floor reflection what to do ?

[DanDan]

Dange, speaking somewhat ignorantly. I doubt that it is possible to explain MP without the Math. I have made the point previously that there are those amongst us for whom Maths is a language. The words and concepts are part of a vocabulary. Sometimes they forget the rest of us. Take a look at this.

This is a screen shot of FM2. With all due respect, Chris clearly thinks MP and all the variants are common parlance. The manual says little more on the topic, apart from MP seems to be just 'better'. There is a command to create an MP version. I would have made that the invisible default and removed all top level references to it.
Better is better, who needs to know why? It is certainly more useful to have ETC's start at zero.
However if anyone shine a light on MP for the common man, Lupo's the one. In fact I am fairly sure he already made a strong attempt earlier in this thread or in another one which I was also in.

However if one redirects focus to the implications of MP, what does it 'do' in a real life situation?
May I suggest that the Front Wall reflection and it's big dipper, are a Minimum Phase system. Frequency does not change with distance/time.
Similarly the boundary LF reinforcement of a Soffit system is MP. The reinforcement is stable and consistent.
Being MP, both of these systems can be adjusted by another MP system. An MP (ideally) equaliser. This can be the simplest shelf Eq for soffit correction.
Or the very exotic Meyer Parametric.
Conversely the Floor Reflection dip frequency does change with listener distance. Pretty phasey I would say....LOL
DD

[DanDan]

Lupo, I think the particular type of reflections you are focussing on will be problematic. You refer to reflections returning to the speaker. This may be true of the side wall ones and the floor and ceiling ones. However, yet again the Front Wall one is unique. The reflection travelling towards the speaker is also travelling towards the listener. I don't think this coincidence, and it's absence in the other axes, can be ignored.
Most of us use the term side wall reflection during RFZ activities. Your side wall reflection would be the 'other' SBIR in my dictionary.
May I hypothesise? My side wall reflection is very simple, the mirror works fine. Your one needs no mirror! However, due to the position of your anomaly, i.e. a virtual speaker inside the wall, I strongly suspect that modal issues are clouding your measurements and attempts to deal with it.
Let's face it, if you had enough side space you could treat this as a width mode issue.

I don't really understand what you are saying about the Genelec figure. I presume they have measured up to 20dB dips in reality. They claim 6-20dB are typical.
Genelec Oy - Wall behind the Loudspeaker Cancellation

DD

[Dange]

Originally Posted by DanDan

Dange, speaking somewhat ignorantly. I doubt that it is possible to explain MP without the Math.

I'm all for the Math..ss, just thought a minimum maths explanation would be better suited to this forum. Having said that starting off with a layman's explanation and moving on to more complex things is the way I like to get into new concepts anyway

[DanDan]

Dange, I am 100% with you. However our wish won't come true. MP I believe is an independent concept. It is not limited to the topics here. A bit like Logs in that respect. I would regard it as 'Pure' Maths. Try a Google and you will see what I mean. Lupo and SAC have gone at this before, pretty valiantly but in vain IMHO. The explanations will inevitably contain words like 'stable' 'causal' and so on. All of these need new definitions within the Mathematical context, and outside of normal english. And that's just to get started.
The other way to consider MP is by way of how it affects us in our little interest sphere. I have given some examples. Lupo was more succinct, with real minimum math, just the equal sign!

Minimum phase = minimum frequency dependent delay.

This totally works for me though.

DD

[SAC]

A precise definition of minimum phase is a detailed mathematical concept involving positive real transfer functions, i.e., transfer functions with all zeros restricted to the left half of the s-plane.


For Lupo - in response to another question about multiplying by "i"



One significance of a minimum phase system is that it allows us to use the Hilbert transform to use the Hilbert transform to derive the imaginary phase from the real magnitude (and visa versa). As for a minimum phase system, the phase is not independent of the magnitude.


It is the Hilbert transform that enables for the definition of the Heyser analytic from a real-valued time response, where the real and imaginary planes are related directly by the Hilbert transform. And as such the envelope (magnitude) of the time signal can be derived (ETC), as well as decay time calculations, arrival times and propagation delay...


