I'm sorry guys but I just don't see the point in this discussion. I can't see how anything that I do would change or why I should even consider the 1/4 wavelength thing in any real situation. So have fun, but I'll stay out of it from here. I've stated my points, there is nothing more to say.
Ydope said:
You are actually technically correct. I really shouldn't have said the 1/4 wave cancellation is a modal effect. What I should have said is that modal analysis will show it to occur. Modal analysis is just one means of determining the in room SPL as a function of source and measurement positions. Modal analysis is not limited to application below the Schroeder frequency. It applies to all frequencies. However, it becomes a less attractive means of analysis as the modal density becomes high.
Markus,
I don't think the problem is with what simulation can or can not do. The problem is that we don't simulate the real room. If we were to design a room on paper with certian specifications for the variation of the surface admittance over the floors, walls ceiling, etc and then build that room I am confident that we would find predictions in pretty good agreement with measurements. The problem is we don't build that room. Typically for a home environment we start with an existing room and try to simulate it by guessing what the admittance over the room surfaces are.
The other point I would make is that it just isn't cost effective to apply simulation at the required level to a "one off" home listening room. Now if we were trying to design a "standard listening room" which would be duplicated by the thousands, then that would sway the equation towards making simulation cost effective.
markus76 said:
I think John provides a reasonable explanation below. In reality, people designing studios to use more complicated simulation. Software such as Dirac and Odeon, if I recall correctly, really allow you to model the environment more accurately. Normally the results are close enough such that little tuning is required.
The thing is very few are going to spend a fortune just for one private listening room.
john k... said:
HiFiNutNut said:
Hi HifiNutNut,
Have you solved your issue with the dip?
I have had similar issues but luckely l have found the source of problem.
I do not have understood if your listening position is "far" from the wall behind you.
Have you measured with one or two speakers simultaneous?
One pink noise generators or two?
I agree about many times problems in the 100-400Hz are more
fastidious than below 100Hz
Cheers,
Paolo
Markus
Do you mean just the LFs and the multiple subs?
The problem is that at LFs the leakage and HVAC and all those litttle things that we ignore in our simple simulations add up. A real room with real furnishings and real wall stffnesses, etc. would be horrendously complicated. At Ford we did some auralizing simulations for cars - a much easier problem. It was marginally effective - good first cut, but no better than just looking at the measurement data. There are issues each way you go and no one approach is ideal - except trying it of course.
I agree with John, the problem is too big to be handled effectivily in detail, but we can use the models to find trends and approaches that seem to hold out advantages. This was exactly how I came to my approach on multiple subs. I did some models found some trends and then implimeneted several applications. The trends held up - in general - but each application was different. Each one had some constrains that did not allow for exact application of the ideals and needed something unique to be done to get the details right.
The room that we had in Thailand would have been an ideal case
to model because it was so simple in construction - it was a sealed perfect box and only had one couch and very precise absorption placement. But that is a very rare situation.
Do you mean just the LFs and the multiple subs?
The problem is that at LFs the leakage and HVAC and all those litttle things that we ignore in our simple simulations add up. A real room with real furnishings and real wall stffnesses, etc. would be horrendously complicated. At Ford we did some auralizing simulations for cars - a much easier problem. It was marginally effective - good first cut, but no better than just looking at the measurement data. There are issues each way you go and no one approach is ideal - except trying it of course.
I agree with John, the problem is too big to be handled effectivily in detail, but we can use the models to find trends and approaches that seem to hold out advantages. This was exactly how I came to my approach on multiple subs. I did some models found some trends and then implimeneted several applications. The trends held up - in general - but each application was different. Each one had some constrains that did not allow for exact application of the ideals and needed something unique to be done to get the details right.
The room that we had in Thailand would have been an ideal case
to model because it was so simple in construction - it was a sealed perfect box and only had one couch and very precise absorption placement. But that is a very rare situation.
markus76 said:
It isn't the limitations of the software. The software is actually rather trivial. It is the limitations of the user to define the required input, and the computational power required. The average user doesn't have a clue as to how define the input nor does he have the necessary computational power in a PC. It's really a catch 22. If you want to correctly define the input you would need to make hundreds, maybe thousands of measurements of your room to determine what the surface admittances are.
