impulse response....

[Hakka]

I started playing around with the impulse response measurements on REW yesterday. 1st graph is impulse response on both subs and both mains, second graph is impulse response of subs only.

I'm not sure if I'm interpreting these measurements correctly or not. Looking at the first graph the RT60 of my room seems to be around 300ms, from what I understand this is ok??

Looking at the second graph, RT60 using subwoofers only the level doesn't get down to -60db. I'm thinking this has to do with the fact that my room has all concrete walls and floor, and no bass absorbers(or should I say no room for bass absorbers). I have also attatched a LF waterfall of mains and subs, as you can see there are some ugly bits below 30hz.

The last graph is one I found very interesting, the red trace is impulse response at the primary listening position, the green trace is the same thing but with me holding the mic approx 3 feet above my head. In this position the mic is above all the acoustic panels in the room and gets all the reflections from the side and screen walls. (I was originally planning on removing the panels for this measurement but couldn't get motivated).

All suggestions welcome.

Harry.

[Hakka]

Here are some more graphs from that session, again I'm not sure if I'm interpreting them correctly. They are energy/time graphs, some at dbfs, some %fs. Looking at the second graph(it has 3 traces right and subs, left and subs, both and subs), I have zoomed in the horizontal axis, are those spikes around 2ms and 3.5ms early reflections from the floor and ceiling maybe?
I'm thinking the first graph shows basically the same thing although its a bit hard to cmpare them with the axis not the same.


I'd really like to understand these measurements better, all input is welcome.

Harry.

[JohnM]

On the 2nd graph in your first post, the reason the response doesn't get to -60dB is due to the signal-to-noise ratio of the measurement. For low frequency measurements it can be difficult to get enought signal energy compared to the background noise levels, so the noise floor might only be 40 to 50dB below the peak of the measurement. You can increase the signal-to-noise by using a longer sweep and/or by using multiple sweeps per measurement. RT60 is a calculated figure based on looking at the linear portion of the energy decay after the initial peak (there are specifications relating to the various parts of the response that are used to fit a line to) and determining from that how long it would take the level to drop by 60dB. It is usually based on band-filtered results though, so that the decay in different parts of the frequency range can be compared. That feature is not yet in REW.

The %FS energy-time graphs are the easiest way to spot reflections for main speakers (the wavelengths for the sub are such that this doesn't really apply in the same way). The spikes after the main peak are due to reflections as you guessed, and you can estimate how far away from the speaker the reflecting surface is by looking at the time to the spike. You can only do this for a single speaker at a time though, the short gap between the spikes on the darker trace that are both at almost 100% are due to arrival time differences between your main speakers, the path from one speaker is slightly longer than the other (by about 7cm) after your receiver's distance adjustments have been applied. The peak at about 60% in the pink trace is a reflection from a surface which caused the sound to have about 24cm further to travel to get to your mic. Is that speaker very close to a sidewall? You work out the distance using (time between the peaks) * 343m as the speed of sound is around 343m/s.

[Hakka]

JohnM wrote:

On the 2nd graph in your first post, the reason the response doesn't get to -60dB is due to the signal-to-noise ratio of the measurement. For low frequency measurements it can be difficult to get enought signal energy compared to the background noise levels, so the noise floor might only be 40 to 50dB below the peak of the measurement. You can increase the signal-to-noise by using a longer sweep and/or by using multiple sweeps per measurement.

I was wondering what the longer sweeps were for.

JohnM wrote:

the short gap between the spikes on the darker trace that are both at almost 100% are due to arrival time differences between your main speakers, the path from one speaker is slightly longer than the other (by about 7cm) after your receiver's distance adjustments have been applied.

Correct, my couch is more comfortable if I sit slighly off centre. I might move the couch to centre my listening position relative to my mains.

JohnM wrote:

The peak at about 60% in the pink trace is a reflection from a surface which caused the sound to have about 24cm further to travel to get to your mic. Is that speaker very close to a sidewall? You work out the distance using (time between the peaks) * 343m as the speed of sound is around 343m/s.

Yes my mains are very close to the sidewalls, my room is very small. I think 24cm would be pretty close. I do have an acoustic panel on that wall but it is only 1 inch thick and not spaced from the wall, I will try modifying the panles and maybe adding some ceiling treatments now that I know how to measure the results.

Thanks for your help John, very much appreciated.

Harry.

[DrWho]

Here are a few chapters from "Sound System Engineering" written by Don Davis that talk about room acoustics and how they relate to what the ETC is showing:
http://forums.klipsch.com/forums/sto....p209_p217.pdf
http://forums.klipsch.com/forums/sto....p218_p227.pdf
http://forums.klipsch.com/forums/sto....p228_p234.pdf
Sorry that the text is a bit fuzzy - I wasn't the one that uploaded them...

One thing I wanted to point out is that RT60 has no meaning in a small acoustic environment because you don't have a dense reverberant field. Instead, you have a semi-reverberant field and it must be treated differently. In other chapters of the book he defines mathematically what determines the difference between a "large room" and a "small room" - the actual number is trivial, but it is on the order of 200,000+ cubic feet...

