Post Your Frequency Response Curve!

[pjwd]

I agree. 5dB per major division is a good compromise and avoids 'too much data' which would obscure the response trend and generally make the graph harder to read. Graphs like I posted can always be clicked upon to see a full size version.

Another suggestion if we're really getting down into the weeds here - and one I'd like to see - is the posting of the original .mdat files from REW which preserves ALL the data, the good the bad and the ugly for those who want to get into the specifics of RT60, distortion, step response, waterfall plots etc. It's all right there in the .mdat file. It's easy enough to do by sharing the file on your Google drive or whatever you may have.

To show how foolish I am, I'll go first. Mine can be downloaded HERE. Note that my subwoofers are set really high in these plots - normally they're much more moderate in level.

Brave !

I had a look - I am not familiar with REW but that looked like it was not time gated so I presume it includes a fair bit of reverberant info. This has always puzzled me about in room plots - if its long enough to get reliable bass information it is not getting very clean info
I have done ground plane plots at listening position to try and get better bass info and that seems to help a bit by taking out the floor reflection but there is still plenty of them
The REW impulse plot looks different from " normal" - it is hard to see where the reflection points are - I also noticed left and right phase were quite different - I am assuming different reflection points

I quite like a 3 db taper from 1K down to 20K and a bump in the bass of about 5db or so - as you note yours is quite " goosed" in this plot
Do you have have any plans to try and get a more even output - does the speaker measure better in an anechoic type environment ( I measure outside and up in the air for the crossover design but its a big task to haul outside when finished )

Thanks for sharing - its great to see how other folks go about it - I am doing a crossover using convolution filters at the moment and will post when I get down the track a bit ( ie they are half respectable)
Phil

 

[Bjorn]

Very impressive.. assuming that is listening position and not an average

It's at ear height at listening position.

Your range is too broad to discern anything, can you please change to 40-90DB and repost?

Too broad to discern anything?! Note that you can press the image to get it open in separate window where you might find it better read.
Using a 40-90dB scale is way too zoomed in IMO and is more difficult to read. That's ok if one is measuring in nearfield, but for in room meausurement in a living room I think the scale I used is better.

Let's look at one meausurement with different scales or decadent to see. All with 1/24 octave smoothing.

First same decadent as I previously used:


Increasing the scale with one noch and we the following:



Using a 40 to 90 dB scale like you suggest looks like this:


I prefer to use either the top one or the second one if the response is very even. The third one with 40-90 dB scale makes it more difficult to get an idea of the overall response IMO, and will make many graphs look quite uneven when they are in reality not. Obviously one you add more smoothing, but it's very zoomed in. You mileage may vary of course. This is just my opinion and what I prefer.

 

[pjwd]

It's at ear height at listening position.


Too broad to discern anything?! Note that you can press the image to get it open in separate window where you might find it better read.
Using a 40-90dB scale is way too zoomed in IMO and is more difficult to read. That's ok if one is measuring in nearfield, but for in room meausurement in a living room I think the scale I used is better.

Let's look at one meausurement with different scales or decadent to see. All with 1/24 octave smoothing.

First same decadent as I previously used:
View attachment 134163

Increasing the scale with one noch and we the following:

View attachment 134164

Using a 40 to 90 dB scale like you suggest looks like this:
View attachment 134165

I prefer to use either the top one or the second one if the response is very even. The third one with 40-90 dB scale makes it more difficult to get an idea of the overall response IMO, and will make many graphs look quite uneven when they are in reality not. Obviously one you add more smoothing, but it's very zoomed in. You mileage may vary of course. This is just my opinion and what I prefer.

Thanks .. is this an ungated plot with a rectangular window .. or some quasi anechoic measurement and what software are you using
I like that presentation
I use clio 12 and although it's quite good I suspect I could use it a lot better. The manual assumes a much more knowledgable person than me

Phil

 

[Bjorn]

Thanks .. is this an ungated plot with a rectangular window .. or some quasi anechoic measurement and what software are you using
I like that presentation
I use clio 12 and although it's quite good I suspect I could use it a lot better. The manual assumes a much more knowledgable person than me

Phil

It's ungated with 1/24 oct smoothing. And meausement of one single speaker at the listening position in a living room. Distance about 2.3-2.4 m.

The software is REW (room eq wizard), which is a freeware. Many professionals also use REW now and has left Clio and other expensive softwares which were more common in the past.

REW room acoustics and audio device measurement and analysis software

REW is free software for room acoustic measurement, loudspeaker measurement and audio device measurement and analysis

www.roomeqwizard.com

 

[sbo6]

It's at ear height at listening position.


Too broad to discern anything?! Note that you can press the image to get it open in separate window where you might find it better read.
Using a 40-90dB scale is way too zoomed in IMO and is more difficult to read. That's ok if one is measuring in nearfield, but for in room meausurement in a living room I think the scale I used is better.

Let's look at one meausurement with different scales or decadent to see. All with 1/24 octave smoothing.