John
I think that its even more difficult than that. A wall (or window, door) that flexes doesn't actually have a single wall impedance - its a fluid coupled structure - that would be very difficult to measure and specify. And then there is the problem that the real situation may not even have a closed form solution like both of us use in our "simpler" models. One would have to use a full finite difference model like are just now being developed (see the latest JAES).
I think that its even more difficult than that. A wall (or window, door) that flexes doesn't actually have a single wall impedance - its a fluid coupled structure - that would be very difficult to measure and specify. And then there is the problem that the real situation may not even have a closed form solution like both of us use in our "simpler" models. One would have to use a full finite difference model like are just now being developed (see the latest JAES).
I'm well aware of real world limitations but nonetheless there's no reason to stop thinking of ideal tools. Today I can do high end acoustic measurements with my PC at home that were far too expensive only 20 years ago.
Earl, how far is your dummy head project? Would love to hear your home theater with different speakers.
John, software isn't only about algorithms but mostly about usability. Do you know CARA
Best, Markus
Earl, how far is your dummy head project? Would love to hear your home theater with different speakers.
John, software isn't only about algorithms but mostly about usability. Do you know CARA
Best, Markus
markus76 said:
Markus
Interesting that you should ask. I ran into a guy that I knew many years ago at my sons holiday play and he was using a Neuman dummy head to do some recording. We talked and he is very interested in what I want to do and believes that the system would work ideally in that application and offered his help.
Now I need to arrange to get "other" loudspeakers in my room to record and compare. This later aspect is now the problem as there aren't many people with "good" loudspeakers that can bring them here. In fact it may not even be a reasonable thing to do at all and I may have to go to the few installations that I know of that are even remotely comparable to mine. I am definately going to record the Summas, Abbeys and Nathans all in the same exact situation as I am personally very interested in this.
gedlee said:
Will you respond to "CABS" by CELESTINOS and NIELSEN as proposed in the November issue? It's basically a double bass array (DBA) as described by Goertz/Wolff/Neumann 5 years ago (sorry, only available in German): http://www.monkeyforest.de/Page10383/Aktuell_dt/DAGA2003-Paper.PDF
Best, Markus
gedlee said:
Remember I am looking at this from the position of 30+ years experience in CFD (computational fluid dynamics) and related areas. This is really just a different type of aeroelasticity problem, i.e. coupling of a fluids problem to a structural problem. The problem is riddled with details, but still isn't all that complicated.
markus76 said:
I looked briefly at their web page some time ago but you can't tell much from that.
Form a mathematical point of view room acoustic is a simple interior problem for which the solution is governed by boundary conditions at the bounding surfaces. As with any mathematical problem it is the boundary conditions which make the solution unique. In this case the boundary conditions are what happens on the surfaces that define the room. The solution offered by any room simulation tool, mine, Sound Easy's FEA analysis, CARA or otherwise, is only going to be as good as 1) how well the interior physics is modeled and 2) the representation of those bounding surfaces; How much acoustic energy is lost to the surface? How much is stored and then returned to the interior. When a wall flexes is it because it is acting as an accelerating mass in response to an applied force, or a spring?
The interior is easy. The wave equation is suitable. And there are any number of way we might attempt to solve the wave equation. But what of the boundary conditions? We might parameterize them. For example we might assume that all sheetrock walls have similar properties, windows have a different set of properties, HVAC vents some other properties, doors.... and then segment areas of each surface in to the appropriate generic family. But if your wall or door or window doesn't fit the generic model the results still won't match with measurement.
We all know how sound waves propagate. The problem is what happens when the wave interact with a surface.
I tend to agree. Normally CDF is used to solve very specific problems. Very seldom if any is it used to model blanket range scenarios. Even with todays PC computing power, it takes too much computing time could take a few days; and then if the results are not good enough, human intelligence still is needed.gedlee said: