REW Graph analysis

[pseud]

I have a 15*12*10 (in feet) room with a 5.1 setup.

I ran the REW using a UMIK-1 and a laptop's HDMI output and got the attached info.
https://drive.google.com/open?id=0B8SRo782zUzxVzlQeHF1YldGRnM

Couldnt make out with the graph, the kind of improvements I need to make.

Also, its a borrowed mic, so if I need to re-run again, please advise.

 

[John Mulcahy]

Well, if the mic sensitivity figure is present and correct the measurement was made at a very high level - if it really was that loud you should measure at a much lower level, but perhaps the cal file doesn't have a sensitivity figure and you just need to go through SPL calibration to get correct figures or something else is awry - the measurement info says the input volume is set to zero, should check that and set it to something more sensible - try 53%. Also don't appear to have selected an output device or output.

Looks like a correct measurement despite that, but with some pretty serious resonances - particularly the one at 35 Hz. Try different speaker and listening positions to see if you can reduce those. The room would likely need pretty substantial treatment to tackle them. Is there any treatment (or even furniture) in the room at the moment?

 

[pseud]

Thanks John for your feedback.

First of all, thanks for creating this wonderful software.

I thought the default calibration for UMIK which gets loaded will handle the sensitivity aspects. If not, I will check on the same. If you can help with the settings for the same, it will be useful.

The room has a thick carpet and has acoustic tiles in the ceiling and heavy recliners which cover almost the entire space opposite the front speakers.

The back wall has the screen and the side walls are bare.

I will run it again, as I am sure I cant bear such loud sound as it appears in the measurements and check again.

 

[Todd Anderson]

Can you show a couple of photos of the space?

 

[EarlK]

( To the best of my knowledge ), the UMIK-1 calibration file downloadable from miniDSP is the only one that includes the sensitivity adjustment ( code ).

 

[pseud]

Can you show a couple of photos of the space?

Here you go
https://drive.google.com/open?id=0B8SRo782zUzxV1ZWNDhJTi1MMDQ

 

[pseud]

( To the best of my knowledge ), the UMIK-1 calibration file downloadable from miniDSP is the only one that includes the sensitivity adjustment ( code ).

From the minidsp site I realised that I need to use the 90 degree calibration file if I am running multichannel and face the mic vertically. Could that be the issue? I had used for the earlier readings the cal file based on the mic code in the minidsp site.

 

[EarlK]

If your ( mic ) calibration file is loaded properly into REW ( & that .cal file includes the sensitivity adjustment ) then REW should register proper SPL readings .

It doesn't matter if you use the 0deg or the 90deg calibration file ( either one should display the proper SPL level ).

BTW, you should own your own UMIK-1 mic / it's somewhat farcical to think one will solve all their room issues by taking a single reading with a borrowed mic ( & then soliciting advice from an online forum ) .

I looked at your file & ( sadly ) your room has the acoustics of an empty swimming pool .

I would recommend that you treat at least the front third of all your walls ( surface area ) with 2" deep OC-703 panels ( OC = Owens Corning ) covered in burlap ( if you decide to make them yourself ). That'll help dry out the "splash" above 300 hz and start to improve the intelligibility of your system .

For treatment ideas below 300zh, you'll need to consult a professional outfit like GIK Acoustics to see what products they can provide to you that'll mitigate the mess ( seen ) in the lower frequencies .

 

[pseud]

I did two more measurements. the first one is by reducing the output, using the 90 deg cal and uses the HDMI output of my laptop and uses java drivers.

I measured it multiple times, moved the speakers, kept the door open, no difference in the measurements. So I thought of switching the output from hdmi to headphone.

The second one uses Asio and headphone out to measure the same.

The second curve is remarkably different, so not sure whats wrong.

 

[EarlK]

How-to-Build-Your-Own-Acoustic-Panels

DIY-Acoustical-Panel Build Thread

 

[John Mulcahy]

The headphone out measurement looks like a loopback, response is just the inverse of the cal file. The input isn't coming from the mic.

 

[pseud]

You are right. There was something wrong with the headphone out. Once I reinstalled Asio and restarted I got a similar graph with the headphone out. I also tried Dirac and got very much the same.

I will first try out bass traps to check out the bump at 35 hz, once I sort that out will look at treatment to the wall.
Not sure what to do with the dip around 80hz.

