REW and room acoustics - what to the graphs tell you?

Posted by: Beachcomber on 25 January 2018

Following a suggestion in a nearby forum, I bought a USB calibrated microphone and installed REW.  Now have lots of pretty graphs - but have no real idea how to interpret them.  On the face of it they don't look great.  The waterfall is very rough and I don't think many barrels would survive it.  The spectograph looks pretty, I guess, but I'm not sure what it means.  What do they all mean, and what can I do to fix any problems - how do I know what I should do?

Waterfall

Posted on: 25 January 2018 by Dave***t

You might find the thread I made about it previously useful.  Especially Huge's contributions.

https://forums.naimaudio.com/topic/working-with-rew (click link).

Posted on: 26 January 2018 by Huge

One approach is to post the graphs on here...

Never mind the embarrassment of showing your defects on here... I won't laugh... Honest... (Actually my room was truly awful 'til I added the bass traps).

Posted on: 27 January 2018 by Beachcomber

OK, thanks for that.  I have rather a lot of plots, but here are just a few:

Impulse Response speakers nearer wall

Waterfall spkrs initial position

Distortion spkrs nearer wall

SPL & Phase spkrs nearer wall

Waterfall spkrs initial position

Spectrogram

From what I can understand it looks like the room is mostly dampening the sounds, except at the 50Hz area, if I interpret this correctly.  There are other graphs as well, such as RT60, Decay and GD (whatever that is).

Cheers

Steve

Posted on: 27 January 2018 by Beachcomber

Hmm  - that didn't work well.  They are on my Flickr page, and I pasted the URL into the window that came up with the mountains icon.  I'll try again.Waterfall spkrs in initial position

Posted on: 27 January 2018 by Beachcomber

Hmmm - it must be possible to do this...

[url=https://flic.kr/p/22sxvPp][img]https://farm5.staticflickr.com/4657/39025426985_cb84d055fd_z.jpg[/img][/url][url=https://flic.kr/p/22sxvPp]Waterfall spkrs forward[/url] by [url=https://www.flickr.com/photos/beachcomberco/]Steve Knight[/url], on Flickr

Posted on: 27 January 2018 by Beachcomber

OK, another go...Waterfall with spkrs in initial position

Waterfall with spkrs in initial position

Posted on: 27 January 2018 by Beachcomber

Damn it!  I've tried a number of different ways but none seem to work.  Is the problem to do with the HTTPS?

Posted on: 27 January 2018 by Beachcomber
View post on imgur.com
Posted on: 27 January 2018 by Beachcomber

Well, I seem to be able to embed a link, which is not ideal but here goes:

 

View post on imgur.com

and

View post on imgur.com

and

View post on imgur.com

plus

View post on imgur.com

with

View post on imgur.com

not forgetting

View post on imgur.com

 Or maybe this will work:

[img]https://i.imgur.com/wk066DN.jpg[/img]

Posted on: 27 January 2018 by Dave***t

Proof, if any were needed, that a preview post function would be helpful.

If I may, it is possible to bring up each image, and on Mac, right click on the image, select 'copy image address', and then paste into the box on this forum.

Which I have done for each of the images here (it seemed to me that one image was repeated, so I didn't repeat it).

Feel free to insert explanations or add in others using the above process.

In general, you need to put a DIRECT link to the image into the box on this forum.  Not any of the other kinds of 'helpful' links which various sites try to offer you.

Posted on: 27 January 2018 by Beachcomber

Oh wow - many thanks for that.  I tried following the various instructions I found lying around but couldn't succeed with any of them.. Much appreciated!

Cheers

Steve

Posted on: 27 January 2018 by Dave***t

No problem.  However, while I can read my own graphs, I don't feel qualified to pronounce on yours.

Hopefully someone who can will be along soon.

Posted on: 27 January 2018 by Adam Zielinski

I see a very uneven bass response, with some of the fundamental bass notes (low E - low A) being very quiet, almost squashed in your room. 

Do you have small speakers or very empty corners in your room?

 

Posted on: 27 January 2018 by Huge

First problem...

Look at the spectrograph.  Even before time 0 you have 50 Hz at about -55dB compared to an average signal level of about 87dB. in other words you have a 50Hz mains frequency sound at about 32dB below signal level.  This either suggests an "earth loop" in your system or something other than the output from the speakers is buzzing at 50Hz.

General observation for all plots...
There's rarely any point in plotting frequencies outside of the 20Hz to 20kHz range (use the 'Limits' control to reduce the size of the X axis to a constant set of values as this makes it easer to get a consistent interpreted 'view' of the system).

There are four observations on the SPL / phase plot:
1   The Y axis is set way too large to be useful (use the 'Limits' control to reduce the size of the Y axis)
2   This is plotted without any smoothing, try using 1/12 octave smoothing to get a more readable trace at HF
3   You are showing absolute phase.  A much more useful plot is 'minimum phase'
4   The major nulls at around 37Hz and 75Hz, suggest some significant problems with room or speaker / speaker interaction at these frequencies (if you can take separate left speaker / right speaker plots we can see if the speakers are interacting with the room as a pair or individually, and when we have better versions of this plot, we can then have a look at the minimum phase plot and the filtered impulse response to see if this is a reflection problem or likely to be fundamental to the speaker positioning).

