Investigating Dipole Bass in a Small Room

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I recently became aware of a theory that says dipole speakers are incapable of producing deep bass in small rooms. Essentially, the theory asserts that below the resonant frequency defined by a room's lowest modal frequency corresponding to its longest dimension, the dipole cannot pressurize the room and so it cannot produce any bass below this frequency.

I decided to perform an experiment to test the theory. I placed a dipole speaker in a small room and measured its frequency response. The results of my experiment are at http://www.landtime.com/perm/smallroomdipolebass.htm.

- Eric
 
Thanks for correcting the URL. Those darn periods. . . .......

I used a Linkwitz Orion speaker. The Orions are a three-way actively crossed-over system. The midrange and bass drivers are configured as dipoles, while the tweeter is not. The crossover frequency between mid and woofers is about 140Hz IIRC. There are two woofers mounted in an H-frame baffle. Since they are sitting on the floor, they have a 2pi radiation pattern, while the midrange is 4pi. Dipole eq, 2pi/4pi equalization, etc, is performed by the standard electronics in the Orion's crossover/EQ. In other words, the speaker was set up to play as usual for full-range music reproduction.

The speaker was placed 1.5' from the side walls, 2' from the wall in front of it, and about 3' from the wall behind. The front/back asymmetry was due to the step in the floor. I would have tried it in different positions but the options were pretty limited in this small room.

The corrections applied to the RS meter measurements are 100Hz/0db, 60Hz/+1db, 40Hz/+2db, 30Hz/+4db, and 20Hz/+9db. There are different sets of corrections floating around the net. I got this particular set from a friend who did his own analysis of the meter. In any case, the raw data shows no significant drop off in the bass below resonance.

- Eric
 
It seems you did not take into account the frequency response of the Orions themselves in the results?

You are correct. I saw no point in doing this for several reasons.

1. The room is swamped by modal resonances.

2. There must be significant room gain too. This is contrary to what John Kreskovsky wrote. If you read the first part of his article, he says that room gain has no effect on dipoles because it's a phenomenon caused by room pressurization -- and since dipoles cannot pressurize the room (in theory), there's no room gain to be had. That doesn't seem to be the case though.

3. The Orions roll off fairly low (-3db/30Hz and -6db/20Hz). This is well below the frequency of interest (56Hz) where theory predicts that there should be precipitous drop in bass levels. If I did adjust the figures for the Orion's rolloff, they would show even more bass production below 30Hz. What is shown, unadjusted, is sufficient to test the hypothesis.

- Eric
 
Couple of things you might want to consider:
1. Room gain, a room that small it is going to kick in early and give a large boost to low frequencies.
2. Distortion from the woofer, It may be very small if they are good woofers but if they produce alot of harmonic distortion when you put the 20-30hz and maybe a little higher like 40hz then you will get some higher harmonics being produced which will interact with the room the same as when you play the actual higher frequency. The spl meter does not know what frequency it's hearing, only spl. This will throw off your measurements, obviously.
3. Why did you choose such a small room,(bathroom?) why not run this experiment in the living room? The statements from the website are for realistic size rooms and Im sure he didnt feel it necessary to discuss such small areas.
4. Also keep in mind that the dipole cancelation effect is not equal in every square inch of the room. It is constantly changing from perfectly in phase(constructive interference) to perfectly out of phase(destructive interference) and everywhere in between. The importance of this is the low decay time/arrival time of the small room. A larger room will have longer arrival times(reflected and direct) and you will notice more profound cancellation. This
Java applet will help you visualise the effect, set it to room mode, 1source/1freq, with boundary, set freq so it appear to be about 10- 20 pulse emissions per second(youll see what I mean),3d mode helps but not necessary.
 
Originally posted by nunayafb 1. Room gain, a room that small it is going to kick in early and give a large boost to low frequencies.

Unless I misunderstood John's articles on his web site, the underlying rationale for the theory is that a dipole cannot pressurize the room and that room gain is a due to room pressurization. So if room gain is present and accounts for the measured response, the theory is based on an incorrect premise.

Originally posted by nunayafb 2. Distortion from the woofer ...

I believe the XLS drivers were chosen for the Orions in part because of their very low distortion.

Originally posted by nunayafb 3. Why did you choose such a small room,(bathroom?) why not run this experiment in the living room? The statements from the website are for realistic size rooms and Im sure he didnt feel it necessary to discuss such small areas.

Small is small, isn't it? Shouldn't the physics be the same in a small room or smaller room?

Actually, it would take some extremely large dipoles to test the hypothesis in a "normal" living room. I just did the calculations for a 20x16x8 room. The transverse distance is about 26' corresponding to a resonant frequency of 22Hz. Is it of any practical significance to anyone if a dipole can or cannot produce high SPL bass below 22Hz? Virtually no speakers can, other than subwoofers of course.

By choosing the smallest room in my house, the resonant frequency could be pushed up well above the point where the speakers roll off. My living room is much too large for the experiment because the lowest resonant frequency there is about 12Hz.

Originally posted by nunayafb 4. Also keep in mind that the dipole cancelation effect is not equal in every square inch of the room.

The room clearly has lots of modes and irregularities in its response, perhaps some of it due to unequal dipole cancellation. I wouldn't want to listen in there, that's for sure. Still, I don't see how this changes anything about the hypothesis.

- Eric
 
Eric,

are you going to measure the sound profile of your bathroom or the validity of John K´s theory?:D

I believe that this missing (or not missing) room gain of the dipole speaker should not be measured against the anechoic frequency response of that speaker but relative to the room gain of a conventional loudspeaker, preferably a closed box, in the same room.

