My Room acoustic measurements

[Guest scumbag]

Hi,

 

I'm just starting to learn about this stuff. I do have some experience in designing speakers so that helps me understand the terminology and it means I have the equipment already to do the testing.

Room acoustics was something that I have read about for some time but it all seemed a bit daunting and potentially expensive to implement. Well I finished the Triptych 3 way speakers and I have assembled some nice gear to feed them with music but there was something not right. I wrote this post because I wanted to show how you can get reasonable results with not too expensive gear and that even a small amount of treatment can improve things. Hopefully the technical points that I mention here aren't total rubbish and I have understood the things I've read. I'm going to go over a lot of stuff that more experienced guys will well and truly know.  I'm sure there are gaps in my understanding so any pointer are appreciated but I hope that someone who is toying with the idea of doing something like this might get something out of my experiences.

 

In my room it became obvious that the bass was woolly and the treble was splashy and tiring. Measuring the finished speakers using gated measurements (i.e. excluding the effects of the room) shows the Triptychs to be very smooth - I designed them that way and used drivers that would give me a good result without having resort to a complex crossover. 

Due to domestic reasons (a severely Autistic daughter) I don't get the luxury of listening to much music at home so I have my stereo and speakers set up in my office. A dream job  The room is about the worse case scenario. 4.1m long, 4.0wide and 3.9m high and with highly reflective surfaces backed with either brick or sandstone walls (it's a converted Bank building) and I have a 650mm deep desk that runs around most of the room except at the point where the doors opens into the main office area. I have 2 windows in the room and the only absorbing surface is the floor which has carpet tiles.

Pretty bad for a listening room really.

 

I bought two HOFA baby bass traps - https://hofa-akustik.de/en/modules/basstrap/details/ - I didn't get the large ones as they would simply take over the room and the babies fit under the desk running around my room's perimeter.

I then got hold of four of these - http://soundacoustics.com.au/product/architectural-panels/ha600-75-hybrid-absorberdiffuser/?removed_item=1 -  the beauty of these panels is that they double as either reflective or absorptive panels. Keep the plywood or plastic curved panel is place and you have a convex reflector and remove the panel and you have a broad-spectrum absorber. I settled on using 2 of the panels to the sides of the room to reduce the early reflections and then 2 on the wall behind my listening position to provide a diffused sound. The idea being to reduce "the room" and hear more of the speaker. As soon as I set them up in their respective positions, the listening experience simply improved. Bass subjectively reduced but the over-hang and lumpiness was reduced. Treble was tamed but imaging became sharper. And I did feel that I was hearing the speakers more and less of the contribution of the room. 

I'm sure the hardcore room treatment people will be thinking, you need more bass traps and panels. I'm sure I could get more but I have to be practical here - I'm in an office, not a recording studio!

 

I have tried quite a few tweaks in my system much to the dismay of the DBT- objectivists on this site - indeed some members on SNA have even felt sufficiently threatened by my tweakings that they felt moved to PM to tell me that I was misguided. But I do believe in measuring things - that's the only way to make a set of speakers that measure decently. Measurements inform and guide and listening in the final arbiter in my opinion. So if you think something has improved then it should be easy to measure a change (assuming that our measurement regimes can actually measure the thing that changed!).

 

To make sure that I was on the right track, I fired up my laptop that has Room EQ Wizard (REW) and got my reference microphone and mic' stand and set up the mic' facing down in a location that was the same equated to my ear height in my listening position.

Firstly, I set up my room with no treatments and measured the response of the speakers in my room.

 

Here are the response graphs. First, the full range:

Fairly flat - as I said. They measure at about 25hz-20,000hz =/- 6dB using gated (anechoic) measurements. Things go astray at the lower frequencies though. REW does not ignore the room contribution which sets it apart from the other speaker measurement software that I use. In most of those programs, they "gate" the sound so that the speakers response is removed after the first reflection. This allows us to approximate a anechoic chamber. Conversely, REW takes into consideration the primary sound and the reflections in the room after that. Which is essential if you want to understand how the room is reacting to the sound from your speakers.