And in regards to equalization, the implication for minimum phase systems is that a change in the magnitude of the signal allows for the one to one corresponding change in the phase of the signal; whereas in a non-minimum phase system a change in the magnitude does NOT



If you know the amplitude response of any minimum phase system, then the phase response can be calculated from the amplitude response. Conversely, if you know the phase response of a minimum phase system, then one can derive the amplitude response. The phase and amplitude responses are two different perspectives of the same event or analytic, and they display a direct correspondence as long as the system at hand is minimum phase.


But as many (most) systems we deal with are not minimum phase, it becomes increasingly important that we be able to evaluate the relationship between the two perspectives. And one important implication of this is that equalization or amplitude cannot be used to correct for room anomalies as the variation in magnitude does not result in a corresponding modification of phase. Hence why equalization cannot be used to correct room and system anomalies resulting from the superposition of non-minimum phase sources.




And regarding the discussion of SBIR and reflections, take a look at the Genelec series of articles and tutorials. Genelec is one of a small handful of really refreshing companies who actually "get it" and disseminate accurate information as opposed to slick marketing hype designed to sell product as is all to common now days.



Placing Loudspeakers correctly in Control Rooms


Wall behind the Loudspeaker Cancellation


and perhaps the best way to solve the issue (and one of the very best descriptions of how to PROPERLY do it!):


Flush Mounting to Wall


And a feature that can help address SBIR issues, that quite frankly few manufacturers do properly if at all - and a feature that if not present in a properly implemented manner (or if the design is not specifically tailored for a particular stated environmental application), should eliminate the monitor as a potential choice!

Room Response Controls

[DanDan]

Nice one SAC. Apart from Hilbert, fully understood here. Also fully agree on the Genelec comments.
And most especially, +1 to the last comment below. I bought my ADAM S3A's largely because I could use the onboard controls to emulate a response I knew well and found very successful. My ADAM's are tweaked to the old B and K curve, very similar to my older monitors, Celestion Dittion 66's. Age showing here now so I will stop. ;-)

And a feature that can help address SBIR issues, that quite frankly few manufacturers do properly if at all - and a feature that if not present in a properly implemented manner (or if the design is not specifically tailored for a particular stated environmental application), should eliminate the monitor as a potential choice!

Eggxactly

DD

[barefoot]

Originally Posted by Mixary

here is a mathematical description for a 110 Hz Floor Reflection.

Originally Posted by DanDan

A mirror, measuring tape, and calculator, are of course useful, but Thomas Barefoot has done us a great favour by writing a 2D Wall Bounce calculator. Good fun. John Sayers' Recording Studio Design Forum • View topic - Wall Bounce Calculator 2D

Marc,

The Wall Bounce Calculator can be used to simulate your floor bounce. Take a look at the attachment. I've made Mf=10000 cm. It's very large so front wall reflections become negligible. You gave a listening distance of 120 cm. So we set Lf=10120. Set the Listening Angle (θ) to zero. And set the Side Listening Distance (Ls)=124 cm. Now the side wall is the same distance as the speaker listening height in your situation. So we just pretend that the side wall is the floor. Make sense?

Looking at the response graph you can see that there is indeed a dip near 110 Hz. However, the dip is only about -2dB. It's not a null. Why? Well, you have to remember that the reflection path length is more than twice as long as the direct listening distance. Amplitude drops off as one over the distance squared. So the direct sound is much louder than the reflected sound. The 180 degree out of phase reflection will only cancel out a small portion of the direct wave.

I don't think the floor reflection is causing the majority of your problem.

Hope this helps!
Thomas

[SAC]

Just a few comments...The only problem with the model is that sound emanating from the side of an enclosure is NOT the same magnitude (not to mention the propagation and group delay and further modification of the signal due to the enclosure material bulk modulus) as the direct signal that then reflects off a near field reflective boundary and then recombines (superposes) with the direct signal.

