A good example of how to use acoustic measurements to place speakers

[Nyal Mellor]

Cross posted from this thread:

Here's the graph showing the before and after of dallasjustice's speaker placement. The frequency resolution is 1/12th octave.

The placement shown was a result of the following:
1) analyzing the 'before' graph to identify root causes of the issues. In this case there is a room mode related peak at ~25Hz another at ~47Hz and a speaker boundary interference related null at ~37Hz. There is no modal support for frequencies between 25Hz and 47Hz which is another cause of the frequency response null or 'droop'. The room mode frequencies and distribution were calculated using a proprietary room mode calculator. In more complex non-rectangular rooms a boundary element model is used to predict modal frequencies and distribution.
2) determining suitable placement to minimize coupling of speakers to the room modes and minimizing speaker boundary interference. In this case we were lucky in that the null for the 47Hz mode was around 5ft from the back wall, the placement of the speakers in this null would also result in coincidence of the speaker boundary interference null (caused by the front wall behind the speakers) at this frequency, therefore further reducing its effects.

Blue is before, green is after.

All of this is before any room treatment. Acoustic treatment will help smooth the modal issues about ~60HZ. EQ will be used to further flatten things out below this. The real objective of this exercise was to set things up so that frequency response was as flat as possible below 60Hz so that acoustic treatment could deal with the remaining above 60Hz issues. It's obviously very difficult to EQ a null such as the 37Hz one, so that's why speaker placement is important, as this exercise shows.

 

[dallasjustice]

funny how fixing the bass can make such a big difference in the rest of the mid-range. Everything sounds awesome Nyal!

 

[Randall Smith]

Being linear is extremely important to revealing everything that is in a recording. Even someone who places little to no importance on measurements has to believe this to be true. Speaker placement is so important. Getting a smooth response isn't always the primary goal though. The best placement is a compromise. Sometimes the location for the best imaging and soundstage isn't the best location for being the most linear. Measuring and listening is required. Understanding measurements and an audiophiles willingness to embrace their importance is a true sign, to me, of someone who gets it. They truly want to achieve the highest level of music reproduction their system has to offer.

 

[amirm]

Thanks for sharing that Nyal. It is great to see such large effects from just placement optimization.

 

[NorthStar]

1. Speaker's positioning.
2. Room threatments.
3. Intelligent EQ (mainly in the bass region).

And you end up with better clarity, articulation all across the entire audio spectrum.

Thanks Nyal.

 

[kevinh]

Impressive results, make the job of passive treatments or eq much easier.

Nyal was this a situation with multiple (3 or more) subs?

Hoave you worked with a multiple sub setup, if so how effective is it in reducing the room modes?

 

[Nyal Mellor]

Impressive results, make the job of passive treatments or eq much easier.

Nyal was this a situation with multiple (3 or more) subs?

Hoave you worked with a multiple sub setup, if so how effective is it in reducing the room modes?

No this was a straight two channel system where the client did not want additional subs.

I have worked a lot with multiple sub setups, it is VERY effective at reducing room modes, especially if you place subs to cancel out the major problematic axial modes. Mode cancellation is kind of magical when you experience it for the first time. You put sound sources in poles of opposite acoustic polarity for the room mode and then that room mode ceases to exist. Sometimes it is not possible to place the subs where they need to go due to practical constraints, in which case the next best thing is putting sound sources in nulls for those modes, though in a room with really stiff walls that won't have that much of an effect.

A before measurement. There were actually FOUR subs in this measurement, but they were placed randomly and as you can see they weren't really being effective!

An after measurement. We took two of the subs out and placed the remaining two in a front / back arrangement to cancel out the 43Hz mode.

Unfortunately the measurements are shown with different scales, but you can see what I am talking about if you focus in on the 43Hz room mode in the before and compare that to the after.

 

[theguesswho]

Hi Nyal, In the first graph, what is the target level? IE: 70db or 75 db etc?
Thanks Wendell

 

[kevinh]

Nyal, that made a huge difference.

Question why go to 2 SW and would the results been better if a third SW was added as per Geddes?
Were the main speakers run full range and did they go down to 50hz?

Thanks for the graphs this is fascinating to me.

 

[Nyal Mellor]

Wendell - not sure what you mean by 'target level'?

Kevin - the above graphs are from a home theater system, so 80Hz crossover. The results with four subs are as you can see in the first graph i.e. four subs, randomly placed with no attention paid to how they interacted with the room modes produced worst results compared to two subs placed specifically in a mode canceling arrangement. The key for mode cancellation to be effective is that the subs have to be 'playing together' over the mode you are trying to cancel. Nearly all modern subs have EQ to bring up the bottom end, and this causes phase shift. So if you use different subs, especially with frequencies below 50Hz where the EQ starts to kick in, then they may not do the magical mode cancellation thing!

 

[theguesswho]

Hi Nyal, target level is the db level that you are shooting for. If you are going for a flat response, which I assume you are, is that at a level of 65db or 70db or 75db? On your graph there are horizontal lines which line are you attempting to match?
Thanks
Wendell

 

[Nyal Mellor]

Hi Nyal, target level is the db level that you are shooting for. If you are going for a flat response, which I assume you are, is that at a level of 65db or 70db or 75db? On your graph there are horizontal lines which line are you attempting to match?
Thanks
Wendell

Hi Wendell...oh you mean the graph I posted in the first post? If so then no target level, really I was just looking at the octave to octave trends and show how one could get rid of a 20dB dip to peak difference in a very small frequency range just by moving things a little.

 

[theguesswho]

Hi Nyal, OH-OK thanks for the answer.
Wendell