Effects of cabinet design on "in room" bass response

[vinylnvalves]

Having recently built a MLTL test cabinet to explore the pro's and con's of this approach for a mid bass (30- 300Hz) enclosure, i was suprised by how room effects contributed to the FR response of the system. Due to the size and weigth of my test cabinet, i carried out measurements in my workshop, with the cabinet in the middle of the 8m x 4m floor. I initially used my usual approach to speaker measurement, placing the mic in the listening position, in this case the distance (2.5m) away from the speaker. The response did not replicate the simulation from MJK'S mathcad work book for MLTL's. Following advice (DIYaudio) i moved the mic closer to the speaker, carrying out near field measurements @ 300mm away the response and simulation are engineeringly similar. Using MJK's room placement predictor some of these room effects can be predicted, with the position of the driver and ports being significant contributors.

Having gone and confirmed that FR of my Onkens are also alot flatter when measured in the near field, in my listening room - I conclude the LF solution needs to be optimsed with the room.

This draws me to my question, ignoring the extreme effects i witnessed (15 db peaks and troughs) of my hard surfaced workshop, are all LF solutions as sensitive to room contribution? After the worst offenders have been precluded, how do i tune driver/port position, baffle size/shape etc to work with the room.

 

[f1eng]

The room, and the position of the speakers in it, makes more difference to the sound than anything else, except the speaker itself IME.

Genelec have some advice on speaker positioning to minimise bass peaks and troughs. There has been some modelling software about which a friend of mine used to help me position my speakers, but that was over 10 years ago so something else, probably easier for the non-expert to use, may be available now.

The other thing having a vast effect on the sound of speakers is radiation from the cabinet. It is impossible to measure, since measurement can't tell what comes from the cone and what comes from the box, but since the radiating area of the box is so much greater than the cone it can be a significant contribution of non-accurate sound to the whole. Recently speaker designers have used computer models to calculate the relative contributions of box and driver which has lead to some very fine sounding speakers.

Getting a good box design is, IMHO, very much more difficult than getting good drivers.

 

[vinylnvalves]

Thanks for that.

I read somewhere on the web (sods law cannot find it now) that open baffle bass was more difficult to integrate into a room than say an infinite baffle or bass reflex. I was therefore wondering would a front loaded bass horn or a line array for example be easier to integrate. Common sense and physics would suggest to me a point source would be the easiest, but an interested in other peoples experience/opinion.

 

[sjs]

If the speakers produce series energy below 100hz the interaction with room boundaries and standing waves can have a huge influence on the FR, whilst the floor and wall construction can influence the impulse response and hence clarity of the LF.

For positioning I have found CARA Cad to be very easy to use and very useful for optimising speaker and listener position. You could probably then compare the results from Cara with in room measurements.

 

[RobHolt]

The room, and the position of the speakers in it, makes more difference to the sound than anything else, except the speaker itself IME.Genelec have some advice on speaker positioning to minimise bass peaks and troughs. There has been some modelling software about which a friend of mine used to help me position my speakers, but that was over 10 years ago so something else, probably easier for the non-expert to use, may be available now.

The other thing having a vast effect on the sound of speakers is radiation from the cabinet. It is impossible to measure, since measurement can't tell what comes from the cone and what comes from the box, but since the radiating area of the box is so much greater than the cone it can be a significant contribution of non-accurate sound to the whole. Recently speaker designers have used computer models to calculate the relative contributions of box and driver which has lead to some very fine sounding speakers.

Getting a good box design is, IMHO, very much more difficult than getting good drivers.

Agreed, and to illustrate the point this measurement of cabinet output, measured at the panel surface of a cabinet we're currently working on, illustrates the effect of simply changing the bracing or cross linking between parallel facing side panels. We've found that for a BBC style ply thin-wall cabinet, bracing does actually help considerably if you treat the 'braces' more as linking rods.

The movement in the two opposing side panels is in anti-phase, therefore getting them to couple within a given frequency range should allow partial cancelation of output, and it does.

Brace too hard and unwanted output starts to appear up in the mid range, exactly as the BBC research found all those years ago.

Here you can see the difference between using one cross link and three - a 15-20dB reduction in output.



crosslink by trebor1966, on Flickr

 

[Tenson]

Just to add to this, the way to measure the cabinet rather than the drivers is with an accelerometer that contacts directly with the surface. You need to average a number of points to get a real representation of what the whole panel is doing though.

