REW room measurement analysis and fix help

[haygeebaby]

Yeah - its a difficult one. After using an axial standing wave calculator I can understand it better.

The area I can sit for best effect on the width axis is about 28cms left/right of the center.

This gap is about 40cm at most. Move your head a little left or right and you don't hear some low range frequency.

Move my head up a little higher - and I also don't hear some bass level frequencies.

I might do the calcualtions and see how things work if everything was set on the short side of the room instead of the long side as it currently is.

But this might be too much of a PITA to implement.

 

This is my current plot in the new position with speakers and listening chair moved 28cms to the right. Taken with mic on the mic stand and no chair. Also with R only and L only speakers as recommended above. Average plot of all the tests as well. Tests where of center pos, left and right 10cm, and left and right 15cm. At listening height of 80cm. With L and R mic combinations.

 

Question - why never use both L&R speakers at the same time to measure?

 

 

 

[almikel]

41 minutes ago, Al.M said:

According to the experts link I attached earlier it’s not possible to resolve the big nulls and modes, they are too severe and yours is a typical one. Changing seat position can resolve it mostly, massive amounts of treatment and EQ can only alleviate but not eliminate it if you still choose to sit where the problem is

I don't disagree with anything said by the guy in the link you attached

 

What he didn't discuss was the option of multiple bass sources to smooth the room's frequency response.

 

46 minutes ago, Al.M said:

And this is after he has had a go at it, not amateurs and semis guessing at it.

sure I'm an amateur, but as the video states, there's no voodoo - there's plenty of science behind the recommendations of Toole, Geddes, Olive etc for using multiple subs to smooth the room's bass response.

 

1 hour ago, Al.M said:

According to the experts link I attached earlier it’s not possible to resolve the big nulls and modes, they are too severe and yours is a typical one.

It is possible, but achieving a reasonable room "time domain" (ringing) response is the hard part in the low end - especially if the room has rigid room boundaries.

 

IME you can resolve big nulls and modes in the frequency domain with multiple bass sources - but great "in room" sound requires consistent reverb times across the frequency bands - which is harder to achieve.

 

I've managed to achieve "reasonable" RT times in my own room and others I've helped design and build, with excellent low bass "in room" response, and "reasonably" consistent RT times, typically rising below 100Hz or so, where absorption gets too big.

 

In all the rooms I've dealt with I've never had to manage low bass in a rigid room - that would be a challenge!

 

For the rooms I have had input to (all of light construction), some broad EQ cut works wonders below 150Hz or so where treatment gets too large.

If applied well, EQ will flatten the frequency response and reduce ringing in the time domain. 

 

cheers

Mike

[almikel]

54 minutes ago, haygeebaby said:

Question - why never use both L&R speakers at the same time to measure?

because it confuses things.

Your ear/brain doesn't "hear" what the mic "hears" - your ear/brain does all sorts of averaging/compensation that you don't realise is going on...it's much better to have the mic "hear" a single speaker, so you know what the speaker is doing in the room...

...keep in mind the room has much more impact than the speaker...measuring a single speaker rather than both at the same time removes the interaction between the speakers and allows you to focus on the interaction between the speaker and the room.

 

Mike

 

 

[andyr]

1 hour ago, almikel said:

because it confuses things.

Your ear/brain doesn't "hear" what the mic "hears" - your ear/brain does all sorts of averaging/compensation that you don't realise is going on...it's much better to have the mic "hear" a single speaker, so you know what the speaker is doing in the room...

...keep in mind the room has much more impact than the speaker...measuring a single speaker rather than both at the same time removes the interaction between the speakers and allows you to focus on the interaction between the speaker and the room.

 

Mike

 

But I want to measure what my ears are hearing in the listening position - with 2 spkrs going. 

 

Then I can tell what needs to be adjusted (via PEQ etc) to 'fix' the listening position.

 

Andy

 

 

Andy

[almikel]

16 minutes ago, andyr said:

But I want to measure what my ears are hearing in the listening position - with 2 spkrs going. 

 

Then I can tell what needs to be adjusted (via PEQ etc) to 'fix' the listening position.

no problem - just jump to a different universe where the microphone can replicate the "ear/brain" transfer function.

