REW and acoustics - help needed please

[Jventer]

Can someone please point me to a resource to teach me how to interpret waterfall graphs.

 

I am trying to measure my room and fix it. Any help will be appreciated.

Edited March 7, 2017 by Jventer

[adamg]

@Jventer have you seen this from Paul Spencer. Setting up the axis scales properly helps if you haven't sorted that out.

http://redspade-audio.blogspot.com.au/2011/05/rew-understanding-decay-and-waterfall.html

 

I am in the same boat. Plenty of discussion threads out there including this SNA thread. A lot of it is bass focused. I'd appreciate seeing some example plots with basic commentary such as 'this room is work well except for X and Y'.

 

The GIK videos are OK for getting a feel for things, including this one which is a basic REW tutorial. You may have seen this stuff?

 

Can you post some of the plots you've generated. Some examples of my output are posted here.

 

I'm going to invest some more time into it so if I find anything worthwhile I'll let you know.

[Jventer]

2 minutes ago, adamg said:

@Jventer have you seen this from Paul Spencer. Setting up the axis scales properly helps if you haven't sorted that out.

http://redspade-audio.blogspot.com.au/2011/05/rew-understanding-decay-and-waterfall.html

 

I am in the same boat. Plenty of discussion threads out there including this SNA thread. A lot of it is bass focused. I'd appreciate seeing some example plots with basic commentary such as 'this room is work well except for X and Y'.

 

The GIK videos are OK for getting a feel for things, including this one which is a basic REW tutorial. You may have seen this stuff?

 

Can you post some of the plots you've generated. Some examples of my output are posted here.

 

I'm going to invest some more time into it so if I find anything worthwhile I'll let you know.

Thanks @adamg

I have spent many hours this week doing sweeps and reading REW stuff.

I have a basic issue, I just want to improve my room for 2 channel. It looks like a lot of the info relates to DSP and HT.

I am grappling to understand all the terms and jargon. 

Will post graphs and look at your post.

 

[Jventer]

I have now read a lot more, including what @adamg has referred me to.  

My interpretation is that I have lots of issues, including bass issues, modal ringing etc.

Big problem is that I still don't understand that much. I really need help please.

 

Room is 4.4 m wide, 4.6 m long, 2.4 m ceiling. Speakers are at this stage about 35cm from the wall. Listening position is 3.5m from speakers. Left wall has a bay window providing space to the left. Right wall has an open arch way opening into another room. Back wall has an opening as well, basically opening into an entrance and stairs going to second floor. 

 

I can do lots of things in the room if I know what I am supposed to do. It is a lounge, HiFi and TV room that will become HT room too.

 

Speakers: Osborn Eclipse Tower Reference. That is 3 way with extra low range.

 

I have lots of graphs and would like to start with these. Measurement is at the listening position. Both speakers together.

 

@adamg, @svenr, @davewantsmoore, @KenTripp - your help will be much appreciated.

Edited March 6, 2017 by Jventer

[Jventer]

Edited.

 

Edited March 6, 2017 by Jventer

[davewantsmoore]

For the presentation of data in REW, familiarise yourself with the "limits" button.    The handy thing about limits is that it defaults to your last setting, and a convenient way to ensure your charts have the same scale  (making them much easier to read/compare)

 

 

General rules:

 

• Where you place the microphone is important.   I assume it's in the listening position for now.
• If you want to take two measurements to be "compared" with one another.  Then the microphone needs to be in exactly the same position.
• If you post charts where the mic is in different positions, then make sure you say this.
• Just measure one speaker at a time (eg.  only L, or only R)

 

Your data shows:

 

• You have a very significant 50Hz hum
• You are probably capturing the output of the L and R speaker together (!?), and/or you have quite a live room, that will benefit from some extra absorption.
• You could do with some EQ for the bass

[Jventer]

@davewantsmoore

Thank you so much for the reply. Much appreciated.