The superposition of direct signals with early first order reflections of the direct signal can indeed result in infinitely deep nulls. Not only that, but the physical phenomenon that is occurring is spatial polar lobing...regions in space where a particular frequency is simply absent. The comb filtering is simply a symptomatic indication of the polar lobing! The comb filtering is not a 'physical phenomenon'...it is an indication of the polar lobing measured at a particular point in space.

And the only portion of the signal with which we are concerned is the frequency which is 180 degrees out of phase at the distance and point in question.

Also, amplitude does NOT "drop off as one over the distance squared."

The sound pressure decreases with a rate of 1/r over distance; NOT 1/r^2 as is so commonly cited. Only the energy density changes at a rate of 1/r^2. The energy intensity is not the same as the sound pressure.

[PaulP]

Originally Posted by barefoot

The Wall Bounce Calculator can be used to simulate your floor bounce. Take a look at the attachment. I've made Mf=10000 cm. It's very large so front wall reflections become negligible. You gave a listening distance of 120 cm. So we set Lf=10120.

I don't believe Marc's monitors are 10m from the front wall. I think the
entire room is about 10m long and the monitor to front wall distance is
around 3m. Marc's room is under a roof and I don't think there are any
vertical walls involved if I've understood things correctly.

Paul P

[barefoot]

SAC, I think don't you fully understand what my calculator is doing. The model doesn’t involve cabinet vibrations. Those would typically only cause small perturbations to the signal from the speaker cones. Sound can diffract around the side of speaker cabinet. It doesn't need to go through the cabinet. And my model accounts for this diffraction. It calculates the difference in directivity of long wavelengths (omnidirectional) versus shorter wavelengths (hemispherical) depending on the width of the speaker cabinet. Once we account for this, we can indeed model the amplitude as 1/R^2.

Furthermore, the model is only concerned with lower frequency reflection effects. Notice that the graph cuts off at 1kHz. It's not intended to be a comprehensive speaker radiation model.

Sure, reflections can indeed cause nulls. But in Marc's specific situation, the floor reflection would not generate deep nulls.

[barefoot]

Originally Posted by PaulP

I don't believe Marc's monitors are 10m from the front wall.

Yes, I understand. But if you read my post you'll see that I set the front wall distance to 10m in order to minimize it's effect and isolate the effect of the floor reflection.

[PaulP]

Originally Posted by barefoot

Yes, I understand. But if you read my post you'll see that I set the front wall distance to 10m in order to minimize it's effect and isolate the effect of the floor reflection.

Right. Sorry about that.

Paul P

[barefoot]

No problem. Looking at it again, I actually set the front wall distance to 100m.

[SAC]

Thomas, ...always a possibility!!!

Diffracted sound behind a enclosure is not the same amplitude as the direct signal, especially above a few hundred Hz! Q and dispersion does indeed matter! I haven't encountered too many true point sources recently.

And while acoustic energy intensity decreases at 1/r^2, the sound pressure decreases at 1/r.

[Nordenstam]

Great posts going on!

Too late and too much saturday at the pub to type properly!

I'll leave this pic here while making some screen shots of some of the hundreds of measurements done today. Hope it's a self evident argument. Reflections are reflections. A source, a boundary(or boundaries) and a receiver. A picture of an antarctic king listening to a Funktion One megablaster says a lot more than 2.718281 words.






Dividing wall between rooms not shown. Both set of reflections occur at once. To complete the picture, add ceiling, other wall, rear wall, double bounces, tripple bounces so on, calculate diffraction, refraction, air and boundary losses, temperature effects, turbulence, and..

Can we agree that SBIR is the right hand side?

[Nordenstam]

Less is more.

Heard that one before?

I haven't.

Not often enough.


Took "%"¤! many hours to try the simple thing.. Did just about every possible permutation of 2, 2.5, 3, 3.5 and four panels to treat that pesky side wall reflection. Some things worked better than others. Two panels was much better than three. And there had to be some on the floor below. But careful, not too much because that didn't work either! A full panel was no good down on the floor, better with some small scraps.. To make matters worse, absorption quite far off seems to influence the result of single "point" treatment. As noted, having some rockwool down on the nearest floor boundary seems essential to treat the WALL reflection. And the corner trap and the ceiling seems to interact as well.. Especially the corner trap. Sigh! Way too complex... Anyway, the important thing: no matter what I did, having two 4" 60kg/m^3 panels consequently made the results worse than the single panel. Took me ages to do the bold move and try that one single panel....

Less obviously IS more. For some things.




So the nearest reflection was down below -20dB. Great. But.. Did lots more measurements, moved the mic back and forward, moved panels and then did the "%¤#% bold thing and moved the speaker and listening position. Moved and moved. Probably more of that to come before it's all settled. Not to talk about moving lots of other things in the room, the readapation to a new modal response and imaging and translation and all. BUT.. Found a spot with better frequency response, and time response is better! It's all below -22dB now. Was strugling with getting below -19dB. That may not seem like much of a difference, but I'm bent on getting it at least below -20db. About -20 to -30dB in the reflections seems to be a general border limit for general music (just don't play any sparse arangement with sharp modern transients!). Having the rear wall as the nearest boundary made it easier to kill reflections (probably due to the polar dispersion pattern of the speaker). Which made the rear wall the primary plase to treat. More on that in next post.

[Nordenstam]

Moving the speakers made the rearwall reflection appear where the sidewall reflectioin used to be. Again, single rockwool panel fared better than any way to use two of those panels. And also again, saw some better effect when treating the floor below as well. Extending to two panels lengthwise was worse. For gravitational reasons, leaning the reflector with a wider gap at the bottom than the top came naturally. This seems to be the best solution (for this problem). Angling the absorber other ways was less effective than a gentle slope from wide distance at bottom to narrow gap at top. Guess it helps making it broadband and stuff. Last one tweaked by changing air gap and angle slightly. Not sure if it's really better than the third measurement shown, it seems more , despite top level being lower. Guess I could move those things around for ever and gain another .5dB.. But.. -22dB initial signal delay RFZ isn't half bad to start with. Time to shift the room around a bit as the speaker move dictates and get to that Haas trigger which was the thing I was supposed to do today. Just had to try that sidewall treatment idea first and one thing led to another..




PS: All measurements and moves doubled to confirm results.

[Nordenstam]

Originally Posted by DanDan

Better is better, who needs to know why?

Some like to know why on a general basis. And its' easier to use a tool when the operative parameters are known entities.

Originally Posted by DanDan

However if one redirects focus to the implications of MP, what does it 'do' in a real life situation?

It foremost means a response can be changed in the same domain using a (minimum phase) equalizer.

Originally Posted by DanDan

May I suggest that the Front Wall reflection and it's big dipper, are a Minimum Phase system. Frequency does not change with distance/time.

Technically speaking, anything having a delay can't be minimum phase. Though the response can be simplified to a phase response curve plus a delay. Removing the delay puts the impulse in the right place timewise. I take it that it means that it should work for EQ'ing down the SBIR peaks a bit. However, the 180' cancellation can't be changed by EQ'ing.


Was the penguin figure alright? There's two "listeners" in a room, the human and the speaker. Both are subject to the same reinforcement and cancellations. Front wall, floor, side wall, ceiling, double bounce, etc doesn't matter.

Originally Posted by DanDan

I don't really understand what you are saying about the Genelec figure. I presume they have measured up to 20dB dips in reality. They claim 6-20dB are typical.
Genelec Oy - Wall behind the Loudspeaker Cancellation

Had the impression that a 20dB dip was pretty hard to attain through a single reflection. In my limited experience, the most bothersome dips have always been the product of several reflections.

[DanDan]

Lupo, I am smiling broadly. I seem to remember that you reckoned my empirical method of moving speakers + testing to optimise locations, was time and effort intensive!
I think it would probably be best to take SBIR to mean the front wall reflection, as heard by the listener, not the speaker. Similarly I think it would be best to consider the side wall reflection in terms of the listener. This would put us in agreement with Thomas' Wall Bounce Calculators.
DD

[Nordenstam]

Originally Posted by DanDan

Lupo, I am smiling broadly. I seem to remember that you reckoned my empirical method of moving speakers + testing to optimise locations, was time and effort intensive!

Hehe.. Well. Given that the choices of placement in a real room are so few, moving through the possible configurations with moving/measuring/moving goes rather quickly. Especially if one does some quick mental computing of room modes to move the speakers in the right direction to balance problem frequencies.

Though, for starters, it's probably a good idea to check room mode charts etc to get an idea of ballpark starting points. Would make it quicker for n00bs that don't take the modal response thing intuitively. I'm using RPG's room optimizer software which goes through five digit numbers of virtual measurements to find ideal positions within the selected limits. Very handy!

Am now so used to my room that I don't have to use the RPG software to find the room modes. Know pretty much where they are. So when I moved the speakers the other day, it only took a short run of move->measure->move operations to find a spot where I coulnd't improve things further. Last night it occured to me that I could run the RPG software to check that I hadn't missed out on something important. Set the operational parameters, waited the ten minutes or so it took to run through 25000 virtual measurements and was very pleased with the result. Within centimeters of the actual placement!





(room is 6x3.9x2.78 meters. One rear corner is angled, making the modal response slightly smaller, so I used 5.9x3.85 meters to compensate)

No need for fancy calculations here though. It's a bog standard golden ratio setup. Though the difference between the worst and best solution found is very small! Without measurements, ending up in the worst case position would be very easy given real life concerns like acoustic center of speaker, deviations from predicted response in real rooms, ec.


Another neat feature of the program is that it spits out position data for the corresponding first order reflection points. Can make life quicker.

Though, as shown in earlier posts, to get the best performance of the treatment of a particular reflection entails fine tuning of the absorber. None of the absorbers I use have ended up exactly at the mirror point..


Am a bit surprised that there haven't been more comments on the drunk posts I made. Probably because they where drunk posts.

But hey.. It did at least surprise me! A single panel giving better absorption than two panels...!! This have far reaching implications.

However, the most mind expanding thing about the exhaustive rounds of measurements was the realization of how big these sound waves actually are. And what it means for treating a refelction "point". If we, for simplicity, assume a spherical propagation from the speaker, a sound is a sphere of about 1 meter in radius after three milliseconds of propagation. This is two meters from top to bottom. Which goes to explain why absorbers on the floor made the reflection point treatment more effective. This also means that corner traps, ceiling traps, etc in the vicinity will effect the treatment.

Complex, indeed!

Originally Posted by DanDan

I think it would probably be best to take SBIR to mean the front wall reflection, as heard by the listener, not the speaker. Similarly I think it would be best to consider the side wall reflection in terms of the listener. This would put us in agreement with Thomas' Wall Bounce Calculators.

Using the RPG program made me realize that I've been abusing the term "SBIR". Had the impression it was for those reflections that does not change with listener position. It's not. RPG's software is ten years old and they use the term SBIR to denote the early reflections. I guess this predates GIK acoustics use of the term. So if we let RPG decide (and I think that's a fair thing to do given their historical precende), SBIR means:

"The SBIR reflects the coherent interference between the direct sound and first order reflections."

Page 18: http://www.rpginc.com/products/roomoptimizer/ROVMAN.pdf

[DanDan]

Lupo, at least your posts have shone a light on the other SBIR's.
To be honest, I think the concept of reflections as heard by the speaker is a bit far fetched. However I don't criticise your logic, surely those high pressures will affect cone movement. As usual creative.
Speaking of which, how about a bursting balloon as a kick drum sample....
No, I'm not drunk....yet, maybe later.

So when I moved the speakers the other day, it only took a short run of move->measure->move operations to find a spot where I coulnd't improve things further.

Were you optimising to minimise your pesky reflections or for best frequency response, or all of the above? Did you ultimately measure a 'real' frequency response and how did this compare with RPG predictions?

That program is fascinating. However as you also noted, the proximity of best case to worst case is at least strange.