 

Then the 20hz-700hz range in closeup.

 

The bass region is very lumpy. If we assume 60dB in the measurement as the reference, we are looking at -10dB dips and +10dB peaks. Or +/-20dB from reference. Yicks. And that's without looking at the effects of reverberation into the high frequencies. 

 

I then added the bass traps and measured their effect with them directly behind the speakers and then in the corners of the room. Interestingly the room nodes reduced reduced most with the bass traps behind the speakers rather than in the corners. 

Here is the "untreated" bass region (red) versus having bass traps directly behind the speakers (blue):

Well it seems the laws of physics are still at work but the peaks and dips are now +/-7dB from reference. There is also a slight smoothing of the bass region - a function of the reduction in the modes. I use Roon and feed my music into a PS Audio Directstream Junior using ethernet. Using REW, I can create a filter to reduce the peaks and even EQ up the dips slightly to flatten this out (dips are less audible though and EQ'ing out dips can tax your amplifier at low frequencies).

 

I then moved to the higher frequencies and measured with the side absorption panels in place (as well as the bass traps) and finally with the bass traps, the side absorbers and some reflective panels to the rear of the room (behind my listening position that is).

Longer RT (reverberation times) are what causes music to sound sloppy in low frequencies and splashy in the higher freq's. The smaller the RT60 the better. I believe that a RT60 of 0.5 or below is regarded as highly desirable.

The absorber / reflectors i used were these - http://soundacoustics.com.au/product/architectural-panels/ha600-75-hybrid-absorberdiffuser/?removed_item=1 - as I said, they are multi-functional so you can insert the panel and it becomes a reflector or remove it to turn the panel into an absorber.

 

Here is the RT60 graph.

Assuming the red graph for the untreated room is correct (I'll do some more measurements as a sanity check), then the RT60 with no bass traps or treatment jumps up to 4.5 in the bass region and spikes at 3.5 at 500hz. It drops down to a more reasonable figure at 900hz-8000hz though.

The fully treated room is dramatically better. It peaks at about 0.7 at 240hz - which is a bit more than ideal but not terrible. From 500hz on though, it hovers around a very respectable 0.5 or thereabouts which is all the better when you consider that I am only using 4 strategically placed panels and 2 smallish bass traps.

 

So the objective measurements tally up quite well with the subjective listening tests. Who would have thought? 

I might get more soundacoustics panels now. I have a feeling that adding some diffusive panels behind the speakers will further remove the room from the equation.

 

Lastly, with EQ in place, here is the predicted bass response:

 

There is no boost applied to the EQ  I've only attempted to reduce peaks. At first when you listen to the setup with this EQ in place, the bass seems to be missing. It's a bit the same as when you add the higher frequency absorbers though. You are hearing what the speakers would sound like if they were in a much larger room which is a more accurate reflection of how they were designed to operate. A bit like removing grunge from your setup with a mains conditioner and feeling that things sound dull even though in reality, you are hearing less pollution and more music.

 

Lastly, the predicted response in room with the same reductions as above but now with a bit of boost (no more than 3dB and no lower than 45hz):

 

Any thoughts or comments would be appreciated.

 

Mark.

Edited July 30, 2017 by scumbag
typo

[HdB]

You might find that dip between the 60 - 80 will respond to a specific bass traps designed for just this narrow band, rather than attempting to use equalisation - not sure if that Variable trap from Vicoustic is this precise and might need a resonance trap (pro-audio Helmholz type)

 

Apart from that, your response graph is remarkable linear - I'd be quite happy with that as long as the mix between direct and reflected sound males it easy to listen too for the whole day - you might ask if Bilbo has used all the diffusers he had cut and if not, maybe borrow a couple to play with - a little goes a long way, surprisingly enough.