First same decadent as I previously used:
View attachment 134163

Increasing the scale with one noch and we the following:

View attachment 134164

Using a 40 to 90 dB scale like you suggest looks like this:
View attachment 134165

I prefer to use either the top one or the second one if the response is very even. The third one with 40-90 dB scale makes it more difficult to get an idea of the overall response IMO, and will make many graphs look quite uneven when they are in reality not. Obviously one you add more smoothing, but it's very zoomed in. You mileage may vary of course. This is just my opinion and what I prefer.

Thanks for sharing with 40 - 90DB, much easier to see dips/peaks. Other view you shared is good to see the slope.
Looks very nice, however usually (IME) a drop at ~1-2KHz prevents the top end from sounding a bit hot, especially in rooms with longer RT60s. Just a thought.

A sub would help or possibly a sweet spot and / or speaker placement movement may alleviate your 45 - 48Hz dip. Looks great enjoy!

 

[Ampexed]

Brave !

I had a look - I am not familiar with REW but that looked like it was not time gated so I presume it includes a fair bit of reverberant info. This has always puzzled me about in room plots - if its long enough to get reliable bass information it is not getting very clean info
I have done ground plane plots at listening position to try and get better bass info and that seems to help a bit by taking out the floor reflection but there is still plenty of them
The REW impulse plot looks different from " normal" - it is hard to see where the reflection points are - I also noticed left and right phase were quite different - I am assuming different reflection points

I quite like a 3 db taper from 1K down to 20K and a bump in the bass of about 5db or so - as you note yours is quite " goosed" in this plot
Do you have have any plans to try and get a more even output - does the speaker measure better in an anechoic type environment ( I measure outside and up in the air for the crossover design but its a big task to haul outside when finished )

Thanks for sharing - its great to see how other folks go about it - I am doing a crossover using convolution filters at the moment and will post when I get down the track a bit ( ie they are half respectable)
Phil

I just leave the REW gating/windowing at the stock settings. This plot was taken at the listening position, so yes, there is quite a bit of room in there. These are quite directional horns, so the room influence is mitigated somewhat. I've done measurements at 1 meter which are of course quite a bit 'pure' and not influenced by the room.

This is an active crossover system with voicing filters implemented at line level, before the amplifiers - the response can be adjusted by quite a degree and can be made almost ruler flat, but 'flat' is not the setting I find sounds best. What you are seeing in these plots is how I had it set at the time of the plot for real world listening, to jazz.

I don't know if you've looked at the picture of my room in my profile, but doing so might give a bit more insight into the graph and what is going on acoustically. The mdat file shows my RT-60 which is actually quite short considering the room size (450mS), and is very flat across the frequency range of 300Hz-10kHz. The step response, while polluted by the listening position distance, is basically just a single impulse since the horns are natively time aligned so everything reaches the microphone at the same instant - 1m distance capture shows this much more clearly than listening position.

I really wish more people would post their .mdat files - please!

 

[Ampexed]

You guys must be kidding me, I said, a quick tip, grow up.

Zooming out too far yields no way to assess and help should it be deemed warranted. Next time I'll put 7 smilees and 6 hearts.. ;-)

There is a cure for all that - let's all post our .mdat files which contains ALL the information and we can zoom in to our heart's content.

 

[Bjorn]

Guys. RT60 is a measurement of a well mixed reverberant sound field. This requires a isotropic sound field, where there are no distinct reflections. We don't have that in small rooms, only in very large rooms like concert halls, large churches, etc.

If we treat the room based on RT60, we end up being mislead. I have had to correct rooms that were treated with RT60 a good number of times.

More here:

Lyd & Akustikk | Small rooms and reverberation time

www.lydogakustikk.no

 

[Ampexed]

Guys. RT60 is a measurement of a well mixed reverberant sound field. This requires a isotropic sound field, where there are no distinct reflections. We don't have that in small rooms, only in very large rooms like concert halls, large churches, etc.

If we treat the room based on RT60, we end up being mislead. I have had to correct rooms that were treated with RT60 a good number of times.

More here:

Lyd & Akustikk | Small rooms and reverberation time

www.lydogakustikk.no

In conjunction with waterfall and other plots, it is adequate for our purposes in the context of this thread - and more to the point, it is all we have (and my room, while not a church, isn't exactly 'small'). I think you're letting perfection be the enemy of 'good enough'.

 

[pjwd]

I just leave the REW gating/windowing at the stock settings. This plot was taken at the listening position, so yes, there is quite a bit of room in there. These are quite directional horns, so the room influence is mitigated somewhat. I've done measurements at 1 meter which are of course quite a bit 'pure' and not influenced by the room.

This is an active crossover system with voicing filters implemented at line level, before the amplifiers - the response can be adjusted by quite a degree and can be made almost ruler flat, but 'flat' is not the setting I find sounds best. What you are seeing in these plots is how I had it set at the time of the plot for real world listening, to jazz.