 

[paulf-2007]

Thanks for that. I read somewhere on the web (sods law cannot find it now) that open baffle bass was more difficult to integrate into a room than say an infinite baffle or bass reflex. I was therefore wondering would a front loaded bass horn or a line array for example be easier to integrate. Common sense and physics would suggest to me a point source would be the easiest, but an interested in other peoples experience/opinion.

you shouldn't believe everything you read on the web, I would say open baffle integrates with the room far easier than a box speaker. Even an open backed box like I use now works as well as it is really just a folded baffle.

 

[karlinamillion]

The room, and the position of the speakers in it, makes more difference to the sound than anything else, except the speaker itself IME.

I 100% concour

The other thing having a vast effect on the sound of speakers is radiation from the cabinet. It is impossible to measure, since measurement can't tell what comes from the cone and what comes from the box...

Are you familiar with acoustic cameras Frank?

http://www.acoustic-camera.com/en/acoustic-camera-en

http://www.campbell-associates.co.uk/saleacocam.htm

 

[Tenson]

That's a fantastic idea! I'd seen something like this used for acoustic measurements to identify the location of reflections in concert halls but this takes it much further. Thanks very much for sharing. Any idea on cost for the low-end model?

 

[karlinamillion]

Cost? no idea! The Campbell lot do rent them out, but not cheap.

Not sure if 'bleed' from the actual driver may cause a prob. I have seen a pic of one very close to the (door) panel being measured, about 10cm.

 

[h.g.]

That's a fantastic idea! I'd seen something like this used for acoustic measurements to identify the location of reflections in concert halls but this takes it much further. Thanks very much for sharing. Any idea on cost for the low-end model?

I worked a bit with a couple of GFAI cameras a few years ago. The company seems to have a grown a bit since then so they must have sold a few. They are fairly specialised devices in so much as the optimum arrangement of microphones and the software tends to vary from application to application. I doubt they would be of much use for loudspeaker R&D. Costs were the wrong side of 100k but they are not that difficult to build and since we would have preferred a different microphone arrangement and had had to write different software to process what we required that was a proposed future project. Hardware parts costs were estimated at about 10k-20k if I recall correctly. GFAI rent out the systems but it is not cheap (I have the figures somewhere for a few weeks plus training).

 

[Tenson]

Not the sort of thing to rent just for fun then!

Not sure if 'bleed' from the actual driver may cause a prob. I have seen a pic of one very close to the (door) panel being measured, about 10cm.

What are we talking about here? Door panel?

 

[h.g.]

Not the sort of thing to rent just for fun then!

Not from GFAI but a few hours work from an acoustical consultant that had one may cost 1k. It depends what you want to do with it. A university project could be free if the topic is appropriate and you are not in a hurry.

The image does not have much dynamic range and so a cabinet panel near a loud driver will not be seen. You need to be interested in the loudest source that the microphones are picking up. So the rear panel in an anechoic chamber could work but the front baffle will not.

 

[f1eng]

Interesting. I wasn't aware of the acoustic camera, but have been aware of FE methods of calculating radiation from cones and cabinets.

The KEF white paper on the LS50 design and Fink Audio consulting on the design of the Q-sound Concept 20 both are very informative. The "deadness"of a KEF LS50 cabinet has to be experienced to be believed.

 

[h.g.]

Interesting. I wasn't aware of the acoustic camera, but have been aware of FE methods of calculating radiation from cones and cabinets.

Yes full 3D numerical simulations can be the most effective way of going about engineering so long as all the relevant physics is contained within the simulation. Unlike an experiment, a simulation contains quantitative information on every physical value at every location throughout the device. This enables questions like "why", "how much" and "what if" to be efficiently and reliably answered. It requires an investment in computer hardware and software and an investment in training people to think in terms of fundamentals rather than black boxes and design formulae. Although dominant in high tech industries, this investment is not necessarily cost effective for a small company in a low tech industry.

 

[f1eng]

Yes full 3D numerical simulations can be the most effective way of going about engineering so long as all the relevant physics is contained within the simulation. Unlike an experiment, a simulation contains quantitative information on every physical value at every location throughout the device. This enables questions like "why", "how much" and "what if" to be efficiently and reliably answered. It requires an investment in computer hardware and software and an investment in training people to think in terms of fundamentals rather than black boxes and design formulae. Although dominant in high tech industries, this investment is not necessarily cost effective for a small company in a low tech industry.

Simulation is -much- less expensive and more accurate than it was in 1972 when I first started doing it! The biggest limits were computer power and the absence of the physical data, particularly damping, for any of the components of our model!

Modelling using more or less off-the-shelf FE methods is trivial in comparison, not to mention many times more accurate.

Fink Audio Consultants do speaker design exploiting modelling for several clients, most of whom, presumably, do not wish it to be known that their products are not all their own design...