 

In this universe you can reduce complexity and make reasonable adjustments in EQ that will be "perceived" as better by the ear by measuring room response 1 side at a time...

 

...are you familiar with the analysis concepts of "superposition" or "degrees of freedom" ?

 

Both are relevant:

• Superposition is used lots in electronic engineering to analyse a circuit that has multiple voltage/current sources - you short circuit/ignore all but 1 voltage or current source, and analyse based on the single source. Proceed to the next voltage/current source and analyse. Repeat until all voltage/current sources have been analysed. Superimpose all results - that's the outcome - vastly easier to analyse individually and add the results than analysing all sources simultaneously 
• degrees of freedom relate to the number of variables in a process - it's vastly harder to determine what's going on when >1 things are changing at once

1 hour ago, andyr said:

Then I can tell what needs to be adjusted (via PEQ etc) to 'fix' the listening position.

until the microphone/software replicates what the ear/brain interface "hears" - you won't

 

Mike

[Grant Slack]

On 30/12/2019 at 1:47 AM, haygeebaby said:

....I've managed to avoid the 40hz null. Can't sit in the center width wise. So I moved the chair 50cm to the right....

So far I've moved my listening spot forward slightly, to the right, and increased my ear height. All 3 help me avoid the standing waves.

But this means all my symmetrically placed acoustic panels are off compared to the new optimized seating position.

....We've opened up a can of worms!

 

Hello haygeebaby,

 

I'm afraid you are throwing out the baby with the bathwater here. It is no secret that a listening position mid-point between walls will be in a null at a certain bass frequency. However, for accurate stereo imaging you do need to be halfway between the sidewalls. Left-right symmetry is vital for helping you localize sound sources in the sound stage.

 

The much preferred solution, as already suggested, is to play with subwoofer locations and/or additional subwoofers. Even move the listening position -- while staying central.

 

The tail (bass) should not wag the dog (rest of the music).

 

Regards

Grant

Edited January 3, 2020 by Grant Slack

[andyr]

12 hours ago, almikel said:

 

...are you familiar with the analysis concepts of "superposition" or "degrees of freedom" ?

 

Both are relevant:

• Superposition is used lots in electronic engineering to analyse a circuit that has multiple voltage/current sources - you short circuit/ignore all but 1 voltage or current source, and analyse based on the single source. Proceed to the next voltage/current source and analyse. Repeat until all voltage/current sources have been analysed. Superimpose all results - that's the outcome - vastly easier to analyse individually and add the results than analysing all sources simultaneously 
• degrees of freedom relate to the number of variables in a process - it's vastly harder to determine what's going on when >1 things are changing at once

Mike

 

 

A remarkably condescending comment, Mike.  As someone with:

a.  a 1st class Honours degree in Maths, plus

b.  2x postgraduate degrees in Applied Maths

 

... yes, I'm familiar with the concepts of 'Superposition' and 'degrees of freedom'! 

 

12 hours ago, almikel said:

 

no problem - just jump to a different universe where the microphone can replicate the "ear/brain" transfer function.

 

In this universe you can reduce complexity and make reasonable adjustments in EQ that will be "perceived" as better by the ear by measuring room response 1 side at a time...

 

 

But, in this universe, you are saying a microphone cannot replicate the "ear/brain" transfer function.

 

Ergo, it makes no difference whether you measure in mono ... or stereo - it's not doing what your ear/brain is doing? 

 

12 hours ago, almikel said:

 

In this universe you can reduce complexity and make reasonable adjustments in EQ that will be "perceived" as better by the ear by measuring room response 1 side at a time...

 

 

Even though the microphone cannot replicate the "ear/brain" transfer function, I can see that the following procedure can come up with a good result:

1.  put the mic at the LP - pointing straight ahead (not angled towards a speaker)
2.  measure - with only one channel going
3.  apply PEQ to that channel to get the optimum plot from REW
4.  measure - with only the other channel going
5.  apply PEQ to that channel.

So that's what I'll try, next time I get the mic & computer out. 

 

Andy

 

Edited January 3, 2020 by andyr

[Grant Slack]

1 hour ago, andyr said:

Even though the microphone cannot replicate the "ear/brain" transfer function, I can see that the following procedure can come up with a good result:

1.  put the mic at the LP - pointing straight ahead (not angled towards a speaker)
2.  measure - with only one channel going
3.  apply PEQ to that channel to get the optimum plot from REW
4.  measure - with only the other channel going
5.  apply PEQ to that channel.

So that's what I'll try, next time I get the mic & computer out. 

 

 

Hello Andy,

 

the general convention in Step 1 is to point the omni-directional measurement microphone straight up at the ceiling, or less than 10-15 degrees forward of straight up, for the type of test you describe. This minimises measurement errors caused by the imperfect polar response of the "ideally omni" microphone. Pointing it forwards will actually maximise that error, which, although I haven't checked, is reported to be quite obvious.

 

For a gated measurement of the speaker's own response, then point the microphone at the speaker.

 

Regards

Grant

Edited January 3, 2020 by Grant Slack

[andyr]

1 hour ago, Grant Slack said:

 

Hello Andy,

 

the general convention in Step 1 is to point the omni-directional measurement microphone straight up at the ceiling, or less than 10-15 degrees forward of straight up, for the type of test you describe.

 

Regards

Grant

 

Aah, yes - thanks Grant.  Up it is! 

 

Andy

 

[haygeebaby]

Yeah that’s a good comment Grant.

I’d like to achieve the best sound possible in this room. And the middle is ideally where I’d like to sit as well.

Might be time to contact an expert and learn some more in the process.

 

[davewantsmoore]

 

21 hours ago, haygeebaby said:

Question - why never use both L&R speakers at the same time to measure?

In practise... Your L and R speaker will have a different response.   What you measure when you play the together is the complex sum, which will have peaks and dips in it due to the not-identical nature of your speakers.

 

For example, there is a distinct dip at 5.5khz in your L+R response...   That could be the result of your speaker not being precisely time aligned with the microphone.   They would only need to be out by a few cm.

 

In short.... I don't know if the peaks and dips in the charts are the result of the L speaker interacting with the R speaker ... the L speaker interacting with the room differently to the R speaker..... or something else.    Just one speaker playing, at a time, is much much clearer.

 

 

The only time you ever want to measure L AND R is when you want to investigate how perfectly L and R are playing together (and you want to compare the L AND R result with L only , and R only....   but that's not what you're investigating here.

 

 

 

4 hours ago, andyr said:

A remarkably condescending comment, Mike.  As someone with....

 

I wonder if Mike was aware of that...

 

 

Edited January 3, 2020 by davewantsmoore

[davewantsmoore]

19 hours ago, andyr said:

But I want to measure what my ears are hearing in the listening position - with 2 spkrs going.

 

With 2 speakers going you will see peaks and dips in the response based on the degree to which your speakers:

• Are not equally time aligned with you
• Are not equally time aligned with the room surfaces
• Don't have exactly an equal frequency response

 

 

 

Errors caused by these differences are not errors which you should correct using EQ.   It is the source of the error which should be corrected.

 

[davewantsmoore]

3 hours ago, Grant Slack said:

the general convention in Step 1 is to point the omni-directional measurement microphone straight up at the ceiling, or less than 10-15 degrees forward of straight up, for the type of test you describe. This minimises measurement errors caused by the imperfect polar response of the "ideally omni" microphone. Pointing it forwards will actually maximise that error, which, although I haven't checked, is reported to be quite obvious.

 

This is true (although if you have a mic cal for different orientations, this can resolve it)  .....  but if doing that type of measurement, then you don't want to just sample the listening position and correct with EQ....   So you missed the important bit, which was telling Andy his procedure was flawed.

 

Some amount of averaging and processing of multiple measurements, can make the process work (like typical "room correction" packages do).... but the before/after charts make them look like they do a much better job than they really do.

 

 

A mentioned earlier.   The "source of the error" is important.... so speakers which have a more uniform directivity, and more perfect time alignment to the listener and room, are better candidates.

 

3 hours ago, Grant Slack said:

For a gated measurement of the speaker's own response, then point the microphone at the speaker.

Yes, it's this sort of measurement where you see what is a candidate for correcting with EQ.

 

IME, the difference between pointing the mic directly at the speaker, and straight ahead (ie. between the speakers) is extremely small (mic dependant).

[Grant Slack]

1 hour ago, haygeebaby said:

Yeah that’s a good comment Grant.