Yes, I have measured the two speakers together. 

Measurements are at listening position. 

 

Interesting the 50Hz hum then again at 100hz and repeating at some 50 and 100hz intervals.

 

What is happening at 400hz?

 

[Guest]

Which listening position@davewantsmoore ? Most people have two ears and the brain doesn't just 'sum' the inputs, or everything would sound 'combed' and 'flanged' to everyone. For a microphone to 'hear' a speaker as a human does there has to be two microphones at ear distance apart, and they have to be power summed, not just have acoustic pressure amplitude and phase added together.

 

@Jventer  are you talking about audible prominences at those frequencies, or are you seeing an artefact of combining multiple sources in your plots, namely multiple speakers which may be one speaker and its virtual images in the room, not necessarily two 'real' speakers.

Edited March 6, 2017 by Guest

[davewantsmoore]

The dip is likely caused by a reflection (eg. from the floor) ....  the reflected sound, is exactly out of phase with the direct sound at 400Hz, and they cancel each other out.

 

You can test this theory, by changing the location of the microphone.    Make it a drastic change to illustrate the effect (like move the mic to be 30cm from the speaker).   You should see that the dip disappears, or moves to a different frequency.

 

8 minutes ago, Jventer said:

Yes, I have measured the two speakers together. 

 

Setup REW to output to just one .... or just disconnect one.

 

[davewantsmoore]

1 minute ago, johnmath said:

For a microphone to 'hear' a speaker as a human does

 

That isn't what is being attempted here.

[Jventer]

3 minutes ago, johnmath said:

Which listening position@davewantsmoore ? Most people have two ears and the brain doesn't just 'sum' the inputs, or everything would sound 'combed' and 'flanged' to everyone. For a microphone to 'hear' a speaker as a human does there has to be two microphones at ear distance apart, and they have to be power summed, not just have acoustic pressure amplitude and phase added together.

 

@Jventer  are you talking about audible prominences at those frequencies, or are you seeing an artefact of combining multiple sources in your plots, namely multiple speakers which may be one speaker and its virtual images in the room, not necessarily two 'real' speakers.

Just a plain 2 channel layout where I have the mic pointing straight to the front - about midway where my ears were if I sat there. 

I can hear the hot bass. 

 

[Guest]

3 minutes ago, davewantsmoore said:

 

That isn't what is being attempted here.

 

I think I understand that, but isn't the objective to improve the *listening* experience?

[davewantsmoore]

1 minute ago, Jventer said:

Just a plain 2 channel layout where I have the mic pointing straight to the front - about midway where my ears were if I sat there. 

 

That's fine.  The precise positioning of the microphone is relatively unimportant (at this stage).

 

... but if comparing multiple sweeps, then they all need to have had the mic in the same position.

[davewantsmoore]

8 minutes ago, johnmath said:

I think I understand that, but isn't the objective to improve the *listening* experience?

 

First objective is to quantify the speakers in-room response ....  and make inferences about the room acoustics.

 

... and you don't need/want the microphone to "hear like a person" ... you need a regular measurement mic.

Edited March 6, 2017 by davewantsmoore

[Jventer]

[Guest]

1 hour ago, davewantsmoore said:

... and you don't need/want the microphone to "hear like a person" ... you need a regular measurement mic.

 

Well that's true in one sense and false in another; and the failure of people to appreciate that nuance has constrained the development of stereo reproduction for decades. But my comments aren't helpful in this thread, so I will shut up now.

Edited March 6, 2017 by Guest

[Jventer]

11 hours ago, Jventer said:

Here are left right and combined results.

Edited March 6, 2017 by Jventer

[Jventer]

Here are a few photos. 

Recently moved into rental, don't like the low ceilings and the paint colour and lots of other things, so planning lots of changes, subject to measurements :).

Do you like my makeshift microphone stand?