 

Lastly, I wouldn't get fanatical about the room having a flat freq response - in the studios, it's obviously necessary, but at home I don't  like a flat response at all - It seems a popular 'thing' these days to use up all the facilities of those dsp units to smooth out all the wrinkles in room response and that tends to suck the life out of the music, as if you'd line cut it - our hearing is quite good at adjusting to non-resonant variations of freq response

 

If you can manage it, a waterfall plot?

[Guest scumbag]

Yes, I'm not too bothered by the dip - I guess one of the Vicoustic traps might work wonders there but so far none have come up on the classifieds in QLD yet. Boo hoo.

Anyway, a bit of EQ fills it in within reason.

Fatigue levels are great over the long term. Choosing a Class-A amplifier and preamp helps here too.

I'm not too fanatical about flat room measurements. I am fanatical about a lot of other stuff but not that. I think it's great that I can get such marked results without too much expense though.

What waterfall plot would you like and what time range would be most useful to look at?

[HdB]

Hey Mark,

 You guys up north are really into 'real room' acoustics - down here, it's still pretty much an unknown language.

 

I think the standard REW contains a good program for doing a 'waterfall' - not sure of the correct name - it's good to illustrate the energy storage/resonances in the room - when measuring with just a few microphone positions, it's not generally a complete picture of what the room is actually doing with music, rather than test signals so your rather remarkable freq response can still be hiding some faults and the reason for trying diffusion rather than absorbtion for to correct dips is that most dips are created by interference either from the boundary effect or standing wave cancelations and the diffusion helps 'spread the sound' and that tends to reduce the dips - sorry for repeating things well known to you

 

It's a bit strange but I've never managed to master the computer 'thing' and testing rooms is still something I get a mate to do for me ( I'll have to get onto this sooner or later, I know) -  it's a really useful way of 'seeing your room' from the point of view of how the energy that you push into the room decays over a set period, especially the low freq area - there are more sophisticated programs available nowadays but the basic ones are much simpler to 'read' and show why 'getting the bass right' is so beneficial - IMO, anyway!

[Guest scumbag]

These are waterfall plots of the "without" in Red and the full treatment in purple - everything is set to REW defaults:

 

And for more clarity - time is set to 1000ms:

[Bilbo]

@scumbag

It's worth reading the Home Theater Shack guidelines about setting the measurement scales in REW.

http://www.hometheatershack.com/forums/rew-forum/934-please-read-posting-graph.html

When measurements are taken at the correct level (75-80dB) and displayed in the preferred scale (45-105dB) your response curve (1/12 smoothing) may not be that flat after all. 

 

[Guest scumbag]

I've measured the speakers using Soundeasy, HOLM and ARTA so I understand how they measure. I wasn't attempting to show the speaker's responses in my post though - only trying to demonstrate how some simple room treatment can affect things. I applied 1/6th octave smoothing as this is regarded as approximating the response of the human ear / brain.

Ironically if  I apply REW's psycho -acoustic smoothing it looks smoother in REW.

https://www.roomeqwizard.com/help/help_en-GB/html/graph.html

Edited July 31, 2017 by scumbag

[marten]

@scumbag

Thanks for your detailed post, including explaining your rationale and the REW data.

I find it a bit difficult to get a clear picture of the waterfall graphs you posted above - any chance you could post the Before and After each as a separate plot rather than combining them both on one plot?  

It could possibly make some aspects of the change effected easier to identify and interpret.

[christosd]

On 7/31/2017 at 11:26 AM, Bilbo said:

@scumbag

It's worth reading the Home Theater Shack guidelines about setting the measurement scales in REW.

http://www.hometheatershack.com/forums/rew-forum/934-please-read-posting-graph.html

When measurements are taken at the correct level (75-80dB) and displayed in the preferred scale (45-105dB) your response curve (1/12 smoothing) may not be that flat after all. 