I wouldn't trust a computer to make these decisions, however I would ask specific questions of it. Note that Meyer use or used to use human operators for their SIM system. Also Air Traffic Control. We're still need for some jobs!
As Mr. Everest said, despite rapid learning and active research for 100years, acoustics still appears to be as much an art as a science.

DD

[Nordenstam]

Originally Posted by DanDan

Lupo, at least your posts have shone a light on the other SBIR's.
To be honest, I think the concept of reflections as heard by the speaker is a bit far fetched. However I don't criticise your logic, surely those high pressures will affect cone movement. As usual creative.

It's not that far fetched. After moving speakers, I had a big dip at the exact frequency corresponding to the distance from woofer to ceiling back to woofer. Placing an absorber in that reflection path solved the issue.

Originally Posted by DanDan

Speaking of which, how about a bursting balloon as a kick drum sample....
No, I'm not drunk....yet, maybe later.

Have built some analogue drum synths. What you describe is the impact part of the sound, the initial impulse that simulates the drum hitting the stick. It lacks the oscillator, the resonance of the drum.

Originally Posted by DanDan

Were you optimising to minimise your pesky reflections or for best frequency response, or all of the above?

The problem was that I didn't get the sidewall reflection low enough. I figured the polar dispersion pattern of the speaker means that it sends more energy to the sidewall than the rear wall. Which turned out to be the case. The rear wall is easier treat as the first reflection point since there's less energy to start with. The sidewall treatment works like a charm now that it got some distance to it.

I don't think it's because the distance means losses in itself. It's too small a distance for that. I think the reason is that when the sound have moved 2 millisecs, it have evolved into a (distorted) sphere of 2 feet radius. This doesn't give much physical room for treatment. When the distance is closer to four millisec, the sound balloon have expanded to ~four feet radius. Which means that corner absorber, floor absorber, ceiling absorber etc comes into play, helping take some energy from the reflection.

EDIT: just have to say that realizing this complex interaction of several absorbers treating a reflection "point" was as close to a mind expanding experience as I've ever had in acoustics.


Once the basic premise was set, that sidewall should be further away than rear wall, actual placement of speaker was done based on modal/SBIR response.

The frequency response is different. Don't know if it can be said to be better or not. It's about the same plus minus magnitude as before, but the problem areas have of course moved around a bit.

Originally Posted by DanDan

Did you ultimately measure a 'real' frequency response and how did this compare with RPG predictions?

Sure! I usually start out forcing the program to compute the response at the current position of a room and compare that to the real world measurement. Most of the time, it's close enough. Sometimes, due to light wall constructions etc, it's so off that it's not useful. The measured frequency response looks about the same as the graph posted above, at least in the biggest problem areas. Close enough to prove that it's not much point in trying to move things around any more.

Originally Posted by DanDan

That program is fascinating. However as you also noted, the proximity of best case to worst case is at least strange.

Notice that speakers/listener is not only displaced on the floor plan as seen from above, it's also displaced in height. Enough of a difference to make the modal/SBIR response significantly different.

Originally Posted by DanDan

I wouldn't trust a computer to make these decisions, however I would ask specific questions of it. Note that Meyer use or used to use human operators for their SIM system. Also Air Traffic Control. We're still need for some jobs!

Sure! I never use the program alone without measurement. But it's a very good guide for finding rough placement!

[Northward]

Minimum Phase & Room Acoustics... Pandora's box is now opened.

[SAC]

Minimum phase is not a problem at all.

It is the fact that many non-minimum phase systems, which by definition exhibit "excess phase" - meaning that you are dealing with signals that are not all aligned in time, and where you lack a one to one correspondence between magnitude and phase between the real and imaginary realms - create complex problems that must be addressed in the time domain.

And anomalies caused by the interaction of non-minimum phase signals cannot be solved by EQ.

[Northward]

Originally Posted by SAC

Minimum phase is not a problem at all

It's not to those who understand / know what it is.

In some aspects it can be a tricky subject though, even for those in the know (and by minimum phase I mean generally speaking both minimum and especially non-minimum phase issues) .

Was just joking about the possible threads on the subject

Back to our usual program.