How do you chose speakers based on room size?

Small rooms are the most difficult. You have at least three acoustic issues in a small room:

1. Early onset of reflections (causes coloration and/or compromised imaging);

2. Worse room modal situation than in larger rooms so the bass is lumpier;

3. Often excessive room gain in the bass region (depends on many factors including speaker and listener locations).

In a very large room, the speaker's power response (summed omnidirectional response) tends to dominate the perceived tonal balance, so that matters a lot more than the on-axis response does (unless you listen nearfield), and often in a large or open-floorplan room the boundary reinforcement is insufficient in the bass region so the speaker tends to sound thin and weak. Also the higher power levels required to reach satisfying SPLs in a larger room can tax a speaker both thermally and mechanically.

I recently did a fair amount of experimenting in the course of investigating a small-room-friendly speaker system, and came to some rather ironic conclusions. For example, in order to get the radiation pattern control that gives good performance in a small room by minimizing early reflections, a physically rather large speaker is required.

Now there is a school of thought that calls for very aggressive use of absorption on the walls of a small room, and this relaxes the radiation pattern control requirement. Unfortunately this approach also eliminates beneficial late-arriving reverberant energy.

I recommend spreading out the bass sources as much as is practical regardless of room size as this smooths the in-room bass response (I can explain why if you'd like). Even a little bit of spacing can be beneficial; for instance, a two-way floorstander with the port on the rear down near the floor has the two bass sources (woofer and port) fairly far apart in two dimensions, and this will usually sound smoother than having the woofer and port displaced in only one dimension. If excess boundary reinforcement is an issue, that can often be addressed by reducing the port tuning frequency (which is easier to do than raising the tuning frequency).

Duke
dealer/manufacturer

J Bailey, your experience with the large-format Klipsch Chorus II's in a relatively small room makes sense to me. They probably benefit from generous boundary reinforcement, and are directional enough to avoid the onslaught of early reflections that one typically is met with.

C1ferrari/Sam, regarding DSP, once in a large room (ballpark 80 feet by 120 feet) we used DSP as a "room/system treatment", and it was definitely an improvement. My speakers didn't have adequate bass output in that size room especially with the positioning we had to use, although they did have enough excursion capability to handle the EQ. A rather sophisticated prosound DSP package was used, and we spent about an hour taking measurements in a wide variety of microphone positions.

Also when Robert E. Greene (of TAS) had a pair of my bipolars in his room some time ago, he experimented with at least one DSP room-correction system (something he is quite fond of). He reported in his online forum that the DSP correction made very little change to what the speakers were doing, which he said was indicative of good in-room behavior.

Two things that DSP cannot do anything about are the loudspeaker's radiation pattern and physical limits (both thermal and mechanical). So to get the most benefit from a DSP-corrected system, in my opinion it's a good idea to start out with speakers that do a good job in those two areas.

Saki70, you have very good questions. I'm going to avoid getting too specific here because I don't want my post to cross the line and become an "ad".

Driver integration is dependent on driver vertical spacing, crossover frequency & slope, and listening distance. Briefly, the ear is poor at resolving the height of a sound source below about 1 kHz, improves dramatically between 1 kHz and 4 kHz (where it peaks), and then actually decreases a bit at higher frequencies but remains pretty good. Steep-slope crossovers give better driver integration (less vertical smearing) at close range than shallow-slope crossovers. In my experience, having a suitably low crossover frequency is more beneficial to driver integration than is close inter-driver spacing. Perhaps Clio09 will post here, as he's the one who really opened my eyes (and ears) on this subject.

Saki70, you mentioned that in a larger room we want to crank it a bit but in a smaller room we want to turn the volume down. One factor that comes into play here is the thermal modulation characteristics of the drivers. Usually the tweeter is more efficient than the woofer so it's padded down, and it normally gets a lot less power anyway. So many speakers run into the problem of the woofer's voice coil heating up more than the tweeter's voice coil, so the woofer has more thermal compression as we turn the volume knob higher. This voice coil heating is virtually instantaneous; a 100-watt transient is like touching a 100-watt soldering iron to the voice coil. Anyway, if the woofer's voice coil is heating up faster than the tweeter's it will have more thermal compression at high input levels. The designer then has to choose an input level at which the drivers are balanced relative to one another, and he will probably choose a fairly high level, let's say 90 dB/1 meter for this example. If we go to 100 dB, the tweeter will get loud a bit faster than the woofer because its thermal compression is less, so the speaker will sound a bit bright on peaks. If we go down to 60 or 70 dB, now the tweeter is softer than the woofer so the speaker sounds a bit dull and lifeless. I think this phenomenon is behind the fact that many speakers do not really "come to life" until you crank 'em up a bit.