 

[Matthew J Poes]

Can I ask that you share measurement images as well. I usually am on my phone so I can’t open your mdat files.

It would be good to know more about your setup. I am sorry if I missed it, but is this just a pair of L and R speakers or with a sub? Was it taken for just one speaker or both (it should be for just one) and if there is a sub how is it integrated with the main speakers?

What is the room like? Those spikes look pretty extreme and suggest a room with nearly no damping in the low frequencies. That would be unusual as normally the walls and furniture will apply a little damping. If you could maybe take a picture of the room that might help.

Acoustic treatments are very important but should really be applied based on first establishing what problems you are fixing. It’s best to treat acoustics after you have first endured the setup is optimized and everything is working ok.

 

[pseud]

The photos are in this link
https://drive.google.com/drive/folders/0B8SRo782zUzxV1ZWNDhJTi1MMDQ?usp=sharing

Using multichannel

Using two channel

 

[Matthew J Poes]

When taking measurements you don’t want to have more than one full range speaker playing at once. It’s fine to take measurements with say your left speaker and subwoofer (or even ten subwoofers) but you can’t measure the L, c, and R and make the same sense of those measurements. I’m blanking on the term for this, but you get a dip in the midrange that isn’t representative of the speakers response or anything to be concerned with. It’s an artifact of having multiple speakers operating in the stochastic region and interfering.

My suggestion is to post measurements of the L, C, and R separately and at least one of those with sub. If you want to be thorough you can do L+Sub, R+sub, etc.

You have clear evidence of room modes, which is a great learning opportunity. You have an axial room mode at 37hz. You also have a big bump in the response very near 37hz. There is another at 47hz and sure enough a dip at 47hz. There is a mode at 73hz, 75hz, and 83hz and while they may contribute to that 80hz dip I am wondering if it’s just poor integration with the mains. It may need its polarity flipped or adjusted and/or delay adjusted (at low frequencies this is doing the same thing so these are just variations to adjust the phase). My suggestion is to measure a single speaker and just the sub and then that same single speaker and sub together and look at all three. If the 80hz dip is always there then it is possible it’s mode related (or even SBIR).

What is the distance of the main speakers to the front wall? Any chance it’s around 3-4 feet? If so it’s likely SBIR off the front wall. I thought it might have been a ceiling related dip but it actually aligns with a peak so it seems rear wall related. If it is rear wall the trick is going to be to move the speakers closer or farther from the wall. One thing I’ve considered in the past is to mount midbass/subs to the front wall and operate them up to 200hz or so. Then integrate everything so that these interferences are essentially canceled out. Another easy fix is to build a baffle wall, butnthat only works if you have a dedicated space.

 

[pseud]

My initial setup had the speakers a feet from the front wall. I moved them to nearly 4 ft from the wall and 2 feet from the sidewalls, although this will not work practically. I did the tests with both, each spkr, and the two speakers with sub. Let me know if you see something more.

 

[Matthew J Poes]

What kind of speakers are these? What brand and model? That’s a strange high frequency response.

You seem to have some strong dips and peaks in the bass. It is very possible this is being caused by room modes and speaker placement.

I guess my best advice at this point without being there to trouble shoot is that there is probably some work to be done with setup, sub integrstion, and bass eq that would help. The room probably needs some treatments, especially bass traps.

Can you share a waterfall 20hz-200hz, scaled 95db to 25db.

 

[pseud]

The speakers are Paradigm Monitor 9.

Some part of the high frequency response maybe because the room is in the basement with no windows and has thick concrete walls.

Below is the waterfall data.

 

[Matthew J Poes]

Ok great I think I now have enough information. Basically your room is supporting very strong modes. The ridges that look basically like mountains which do not angle down toward the front of the screen are largely going to re-elect minimum phase steady state modal behavior. That 37hz mode can be knocked down with a peq filter. They same for the ~100hz mode.

However to really smooth things out and improve the low frequency response, reduce seat to seat variance, a distributed multi-sub setup would be most ideal. Setup correctly you will like see a large improvement and eq will ultimately be more successful. There are different schools of thought on this, but the Geddes approach I aspouse involves a minimum of 3 subwoofers.

This is totally my fault for not saying anything but I’m the future you should scale the time domain/z axis of the waterfall out to 800ms to 1000ms. This helps to see how the bass is decaying. By having it scales to just 300ms you can see that many of the modes never had time to decay at all. Since it’s preferee to see bass decay under 500ms at any frequency what I see is enough to tell me what the problems are. In the future when you scale it as I suggested you will start to see a bit more detail about decay.

