipate (high RT60 time). This results in the "lose" and "muddy" feeling. Conversely, no amount of boost supplied to a room mode corresponding with a null will pull up that band.

Thankfully, REW offers a very simple way to evaluate the room mode interactions. Check out the Waterfall Plot option therein.

More on generating and reading REW Waterfall Plots:
https://www.roomeqwizard.com/help/help_en-GB/html/graph_waterfall.html
http://www.gikacoustics.com/understanding-decay-times/

To clarify:

Room modes do not have to measure as peaks.

The transition frequency is related to the dimensions of the boundaries in a room. It explains when and where bass stops being omni directional as the sound steadily gains forward bias as frequency increases.

That is what is unique about small room reproduction - small doesn't mean bedroom, it means small compared to an auditorium where the physical dimensions are larger than the lowest frequencies being produced. That's where Reverberation Time really comes in to play.

@agarwalro Absolutely right about the energy not dissipating. When predicting room modes, and confirming them with acoustic measurements, you then calculate the higher order reflections that will extend well above LFE and upper bass, in to the mid range. This is why I measure up to 1khz even if I am just dealing with subwoofers. The effectiveness of DSP inputs will be reflected in higher order reflections being eliminated.

If you will allow, and not feel as though I am 'upsetting' your cheerios; Waterfalls are very pretty, but they tell a very narrow part of the story due to the requirement to choose between resolution in time vs. frequency, you can't have both. As loudspeakers are minimum phase devices, a high-resolution (1/24 octave no smoothing) steady-state response measurement with phase across the listening position(s) is easier to work with.

We've got rain coming this later week, so I will try to get a demonstration put together to show predicted room modes correlated to measured response. I've added a new entertainment center in my room, and it has audibly changed my bass response. Show, rather than tell, I suppose.

 

[ATLAudio]

You guys are correcting the FR to flatten it. This is only the first step.

Warrior has a valid point about notching out room modes peaks. Left unchecked, the energy corresponding with peaks will not dissipate (high RT60 time). This results in the "lose" and "muddy" feeling. Conversely, no amount of boost supplied to a room mode corresponding with a null will pull up that band.

Thankfully, REW offers a very simple way to evaluate the room mode interactions. Check out the Waterfall Plot option therein.

More on generating and reading REW Waterfall Plots:
https://www.roomeqwizard.com/help/help_en-GB/html/graph_waterfall.html
http://www.gikacoustics.com/understanding-decay-times/

Depending on the room identifying room modes is a fairly easy task. While looking at time delay graphs are really good, most of the time you can find them with the review of a few sweeps. I usually do these across the front sofa.

That said, I don't know anyone so far (Mitch, Jogre etc) who's complained of coming away with loose and muddy bass after identifying a baseline in REW, and smoothing response. You can do this in REW, in the EQ Window, by clicking 'set target level' which automatically adjusts the level of the target response to provide a good match to the measurement over the range selected for EQ. I usually start here, and tweak it +-1 or 2dB or so, but it's really good at what it does.

In order for high RT60 time to become a serious issue one would have to ignore a very noticeable and serious room mode peak, which are usually the first eliminated. I also have had success in eliminating nulls across my listening area (front sofa area) with boost and distance change, and even with boost alone, so I wouldn't definitively dismiss it. However, you'd be correct to say that you should use it judiciously.

 

[TheWarrior]

Depending on the room identifying room modes is a fairly easy task. While looking at time delay graphs are really good, most of the time you can find them with the review of a few sweeps. I usually do these across the front sofa.

Could you please explain in detail how you 'know' what the room's influence is on a response graph?

 

[agarwalro]

If you will allow, and not feel as though I am 'upsetting' your cheerios;

You can upset the Cheerios all you like, I'm a Honey Bunches of Oats kind of guy .

As loudspeakers are minimum phase devices, a high-resolution (1/24 octave no smoothing) steady-state response measurement with phase across the listening position(s) is easier to work with.

I tried looking this up and couldn't find anything as simple as Waterfall Plots. Could you post a link or few on them.