So to sum up the preceding paragraph, it's not uncommon for a speaker's tonal balance to change with the input power level, going from dull at low levels to "just right" (the Goldilocks zone) to too bright at very high levels. If a speaker is going to work well at a wide range of power levels, either the woofer and tweeter need to have very similar thermal compression characteristics or their departure from linearity should happen at higher input power levels than we're likely to see in the home.

On a related note, low thermal compression correlates with speakers that convey emotion well. Musicians use variations in loudness to convey emotion, and it's nice if the speakers can preserve those dynamic shadings. If a speaker is compressing the peaks by 2 or 3 dB, well then ou lose emotional impact.

So anyway in order to meet the requirement that a speaker work well at low levels in a small room and also at high levels in a large room, a lot of issues have to be addressed. I think it's easier to address them with prosound drivers, but won't claim that's the only feasible approach. But large-diameter prosound woofers and high quality constant-directivity horns & waveguides combine good directional control with the ability to move a lot of air should they be called upon to do so.

I don't think it's possible to design a speaker whose characteristics in the bass region do not change significantly as its position in the room (and/or the room itself) is changed, so I think the best we can do is build in a reasonable amount of adaptability that will hopefully cover most situations.

Duke

Thanks, Halcro. There can be other reasons for the perceived tonal balance to change with level, but I think differing thermal compression characteristics among drivers is usually the main one.

Heykay, even though your 12 by 19 foot room would probably be considered "medium sized", the fact that it's open into another large room means that it's in effect a "large" room as far as the bass region goes. My opinion would be, either go with speakers "voiced" for a large room or consider using a sub with the intention of adding a little bit more energy across the bass region (rather than merely filling in below the main speakers, which is the more common usage for a sub). In many cases an opening into an adjoining room acts as a huge bass trap and can actually be beneficial, provided you can get the bass level up enough to offset the loss of bass energy.

Duke

"What design characteristics would be favorable for speakers in a small room that would not be favorable for speakers that were to be used in a large room and visa/versa?"

Well this depends somewhat on where the speakers are located within the room but in my opinion in a small room you generally want the bass rolloff to start fairly high but it can be gentle. This should synergize better with the anticipated boundary reinforcement than a speaker that is "flat" down to a fairly low frequency. Conversely, in a big room you need speakers with more bass output, relatively speaking, because the room isn't going to help you out as much.

In any size room I like fairly uniform radiation patterns in the midrange and treble region. In a small room I like a fairly narrow pattern that can be aimed to minimize early sidewall reflections, but there are people who like omnis in small rooms (the reflections from an omni are likely to be spectrally correct, which is desirable). I like to have at least a 7 millisecond delay before the onset of reflections other than the floor or ceiling bounce, and that corresponds to a path length difference of about 7 feet. So something like baby Maggies located 3.5 feet in front of the wall would probably be feasible (I'd prefer 5 feet out, but we're talking about the minimums here).

In a medium or large room where we'll have a good 10 milliseconds or more before the onset of reflections from nearby walls, we can freely use speakers that have a wider pattern (dipoles, bipoles and omnis conceptually fit into this category if they can be placed far enough out into the room)). Such speakers give a greater relative level of reverberant energy in the room and in my opinon more closely approximates the sort of soundfield we experience at a live performance. Now there are competing schools of thought that object to "adding" reverberant energy via wide-pattern speakers, but that's another topic for another day.

So to recap, you usually need a lot of bass output in a big room but not nearly as much in a small room, and (in Duke's opinion anyway) wide-pattern speakers are better suited for medium or large rooms than for small rooms.

Saki70, I'd have to see the speaker (and preferably know its crossover frequency) to make a judgment about its radiation pattern.

The room gain you get in a small room plays right into the hands of a good back-loaded horn that exits into a corner. That's an excellent, very high bang-for-the-buck format in my opinion. For instance I don't see how I could compete with what Ed Schilling is doing in that area.

And yup, your example of big Altecs in small Asian rooms is consistent with what I'm talking about.

Mapman, I'm encouraged by how many people find success with the Ohms in a small room. The Ohms aren't just wide-pattern speakers; they are wide, uniform pattern speakers - so their reverberant field is spectrally balanced. I suspect that there is a correlation between how good they sound in a small room and how far away from the walls they are positioned. Having experimented with fairly uniform-pattern quasi-omnis, that was my conclusion: Get 'em out from the walls and they can work great in a small room. My finding was that if they can't be pulled out at least 3.5 feet into the room, then controlled-pattern monopoles work better.

Have you found a correlation between sound quality and distance from the walls with your Ohms?