I’ll mention again that some bass traps could also help smooth this out. While you need a lot for best improvement, if you can get two fairly good/large ones that may be a great start.

In terms of highs, at frequencies as high as 10khz it is unlikely that room reflections would have such a pronounced effect on the sound. Typically the response is reflecting primarily direct sound at those frequencies. The room would need to be highly reflective and the speakers poorly placed for early reflections to cause such issues.

 

[pseud]

I ran Dirac processor and see if I can eq some of the issues and got the below graph.

Will test with bass traps over the next weekend. If it shows improvement, will start with two behind the speaker corner and then try the other end.

 

[Matthew J Poes]

View attachment 4518

I ran Dirac processor and see if I can eq some of the issues and got the below graph.

View attachment 4518 View attachment 4519

Will test with bass traps over the next weekend. If it shows improvement, will start with two behind the speaker corner and then try the other end.

That new measurement already looks significantly flatter/smoother. I huge improvement I think. Of course what matters most is, does it sound better to you? What do you think?

The waterfall plot still shows significan ringing, especially at 37 hertz.

The suckout at around 78hz seems to have a higher Q now and I suspect is less audible. If those waterfalls are both from after the Dirac correction then I think it has had surprisingly little effect on the ringing. Because of how bass works in small rooms, typically a reduction in the peak of the resonance has the same effect in the reduction of ringing. This is a minimum phase property. The mixed phase filters of Dirac should act as minimum phase at this Frequency range and thus should have reduced ringing. There are reasons for why this may be the case, that the filters reduce the amplitude of the peak by not the ringing over time. In this case the bass traps and additional subwoofers are really among the only solutions.

None the less, I wouldn’t be too discouraged. It really looks much better now.

 

[pseud]

I didn't take a waterfall after running Dirac. Ignore the second waterfall it was uploaded by mistake.

 

[Matthew J Poes]

I didn't take a waterfall after running Dirac. Ignore the second waterfall it was uploaded by mistake.

Oh I see. Yes then that Waterfall looks right. It matches what we saw before. That mode at 37hz is ringing for a long time. Very powerful mode. The best solution would likely be a PEQ filter to knock it down.

It’s always good to take new measurements including waterfalls after eq because helps see the improvement.

 

[pseud]

Oh I see. Yes then that Waterfall looks right. It matches what we saw before. That mode at 37hz is ringing for a long time. Very powerful mode. The best solution would likely be a PEQ filter to knock it down.

It’s always good to take new measurements including waterfalls after eq because helps see the improvement.

Below is the waterfall with Dirac eq

 

[Matthew J Poes]

Below is the waterfall with Dirac eqView attachment 4530

That's a beautiful waterfall. While I see ringing present at 37hza nd what looks like 95-100hz, it is much better than before. It is likely that the ringing in the 100hz range would only be impacted by acoustic treatment at this point, possibly a very narrow PEQ filter would help. I would guess its not that audible. At 37hz I think bass traps won't have a huge impact, its just too low, but you could always try. At such a low frequency and with such little ringing left, I would actually think you are a good candidate for a tuned bass trap. Gik has such a product!


http://www.gikacoustics.com/product/gik-acoustics-scopus-tuned-bass-trap-t40/
This is a 40hz tuned trap. Now since this modal ringing is not at 40hz and is in fact very narrow, I would consider asking GIK if they can make a custom trap that is tuned down about 3-5hz lower than stock. They advertise they can do this. They may argue that 40hz is close enough, I'd personally push the issue myself. You could try one, but to be honest it takes a lot of tuned traps to attack problems this low. My recommendation is 4 on the rear wall. Again you can always start with 1 or 2 and add on.

Adding broadband absorption in the low frequencies such as the Gik Monster Bass trap and the Gik Soffit trap will not only help at 37hz, but also help at 40-50hz, 60-70hz, and that ringing at 100hz. Adding some of these would also be good. My suggestion might be to use their range limiter membrane versions for anything you are putting on the rear of the room or on the front wall behind the speakers. Anything on the side walls and ceiling can and probably should be full range.

If those are too costly and you are interested in DIY, feel free to ask and I'm happy to start a new thread in the acoustics section to discuss how to create DIY versions of these.