The distortion plot:
The Y axis is set way too large to be useful (use the 'Limits' control to reduce the size of the Y axis)
On the other hand the value of 0.127 THD is excellent at a volume of about 87dB.
However the significant family of peaks around 1.7kHz indicate a major issue with the crossover.

For the waterfall plot, it's much easier to interpret and relate to other plots if you use a logarithmic X axis, using an X axis the same as the other plots will also be useful for getting better clarity in the midrange.

Lastly taking the spectrograph and Waterfall plots together indicates that your room is generally quite reflective (300-400ms in the mid-range), but is significantly under damped at about 27Hz (is one of the room dimensions 6.7m by any chance?).

 

Get these sorted first and we can then start to look at the filtered impulse response to see if there are any room reflection problems.

Posted on: 28 January 2018 by Beachcomber

Fantastic - thanks for all that.  I will re-do the charts as you suggest and see what that does. 

I'm a bit worried by the crossover issue - the speakers are Ovator S600, and I would have thought the xovers would be OK.  Do you think I should get them back to Naim to have them looked at?

Good call on the 6.7 metres - the room is (at maxima) about 6 metres by 6.4 metres and 4 metres high into the peak of the ceiling (the ceiling is pitched, from about 2 metres or so at the sides going up the 4 metres at the centre).  

I will do as you suggest and come back when done - assuming I can figure out how to post the charts properly :-)

Thanks again.

Steve

Posted on: 28 January 2018 by Adam Zielinski

Also check your speaker cables are connected correctly - no phase reversal.

Posted on: 28 January 2018 by Huge

Since you speakers are Ovator S600, there isn't a crossover point at 1.7kHz, so that distortion peak is most odd (it's in the middle of the BMR's range).
The distortion plot is typical of a 2 or 2.5 way system where the designer has pushed the sensitivity too far, where the crossover is putting too much LF energy into the HF driver and too much HF into the bass/mid driver.  However that isn't the case here.

Please could you do a sweep of each speaker individually and see if the same peak appears for both speakers when each is measured individually.  As drive units tend to be assembled by hand, faults are very often associated with individual units rather than entire batches.

Posted on: 28 January 2018 by Beachcomber

will do.

Posted on: 28 January 2018 by Japtimscarlet

This has pricked my interest to check my room ..

Is the software free / a demo ?

Posted on: 28 January 2018 by johnG

REW is free. Search on 'Room EQ Wizard (REW) Information Index' and the first two links will be all you need.

Posted on: 28 January 2018 by Innocent Bystander

You also need a suitable microphone. IIRC thevREW website dies make a suggestion for acreasonably priced one. 

Posted on: 28 January 2018 by wenger2015

As an alternative, just go on the GIK Accoustics web site, fill out the questionnaire, add some photos of your room and they will work it all out for you and let you know of the best way to treat your room, it’s a free of charge service with no obligation to buy.

Posted on: 28 January 2018 by Innocent Bystander
wenger2015 posted:

As an alternative, just go on the GIK Accoustics web site, fill out the questionnaire, add some photos of your room and they will work it all out for you and let you know of the best way to treat your room, it’s a free of charge service with no obligation to buy.

But that is quite limiting and nowhere near as good as undertaking measurements of the sound behaviour.

incidentally, GIK are happy to receive REW measurement files to help give more accurate assessments, and will then prompt for additional specific measurements if that would help further.

Posted on: 29 January 2018 by Beachcomber
Huge posted:

Since you speakers are Ovator S600, there isn't a crossover point at 1.7kHz, so that distortion peak is most odd (it's in the middle of the BMR's range).
The distortion plot is typical of a 2 or 2.5 way system where the designer has pushed the sensitivity too far, where the crossover is putting too much LF energy into the HF driver and too much HF into the bass/mid driver.  However that isn't the case here.

 

Still to do that sweep - will try to get time this evening.  Just a thought - could that problem (or any of the others) be because the NAC52 is over 20 years old and not serviced?

Posted on: 29 January 2018 by Huge

As that particular issue is so frequency specific, it's not particularly likely be caused by the amplifiers (but that is still a possibility).

As the even order harmonics dominate over the odd order harmonics, the only electronics that would tend to cause that type of distortion would be the bridged poweramps of the 500.  There is still a small chance that it is from the amplifiers, but if it were, I'd still expect to see a wider frequency distribution.


Note that although it's an issue, it may not actually be a problem!

The distortion is dominated by the the less offensive even order harmonics and only present over quite a narrow range of frequencies that are associated with musical and vocal harmonics rather than fundamental tones.  Because of this, the brain can, to a degree, separate out the distortion associated only with the harmonics away from the natural harmonics of the voice or instrument that are associated with the fundamental tones.