I’d like to achieve the best sound possible in this room. And the middle is ideally where I’d like to sit as well.

Might be time to contact an expert and learn some more in the process.

 

Well, I wish you well with that, but truth is, acoustic consultant experts are often not as expert as they (and their clients) think. No telling what half-truths you will be led to act on. Take Herr Lohan in the video linked earlier, I have a few quotes:

“I am a professional audio rooms (mixing rooms, recording rooms) guy, but the principles I am about to describe applies the same to home listening” — not according to Toole

“wall reflections create this big mush of comb filters that can really destroy the sound” — not according to Toole for home listening

“the first thing is to optimise the room dimensions if possible” — not according to Toole

That was in the first 10 minutes. I didn't go on; no telling what combination of truths and not-really-truths are coming from him, just like the internet.

 

Speaking of which, you have been advised in this thread to: attenuate your tweeters based on one L+R measurement (too hasty); align your system diagonally across the room, somewhat (umm...); add a sub at the back of the room (good idea - will need adjustable delay); turn everything 90 degrees and listen along the short room dimension with your seat on the back wall with an optional absorber behind your head (could hardly be worse advice); add some absorbent furniture (decent idea but a bit premature and not about the bass); move seat forward and/or speakers back (sure, why not give it a go); move your subs around (yes! - but only with digital delay lines); and add subs (yes! - same proviso).

 

I know: it sounds hopeless if you can't even trust the hired guns. Some of them are near-genius, and some are near-incompetent -- what else did you expect? Same goes for GPs etc; that's how life is. That's why there are such things as second and third expert opinions, eg in medical. My optician was a real guru -- wrote the textbook used to teach opticians with -- and used to say to me, "I wouldn't trust most opticians to prescribe glasses for my cat!" Well now he has retired, and I am at the mercy of the mob! I'm sure audio consultants are no different.

 

But hope springs: what you do have, is your intelligence, your microphone, and your wallet. The first of those is probably telling you that: you need to keep the layout symmetrical and in-line; room treatments won't make that null go away without at least 4 ft thick loose fibre on the full length of your side wall to stop a null caused by sitting mid-width; and the research is clear that mobile and/or additional subs are the technical solution to this, along with intelligently-applied bass EQ. The rest you can do with your wallet and your microphone.

 

cheers

Grant

[almikel]

21 hours ago, Grant Slack said:

the research is clear that mobile and/or additional subs are the technical solution to this, along with intelligently-applied bass EQ.

agreed - by "mobile" I assume you mean, "placed where the sub/s have the smoothest room bass response at the listening position".

 

21 hours ago, Grant Slack said:

“I am a professional audio rooms (mixing rooms, recording rooms) guy, but the principles I am about to describe applies the same to home listening” — not according to Toole

I don't think Toole would dispute that recording rooms are similar to listening rooms and the same principles apply for achieving good "in room" sound.

 

21 hours ago, Grant Slack said:

“wall reflections create this big mush of comb filters that can really destroy the sound” — not according to Toole for home listening

Toole specifically discusses leaving 1st sidewall reflections alone if the speakers have a smooth off axis response - front/rear wall reflections do contribute to comb filtering/SBIR.

 

21 hours ago, Grant Slack said:

turn everything 90 degrees and listen along the short room dimension with your seat on the back wall with an optional absorber behind your head (could hardly be worse advice)

agreed this is bad idea

 

21 hours ago, Grant Slack said:

“the first thing is to optimise the room dimensions if possible” — not according to Toole

Toole discusses there are no "golden ratios", and that all rooms have modes, ...I'll need to dig deeper into my Toole texts, but I expect he would still recommend avoiding even multiples of room dimensions to minimise the stacking up of room modes.

 

21 hours ago, Grant Slack said:

move your subs around (yes! - but only with digital delay lines); and add subs (yes! - same proviso).

I agree with this with one proviso...depending on sub/s position relative to the mains wrt to the listening position, and the crossover filters used, for optimum time alignment of the subs with the mains, you may have to delay the mains more than the subs...

This is often a bridge too far with analog setups, where the only delay available is with the sub/s.

If absolute time alignment (between sub/mains) isn't critical (and IMHO it isn't), delaying the sub/s a cycle or so to achieve a smooth frequency response with the mains is a good outcome.