One of the interesting conclusions of the small room is that the "revererant field" isn't dense enough - meaning that there aren't enough reflections spaced out enough. Instead, you get a lot of early reflections all near the same time, which just muddies up the sound. The first thing you want to do is try to extend the Initial Signal Delay (the time between the initial impulse and the first reflection). This can be done with either absorbtion, diffusion, or careful speaker positioning (usually a combination of all three). Once achieved (haha, so easy to say), the next step is to ensure that the semi-reverberant field approximates as close as possible a naturally decaying reverberant field. To do this you're going to need reflections - which leads to the conclusion that absorbtion at early reflections is a poor approach because it soaks up reflections needed to create a more diffuse field. Instead, you should concentrate on redirecting the sound throughout the room (diffusion) - making sure the sound travels a distance as long as possible before arriving at the listening position.

I hope that made sense...I have absolutely no personal experience with the treatment process, but I've watched it done a few times and have been trying to catch up through reading. But that also means that I don't have as strong an understanding as someone like Davis or any of the guys from Syn-Aud-Con who are playing with this everyday:
http://www.synaudcon.com/

[Hakka]

Thanks for the info Mike, you make some interesting points about small room refelctions. I will have a read of those articles tonight (they might be easier to read on the 2.2meter screen )

Harry.

[Hakka]

JohnM wrote:

The peak at about 60% in the pink trace is a reflection from a surface which caused the sound to have about 24cm further to travel to get to your mic. Is that speaker very close to a sidewall? You work out the distance using (time between the peaks) * 343m as the speed of sound is around 343m/s.

I did some measuring (with a tape measure), the sound coming off the sidewall needs to go approx 40 or 50cm further than the direct sound. When I took the readings in REW I was sitting next to the mic, I think 24cm would be spot on so it seems that reflection was coming from me??

The pink trace is the left main, the purple is the right main and the black trace is both. The mains are symetrically placed in the room so that spike would have been there in all 3 traces if it was coming off the sidewalls. I guess I must have been leaning forward over the computer or something for one of those sweeps.

Harry.

[Hakka]

I remeasured last night using the longer sweep settings. The reflection at around 5ms is the ceiling refelction, I think the ones around 8 - 9ms are off the back wall/ceiling or sidewall/ceiling. My acoustic panels seem to be doing their job as I cant see any reflection from the sidewalls or the screenwall.

The second graph is the same as the first but with more zoom. The ceiling reflection at about 5ms is made up of 4 peaks, could this be attributed to the different distances between the tweeter and woofers from the ceiling?

The level of the reflections seems to be quite low compared to the original sound.

Any thoughts?

[atledreier]

It seems this thread died. I'd like to know more about this stuff too.

Any thoughts?

[DrWho]

Two thoughts on the last measurement posted...

First, it looks like you've got a rather dead sounding room - it would probably work well for movies, but how does it sound for music?

Second, the reflections are certainly attenuated, but they're still specular. If you put some diffusion in those location instead of absorption, you'll see the spikes widen into a bunch of little spikes. This will actually improve the intelligibility and make the room sound more spacious.

[Hakka]

I've rearranged my room treatments and ran about 200 sweeps with REW.

After i posted the graph above I added some black curtains over the ceiling, sidewalls and rear wall, from the front of the room to about half way back, this deadened the room further. Video was good but I ended up removing half of the curtain so it only goes 1/4 of the way back and removed all the rear wall cutains.

I dont have diffusers so I mixed up the arrangement of my DVD racks, which seems to work ok.
I have another rack under construction for the rear, for now there is a wonky stack of DVDs there.

I now have diffusers on the sidewall and backwall primaries(LCR) and put some absorber panels on the ceiling primaries(LCR). There is also an absorber mounted above the rear diffuser to catch a reflection from the centre that bounces off the ceiling then off the backwall.
There are absorbers on the frontwall, one behind each main speaker, they are positioned slightly lower than ear height so they dont do much (if I turn them the other way they block the bottom of the screen). There are also absorbers on the sidewall next to each main. I attached some pics.

I have attatched impulse response graphs for all my speakers. The centre channel seems to disperse sound more towards the ceiling than the others, probably because of the horizontal placement. The reflections around 7 - 9ms are ceiling and ceiling>sidewall.

I also want to add some more side and rear diffusers for that spike in the mains at 9ms, and a better ceiling absorber for the centre.

I might have a go at some of those DIY QRD diffusers for the upper back and sidewalls.

I'm stuck with absorbers on the ceiling, I'm using thumb tacks and diffusers are too heavy if they land on my head.

I will post more graphs as I make changes.

Suggestions are welcome.

Hakka.

[atledreier]

love the CD-cover-based diffusors!

The IR looks better, I think. How does it sound now, compared? I have some strong specular reflections as well, and quite a bit of absorbtion in my room, so I might try some QDR(?) diffusors as well. I think I'll make my own thread, instead of hijacking yours, though. Keep us posted on any progress!

[Hakka]

I've only watched one movie - Apocalypto Bluray (5.1 PCM 24bit 48khz) and it sounded excellent, very pinpoint and detailed.

Hakka.