 

I agree regarding the measurement range comment above.  The rule with any measurement is to only have a y axis on something like a sweep result that reflects the measurement system's dynamic range, and then zoom in from there.  I never chase flat, I just like to characterise different hifi systems and rooms by measuring lots of different SNA'ers systems, and then attribute a low frequency sweep characteristic to the sound I am hearing. It helps keep the fun.

[frednork]

I have been treating my room "by ear" and intend to do something similar. Great thread!

[Guest scumbag]

On 04/08/2017 at 0:41 AM, marten said:

@scumbag

Thanks for your detailed post, including explaining your rationale and the REW data.

I find it a bit difficult to get a clear picture of the waterfall graphs you posted above - any chance you could post the Before and After each as a separate plot rather than combining them both on one plot?  

It could possibly make some aspects of the change effected easier to identify and interpret.

Here it is:

Before:

 

 

After:

 

 

Overlay (again) - the transparancy has been set so that the Red will sit on top of the purple. That is to say, where you see Red (the before), it exceeds the amplitude of the purple (after). So assuming I'm reading things right, the purple is less in amplitude after perhaps 20 milliseconds.

 

 

 

 

 

 

 

 

 

Edited August 5, 2017 by scumbag

[Guest scumbag]

I've got some more acoustic panels on order so I'll add them into the room and play around with positioning to see if I can get more improvement. The surface area of the treated parts is far out-weighed by the non-treated reflective parts here. It's encouraging that a small amount can have some discernible effect though.

I did note from doing some serious listening that I am getting a bit more depth in the sound-stage with things set up the way they are. I am also getting sound-stage height too which came as a real surprise the other day. As indirect / uncontrolled reflections reduce, your speakers become un-entangled from your room acoustics. This all adds to the fun of listening. Less fatigue as your brain does not have to filter out stuff at the sub-conscious level and more insight into the recording. And again, this is a minimal amount of treatment so I can only imagine what an extensive room treatment would do if well implemented!

[marten]

@scumbag:

I think something happened on the frequency scale of the latest two waterfall plots in your thread.  The earlier plots showed the range 20Hz to 400Hz clearly, but the latest two compress that range into too small a plot area to make good sense of. 

I ask because I'm quite interested in seeing what difference your initial treatment made; just the latest plots' scale make it hard.

Edited August 8, 2017 by marten
typo

[Pieface]

1 hour ago, marten said:

@scumbag:

I think something happened on the frequency scale of the latest two waterfall plots in your thread.  The earlier plots showed the range 20Hz to 400Hz clearly, but the latest two compress that range into too small a plot area to make good sense of. 

I ask because I'm quite interested in seeing what difference your initial treatment made; just the latest plots' scale make it hard.

 

The x-axis has been changed to linear rather than logarithmic scale. Best to set on the logarithmic scale so each octave occupies the same distance is my understanding.

[Bilbo]

@scumbag

Mate, you've got to set the "limits" on your measurements properly on a logarithmic scale.  Your earlier posted plots were at least readable but the latest ones are off the scale (excuse the pun)!

Edited August 8, 2017 by Bilbo

[Guest scumbag]

I thought I'd posted these on Friday. i must have gotten side tracked. Hope these are more useful. The first one is the room with no treatment. The 2nd with all treatment in place:

 

[Guest]

Here is a graph from a lounge after intensive treatment to look at, response curve at the time was a little nasty, but things have changed since then

Posted for something of a reference and another graph to look at. Was done by @davewantsmoore about 6 months ago.

More is better, keep treating 

[marten]

Thanks, much better view.   Since you said you noticed a significant sonic improvement, I was keen to see what that might look like on paper.

Not as obvious a change from Red to Green as I expected, but there are changes:

1) From about 100Hz through 2KHz, the "average" reverb time (ringing duration) has definitely been reduced across the board, although not dramatically so. But that change can still be audible.

2) Prominent ringing spur around 45Hz - big change from Red to Green.

Now it's close to 50Hz which may/may not be related to the 50Hz AC in equipment.  Amikel, at 45Hz, do you agree it's unlikely to be the "AC effect"?  If it was, it should really have been present in the Green as well.