 

I'm happy to compromise true time alignment (between subs/mains) for a smooth "in room" frequency response.

 

22 hours ago, Grant Slack said:

acoustic consultant experts are often not as expert as they (and their clients) think.

I stand by my recommendation of @Paul Spencer - many years real world experience in room analysis/treatment and speaker design.

 

cheers

Mike  

 

 

[Grant Slack]

Thanks Mike, good comments all round. I hadn't thought about when the subs are further away than the mains!

45 minutes ago, almikel said:

I don't think Toole would dispute that recording rooms are similar to listening rooms and the same principles apply for achieving good "in room" sound.

Two major differences. Recording rooms vary in size up to 'non-small' performance spaces for orchestral recordings. The large ones behave quite unlike a home and need very different treatments. The small ones, along with mixing rooms, are made very anechoic in a way that no home should be -- according to Toole, AFAICT. I am pretty sure I have seen Toole, either in video or text, say that acoustic consultants as a group have a knowledge and a toolkit that is not right for homes, and are not well versed in the needs of homes. Which is implying that the  principles vary. However, if someone asks for that statement, I don't think I know where to look. My apologies.

 

regards

Grant

 

[almikel]

On 04/01/2020 at 4:29 PM, almikel said:

I don't think Toole would dispute that recording rooms are similar to listening rooms and the same principles apply for achieving good "in room" sound.

 

On 04/01/2020 at 5:28 PM, Grant Slack said:

Two major differences. Recording rooms vary in size up to 'non-small' performance spaces for orchestral recordings. The large ones behave quite unlike a home and need very different treatments. The small ones, along with mixing rooms, are made very anechoic in a way that no home should be

Hi Grant,

I had a bit of a brain fade and was thinking "control room"/mixing room instead of the space where the musicians make the music, which is of course the recording room, and can vary greatly in size from large auditoriums to bedrooms (in which case the control/mixing room is usually the same space as the recording room   )

 

I went scanning through my copy of Toole's Sound Reproduction - Loudspeakers and Rooms, mostly to confirm the assertions I made above on my interpretations of Toole were accurate.

I keep reading sections and get dragged down rabbit holes not relevant to this thread.

 

In this specific text, Toole primarily focuses on the size of the room and the differences in the sound field from large (auditoriums), medium (offices and industrial spaces), and small (domestic listening rooms and control rooms) rather than the purpose of the room.

 

I've been in a variety of control/mixing rooms, both professional and amateur. None of the good mixing rooms were overly anechoic sounding - and the mixing rooms that were overly anechoic/dead never produced music that sounded "right" in normal spaces. 

 

Some key nuggets from the text, all from Chapter 13 "Making (Bass) Waves below the transition frequency":

• Everything matters: room dimensions, placement of listeners, loudspeakers, wall construction, where you put a door, and on and on.
• There are no generalized “cookbook” solutions, no magic-bullet room dimensions
• Without your own acoustical measurements, you are “flying blind.”
• Without high-resolution measurements, you are myopic
• With good acoustical measurements and some mathematical predictive capability, you are in a strong position to identify and explain major problems.
• There are indications that some combination of low-frequency acoustical treatment, multiple subwoofers, and equalization will be helpful
• The idea of optimizing room dimensions has not been abandoned, but future investigations must take into account where the loudspeakers and listeners are located.

It's been ages since I reviewed one of Toole's texts - a worthwhile read...my takeouts:

• getting the "in room" bass right is important - IMHO it should be the 1st priority for achieving good "in room" sound"
• to get the "in room" bass right you need accurate and high resolution measurements...and much more difficult...the ability to interpret the measurements and make changes that improve the measurements (then rinse/repeat etc)
• a combination of treatment focused on low frequencies, multiple subs and EQ in conjunction with measurements is required to achieve great "in room" bass

 

On 04/01/2020 at 5:28 PM, Grant Slack said:

I hadn't thought about when the subs are further away than the mains!

If accurate time alignment between subs and mains are a goal (and IMHO a smooth frequency response between mains and subs is more important than time alignment), then more than distance is involved - the group delays of the high pass/low pass crossover between the mains and the sub/s needs to be considered.