We've got rain coming this later week, so I will try to get a demonstration put together to show predicted room modes correlated to measured response. I've added a new entertainment center in my room, and it has audibly changed my bass response. Show, rather than tell, I suppose.

I look forward to your approach. Especially, since you say, its results are better than linear corrections through EQ to get flat FR and time domain control of room response with EQ and bass trap/diffuser application.

 

[agarwalro]

after identifying a baseline in REW, and smoothing response. You can do this in REW, in the EQ Window, by clicking 'set target level' which automatically adjusts the level of the target response to provide a good match to the measurement over the range selected for EQ. I usually start here, and tweak it +-1 or 2dB or so, but it's really good at what it does.

What manner of sorcery is this! Time to update my 10 year old copy of REW.

 

[ATLAudio]

What manner of sorcery is this! Time to update my 10 year old copy of REW.

Let me know what kind of fun you have!

 

[agarwalro]

Let me know what kind of fun you have!

Surely will. Probably won't happen before Thanksgiving though .

 

[TheWarrior]

You can upset the Cheerios all you like, I'm a Honey Bunches of Oats kind of guy .

I tried looking this up and couldn't find anything as simple as Waterfall Plots. Could you post a link or few on them.

I look forward to your approach. Especially, since you say, its results are better than linear corrections through EQ to get flat FR and time domain control of room response with EQ and bass trap/diffuser application.

Well then, I'll try to keep things 'sweet' for you.... But believe me, if I could shorten these responses I would! (but it's all condensed info coming from a 500p book)

As for frequency vs amplitude with phase as opposed to waterfall, that is information I am sharing from Floyd Toole's 'Sound Reproduction'. Floyd goes in to it a bit in his CIRMMT video on youtube. But the compromise between frequency vs time is the issue, and can give misleading information that would be wrongly used to input DSP corrections with.

Back to your RT comment: "Voices and musical instruments have limited sound power output, so (concert) halls are kept as reflective as possible to keep loudness high, without interfering with 'intelligibility' of the music. It is a difficult compromise. Typical values for RT in performance venues are in the range of 1-2 seconds, optimized for the music most commonly performed... In sound reproduction, the essential hall reverberation is captured in the recording, so nothing additional is required. Typical RT for domestic listening rooms and recording control rooms are 0.2-0.4 seconds. The reverberation in classical recordings completely overpowers that of the listening room." Pg 281 'Sound Reproduction 3rd Edition'

Point being that RT is not often an issue in small rooms. Understanding the transition frequency is much more important to identifying and treating the naturally occurring room modes or standing waves resulting from the parallel surfaces of your home. As you pointed out, the energy from the first reflection doesn't just dissipate, it keeps bouncing, creating higher order modes that are higher in frequency than the originating modal frequency. That is why I don't recommend adding DSP filters without predicting and verifying room modes, first, due to the near guarantee of unknowingly creating new problems(which may simply, be inaudible). Knowing that the energy keeps going, means you can calculate and KNOW which mode is the issue, and based on it's relation to your LP and your speakers will give you the information needed to constructively or destructively drive those modes.

My 'approach' is outlined in Ch.13 of the second edition, and Ch.8 of the third edition of 'Sound Reproduction - The Acoustics and Psychoacoustics of Loudspeakers and Rooms.' And I'll clarify yet again, my approach has been nothing but information gathering to get started. Applying DSP to create your own idealized graph is a misnomer. It is a complex subject without many simple answers, but gaining a better understanding of the subject of Sound Reproduction enables you to make informed decisions. But you can't do that, without knowing your room's influence.

From Ch. 12 p345 "When one sees a very smooth high-resolution steady-state room curve after equalization, there is a high possibility that something inappropriate has been done, and the sound quality may have been degraded."

When you consider the amount of time being spent measuring and tweaking DSP filters, it is amazing for me to see the resistance to the fact that the physical dimensions WILL correspond to the acoustic measurements. But clearly, my saying that is insufficient for those who haven't read the book. So I will do my best to upload my step by step correlation of predicted modes(and my room has a bunch!) to the acoustic measurements. I have to start a new photo host because all my other threads are picture-less after Photobucket changed their terms.