 

Since it's really difficult to make treatments work as low as 45HZ, I'm a bit at a loss about the big change from Red to Green. 

If anyone with more expertise/experience in interpreting these graphs could chime in to explain that one, it would be nice and we can all learn something.

[Jventer]

@almikel @davewantsmoore ?

 

[Guest scumbag]

20 minutes ago, marten said:

Thanks, much better view.   Since you said you noticed a significant sonic improvement, I was keen to see what that might look like on paper.

Not as obvious a change from Red to Green as I expected, but there are changes:

1) From about 100Hz through 2KHz, the "average" reverb time (ringing duration) has definitely been reduced across the board, although not dramatically so. But that change can still be audible.

2) Prominent ringing spur around 45Hz - big change from Red to Green.

Now it's close to 50Hz which may/may not be related to the 50Hz AC in equipment.  Amikel, at 45Hz, do you agree it's unlikely to be the "AC effect"?  If it was, it should really have been present in the Green as well.

 

Since it's really difficult to make treatments work as low as 45HZ, I'm a bit at a loss about the big change from Red to Green. 

If anyone with more expertise/experience in interpreting these graphs could chime in to explain that one, it would be nice and we can all learn something.

The intensity of the peak is being reduced due to the fact that I introduced 2X HOFA Baby Bass Traps after the initial measurement, the peak hasn't moved though, it's still at around 42hz throughout so nothing odd there, I don't think. The level at 50hz is higher but it is on target with the mean level that I require so perhaps it just means there is less destructive cancellation of modes at that point??? Guessing here.

 I have a screencap of the room modes that REW has calculated for my room that might shed some light on this.

 

I know this is not a definitive result as I am treating a really bad room with a minimal amount of material. 

 

[Guest scumbag]

11 hours ago, 125dBmonster said:

Here is a graph from a lounge after intensive treatment to look at, response curve at the time was a little nasty, but things have changed since then

Posted for something of a reference and another graph to look at. Was done by @davewantsmoore about 6 months ago.

More is better, keep treating 

That looks a lot better than mine!

Have you got the "before" waterfall by any chance?

Edited August 14, 2017 by scumbag

[marten]

Thanks for pointing out the use of the Hofa baby bass traps.

 

Although I could not find the specs for the "Baby" trap, it's larger version (the "Hofa Bass Trap") has the following graph:

 

 

Given the limited effect under  60Hz, I am still a bit perplexed about how that 45Hz ringing spur was so dramatically reduced in the Green plot.  I get that the absolute height of the peak hasn't shifted that much, but I thought the amount of ringing at 45Hz in the Red plot could muddy the bass appreciably.

[Jventer]

Dont have that much experience but agree with @marten- good results re the ringing.

 

[almikel]

2 hours ago, marten said:

Thanks, much better view.   Since you said you noticed a significant sonic improvement, I was keen to see what that might look like on paper.

Not as obvious a change from Red to Green as I expected, but there are changes:

1) From about 100Hz through 2KHz, the "average" reverb time (ringing duration) has definitely been reduced across the board, although not dramatically so. But that change can still be audible.

2) Prominent ringing spur around 45Hz - big change from Red to Green.

Now it's close to 50Hz which may/may not be related to the 50Hz AC in equipment.  Amikel, at 45Hz, do you agree it's unlikely to be the "AC effect"?  If it was, it should really have been present in the Green as well.

 

Since it's really difficult to make treatments work as low as 45HZ, I'm a bit at a loss about the big change from Red to Green. 

If anyone with more expertise/experience in interpreting these graphs could chime in to explain that one, it would be nice and we can all learn something.

agree that at 45 Hz is won't be power supply, as that would be on both and be much closer to 50Hz

...did the microphone move at all between measurements? if so that will explain the difference

Mike

[Guest scumbag]

I kept the mic' as still as I could. I guess I might have at some point moved it a tiny amount so that would perhaps account for a small change.