[davewantsmoore]

Remove the couch/chair (keep the mic in the same place) and compare .... or first just move the mic forwards ~ 1m

[davewantsmoore]

2 hours ago, Jventer said:

Do you like my makeshift microphone stand?

 

Yes.

 

How loud (subjectively) are your measurement tones?  Some of those readings are quite low.

[Jventer]

8 hours ago, davewantsmoore said:

 

Yes.

How loud (subjectively) are your measurement tones?  Some of those readings are quite low.

Good point, I first started making measurements at a certain level, then went higher, and then last night I forgot about the higher level and went down to the lower level.

I will in the next two days start at the same position but at higher level. Then do measurements at different points as you suggested. 

Thanks a lot for the assistance.

[aechmea]

@Jventer

Back of an envelope calculation (if just floor bounce) ...

3m from speakers 1m above floor;  floor bounce cancellation is at 285Hz (roughly).

 

However, the lounge will most definitely be providing pretty drastic path difference reinforcements and cancellations.  So I would get it well out of the way as @davewantsmoore suggested.

I get all sorts of peaks and troughs simply from my leather chair being in close proximity to the mic;  been there, done that.

 

[OT:  I have often thought that the chair is actually there when we listen so maybe we hear some of those peaks and troughs anyway.  I suppose that when a body is sitting there and is an absorber rather than a reflector, then the effect might not be as pronounced.  Could be measured I suppose.  3 measurements; nothing there, chair only, chair with sitting person.  Compare.]

[davewantsmoore]

4 hours ago, Jventer said:

measurements at a certain level

 

Measure at ~90dB+ in the listening position.    You will find this is quite loud (I would recommend hearing protection if you are close by)

 

51 minutes ago, aechmea said:

OT:  I have often thought that the chair is actually there when we listen so maybe we hear some of those peaks and troughs anyway.  

 

Like you say, working out what is the cause of what you see in the charts is the current mission.

 

Correcting (using EQ) the peaks and troughs caused by a couch has a lot of caveats, and is generally a bad plan..... they'd likely be better off left alone.    Like many people love to quip - humans don't hear like the microphone.

 

[andyr]

Perhaps Dave, John & others could comment on my below statements (as I too am learning how to use REW):

1. if you look at Jventer's last graph and mentally smooth it out ... you get a basically horizontal line @ 45dB below 600hz and at 48dB above 600hz.
2. the lower line extends to 15hz (which is sensational!  ) but there is a steep roll-off from about 16Khz.
3. what do you suggest is the cause of this 16Khz roll-off?  (NB: we'll assume Jv is using a calibrated mic.)  Is it just your typical room absorption?
4. my own smoothed REW plot is tilted - ie. it's a line which is higher at the LF end and lower at the HF end (sorry, I can't post it now bcoz I'm away from home).  This is with the (calibrated) mic at the listening position.
5. yet when I sit down and listen ... I am not overwhelmed with bass.  In fact, my system sounds balanced - and people who have come round to listen ... have agreed. 
6. looking at Jv's (mentally smoothed) graph and comparing it to mine ... I would say Jv's is 'light on' in the low bass ... and too bright in the treble.
7. what do you guys say.

 

Thanks,

Andy

 

[davewantsmoore]

2.    Measuring at 16khz is tricky....   Moving the mic slightly, or surfaces nearby can change the data captured.....   what I see in the chart if that it is within +/- 2.5dB above 1khz  (ie. there is no massive rolloff) .... but even if you did "call" this a rolloff .... it is likely to do with where the mic is, the couch, etc. etc ..... and if you took many more measurements (including close to the speaker, carefully aligned to the tweeter axis) you'd see different things.

3.  No 

4.  Good.

5.  Expected.

6 (and 1)  I see JVs data as basically flat (if you ignore large dips, which are likely modal or SBIR cancellations) .... but agree it should be more tilted like yours  (which is why I commented he "needs some EQ for the bass")

7.    More data is required before some of these inferences can be supported.