I don't know if you've looked at the picture of my room in my profile, but doing so might give a bit more insight into the graph and what is going on acoustically. The mdat file shows my RT-60 which is actually quite short considering the room size (450mS), and is very flat across the frequency range of 300Hz-10kHz. The step response, while polluted by the listening position distance, is basically just a single impulse since the horns are natively time aligned so everything reaches the microphone at the same instant - 1m distance capture shows this much more clearly than listening position.

I really wish more people would post their .mdat files - please!

Thanks for the detailed response .. your room and setup look great.
I understand that the directionally of horns mitigates room reflection to some degree.. have you taken off axis plots at 1m

 

[pjwd]

Guys. RT60 is a measurement of a well mixed reverberant sound field. This requires a isotropic sound field, where there are no distinct reflections. We don't have that in small rooms, only in very large rooms like concert halls, large churches, etc.

If we treat the room based on RT60, we end up being mislead. I have had to correct rooms that were treated with RT60 a good number of times.

More here:

Lyd & Akustikk | Small rooms and reverberation time

www.lydogakustikk.no

I know using reverberation for small rooms is definitionally incorrect and classic formula cannot be applied but no one told sound .. it bounces around our rooms regardless of what it is called and the notion that small room decay is complete within 33ms is not always the case .. it depends on mass and rigidity of the envelope.
There are a lot of unstated assumptions in that document about room construction.

 

[jpwelcher]

Measurement from the listening position using Rew, Minidsp umik-1 mic with 1/3 smoothing. The red line is with a Macintosh MEN 220 and the green line is without.

 

[Jim Smith]

Be careful!

In my experience, when I arrived at a considerable number of RoomPlay clients' homes who had used REW or other measurement techniques, when I listened to the music, sometimes the ME Factor (Musical Engagement) was not bad at all, but the client had simply fallen into listening to the sounds, rather than the music.

 

[sbo6]

Be careful!

In my experience, when I arrived at a considerable number of RoomPlay clients' homes who had used REW or other measurement techniques, when I listened to the music, sometimes the ME Factor (Musical Engagement) was not bad at all, but the client had simply fallen into listening to the sounds, rather than the music.

IME I think a lot has to do with correlating what you see in a graph to what you hear. Flat lines are not the most musical and engaging sound!

 

[Ampexed]

All measurements, whether electrical or acoustic are pretty meaningless in the hands of anyone lacking the training and experience to be able to interpret them properly. REW gives only the roughest indicator of what's going on in a particular room based only on a frequency response plot; there's a lot more to it than that one simple measurement.

 

[Bjorn]

Measurement from the listening position using Rew, Minidsp umik-1 mic with 1/3 smoothing. The red line is with a Macintosh MEN 220 and the green line is without.

View attachment 134937

I would recommend change the aspect to 5dB on the Y axis. 1dB makes makes it difficult to get an overview and almost everything looks bad.

 

[snatex]

All measurements, whether electrical or acoustic are pretty meaningless in the hands of anyone lacking the training and experience to be able to interpret them properly. REW gives only the roughest indicator of what's going on in a particular room based only on a frequency response plot; there's a lot more to it than that one simple measurement.

Please share what other tools and measurements you use in REW to dial in your system. Please also explain how and why you use them. Thanks.

 

[Ampexed]

Please share what other tools and measurements you use in REW to dial in your system. Please also explain how and why you use them. Thanks.

I use the impulse (step) to adjust time coherence (the woofer horn and HF horn can be moved independently), waterfall, spectrogram and distortion, with group delay occasionally. I don't use any of the room simulation capability. The why's of all these would make a very lengthy post. In my electronics engineering I also use many types of measurements beyond the basic ones to zero in on a particular area of interest or to try to find the reason something isn't right - I use REW in a similar way with the types of measurements depending on how the room is shaping up and what the problems are along the way.

 

[Tangram]

All measurements, whether electrical or acoustic are pretty meaningless in the hands of anyone lacking the training and experience to be able to interpret them properly. REW gives only the roughest indicator of what's going on in a particular room based only on a frequency response plot; there's a lot more to it than that one simple measurement.

I agree, me included! I have an interest in the science of room acoustics, but only to a point. In my experience with REW, it is VERY difficult to make a challenging room look good on REW plots without using EQ (which I don’t use, preferring an all-analog signal chain). Once the standard acoustic treatments are deployed, and listening and speaker positions are close, smoothing out FR plots is like whack-a-mole. For me, the waterfall plots are the most important, I assume because my room is fairly small.

I had a room acoustics consultant in for a day in June to optimize my system. He made a bunch of tweaks but I didn’t feel the sound was any better than before. And to add insult to injury, I “unwound” every tweak over the course of a month. Expensive lesson but without EQ, he couldn’t do much.

 

[ecwl]

I had a room acoustics consultant in for a day in June to optimize my system. He made a bunch of tweaks but I didn’t feel the sound was any better than before. And to add insult to injury, I “unwound” every tweak over the course of a month.

Interesting. I wonder if you have the frequency response and waterfall plot before and after the room acoustics consultant’s changes. And whether the acoustics consultant actually made the graphs worse too.