 

If you have a symmetric high pass/low pass crossover between the woofer in the mains, and the sub/s, then they will have the same group delay profile - so distance to the listening position is the primary consideration for delay, and if the subs are further from the listening position than the mains then the mains may need to be delayed.

 

In a more typical mains + sub environment, with say a sealed woofer in the mains without a high pass filter the woofer in the mains speaker will have a 2nd order high pass rolloff - for simplicity let's say it rolls off at 80Hz.

Lots of commercial subs have a 24dB low pass filter - let's say we set that at 80Hz.

 

All filters have delay, and the steeper the filter the more delay (also the lower freq the filter the more delay)

 

In this scenario the low pass filter of the sub would have more delay (as it's steeper) than the high pass filter of the woofer in the mains (the driver rolloff in the speaker enclosure), so even if the sub was co-sited with the mains it would "lag" the mains requiring delay of the mains to achieve "time alignment".

 

I must admit, achieving "time alignment" between mains and subs is a rabbit hole as to what's the best approach but with my fully digital setup I've gone with aligning the impulse responses at the listening position between my single sub and the mains woofer closest to the sub, and adjusted room EQ from there to get the smoothest bass response across the listening couch...I'll add another sub at some stage...I've got all the bits, just need to make sawdust.

 

Interestingly my single sub is closer to the LP than the mains, but using the impulse response alignment method, the mains require more delay than the sub...pretty sure its a symmetric 24dB LR electronic crossover, but who knows what the acoustic crossovers are??

 

Ultimately it's about a reasonably flat FR response at the listening position (dial your own room curve) and good room decay - a good sounding room requires consistent reverb times across the spectrum.

 

cheers

Mike

[haygeebaby]

Just an update for you guys.

With the last seating position change we avoided some of the bass nulls in the room.

Since them I have been planning sub woofer placement to further improve the bass frequency in the room.

Below is the new room frequency response with a small sub woofer located in the rear left corner of the room.

The results have been very good. I am really enjoying the added lower frequencies.

 

The last graph and then the most recent with sub woofer in purple.

 

[almikel]

 

On 24/01/2020 at 2:50 PM, haygeebaby said:

The last graph and then the most recent with sub woofer in purple.

 

Is that measurement with the listening position centered again and just adding a single small sub?

 

If so that's an awesome result

 

On 24/01/2020 at 2:50 PM, haygeebaby said:

I am really enjoying the added lower frequencies.

 

You still have a broad dip 500Hz - 3kHz or so - if you have EQ capability some broad gentle boost (low Q, low gain) may help in this region.

 

EQ has a deservedly bad reputation, as poorly applied it can sound truly dreadful...and EQ boost is generally not recommended for good reasons...

 

...but deployed well, EQ works wonders IMHO...

 

Keep EQ broad and low gain (cut and boost) and it works fine.

 

Careful EQ cut of peaks can be narrower (like the ~30Hz peak in your measurement) - just check the outcome at multiple listening positions...the narrower the EQ applied, the more position dependent it becomes.

 

Never apply EQ boost to sharp dips.

 

There's plenty of good science that says the ear is not at all sensitive to sharp peaks/dips in the frequency response (high Q), but is very sensitive to small gentle changes in frequency response (low Q)...

...it's likely you would find the gentle dip between 500Hz - 3kHz much more audible than the large peak at 30Hz.

 

The graph makes you want to fix the peak at 30Hz first...it's quite steep (ie high Q), so although it looks bad, it may not be audible...I'd want to see the room's time domain response before deciding if it needs treating...if it's ringing a long time then it needs to be dealt with (a REW waterfall will show this).

 

At lower frequencies in typical rooms (say below 250Hz), ie within the modal region of the room, cutting the modal peaks with EQ will reduce the peaks and the time domain ringing...not intuitive, but true based on good science and demonstrable with measurements...

 

...and if your room bass sounds fine (ie not ringing/overhanging) - ignore that peak!!!

 

cheers

Mike

 

 

[JPete9]

Thats not a bad response. I'd try and stop all measuring for a little while and listen to your  music...even just for a short time and then come back to it.  1. when you're deep into something, sometimes stepping away can help with perspective/clarity. 2. it might not measure to absolute benchmark perfection...but it probably sounds pretty great.