You surely have noticed that John K. did this in all his diagrams. Would be nice to see that comparison in your experiment too.

I have heard a few comments of dipole owners who found their speakers working "according to John" in their rooms. So I am not yet a believer in your results.

But thanks of course for looking deeper into this problem.

Rudolf

BTW: Are your room dimensions really translating to 1.12x1.65x2.25 m? If that is possibly true you could not honestly expect a speaker with the size of the Orion to work as a dipole in that space. It could behave more like an infinite baffle with some big leaks inside.
 
the underlying rationale for the theory is that a dipole cannot pressurize the room and that room gain is a due to room pressurization
Ok, good point, I missed that from his article, but there are other factors in this little room, kind of a weak answer but think about it for a minute. What is the rolloff like in the living room? I bet it rolls off in accordance with the equations that you used.

Small is small, isn't it?Shouldn't the physics be the same in a small room or smaller room?
Yes same laws, but the results can be different, in a "normal" size room the pressure remains constant see the link you posted.
Something is obviously different in this extremely small room, hence the unusual results.

it would take some extremely large dipoles to test the hypothesis in a "normal" living room.
No, it would take a normal size dipole like everyone else is using. Did you ever ask yourself why everyone is talking about their "dipole rolloff" and why everyone who has tested dipoles in their listening room has measured rolloff below this room mode? Why dont you give it a try in a different room and analyze the correlation between fundamental room mode and frequency rolloff?

The reason he mentioned small rooms is because large rooms have longer paths for standing waves to travel and the energy in the colliding waves (front and rear)whether constructive or destructive, is too low to cause major fluctuations in the target spl. I would imagine that in a room as small as yours the waves are coming at the mic from so many different angles that they cant actually fully cancel. You might try Frequency Response Plotter and see if the 1hz resolution changes the small room response, ie. displays narrowband dips/peaks in the response that were not revealed before.
 
Rudolf,

Your message doesn't make any sense to me, but thanks for the joke.

How could the dipole be behaving like a leaky infinite baffle in this small room when the radiation from both sides of the driver is being injected into the same room? What do you think is absorbing the rear wave?

The plane of cancellation through the baffle was completely unobstructed (except for the floor of course). The two faces of the drivers were moving in opposite phase with respect to the baffle. What more does it take for a dipole to function?

Comparison to box speakers may have helped John to develop or illustrate the theory, but it has no impact on the behavior of the real dipole that I measured in a real room.

I haven't read anything about this theory that qualifies "small" in a "small room". In normally sized rooms (like the example I gave in a previous post) the fundamental frequency is so low that the entire theory is moot. In the small room that I used for testing, you're implying that test is invalid. So what exactly is a small room that will exhibit the predicted rolloff?

Can you explain how the room gain is affecting the test when the hypothesis is predicated on the dipole not being subject to room gain?

Like I said, your message is confusing... :confused:

- Eric
 
nunayafb said:
No, it would take a normal size dipole like everyone else is using. Did you ever ask yourself why everyone is talking about their "dipole rolloff" and why everyone who has tested dipoles in their listening room has measured rolloff below this room mode? Why dont you give it a try in a different room and analyze the correlation between fundamental room mode and frequency rolloff?

The lowest resonant mode in my listening room is 12Hz (diagonal of approx 45'). To get 12Hz out of a dipole would require significantly more displacement than this "little" two-woofer dipole has. To test well below this frequency to find out if there was precipitous loss of bass would take some really massive dipole woofers IMHO.

It might be worth throwing in some anectodal evidence here since you brought it up. I had set up the system in an 11x17x8 room so a friend could get an idea about what they might sound like in his similarly-sized room. (I consider this to be a small room for listening. Is it small enough to qualify as "small" in the hypothesis under test?) The longest diagonal distance is 21.5', resonant frequency is 26Hz. We played the sweep tones just to get an idea of how bad the resonant nodes and suckouts were. There were a few biggies to be sure. But we didn't use a meter, we just listened. At 20Hz, everything in the room was rattling as much as at 30Hz. Subjectively, I was surprised at how loud it was at 20Hz.

- Eric
 
I have corrected an error in the experiment's discussion section. I mistakenly took the corner-to-corner distance as defining the lowest resonant mode. In fact, it is the longest wall-to-wall distance that defines the resonant frequence. The oblique modes that can be energized between corners have higher frequencies.

This doesn't change the results of the experiment. Instead of looking at the bass response below 56Hz for dropoff, the lowest resonance is 75Hz. Bass output was as constant below this frequency as it was below 56Hz.

- Eric
 
Eric Weitzman said:
I have corrected an error in the experiment's discussion section. I mistakenly took the corner-to-corner distance as defining the lowest resonant mode. In fact, it is the longest wall-to-wall distance that defines the resonant frequence.
Eric,
hopefully I can make my point less confusing this time: :(

If we still consider your Orion working effectively as a dipole in your experiment (which I doubt BTW), the relevant "longest wall-to-wall distance" for determining the lowest room mode is NOT the vertical distance, since room modes in vertical direction are only faintly excited by the Orion. Decisive is the horizontal on-speaker-axis distance of 5.5´, which results in a lowest room mode at ~100 Hz.

When looking at your diagram, I see a kind of peak at 100 Hz and a drop of 20 dB from there to 60 Hz, at least in the sweep diagram. That´s a 20 dB drop in less than half an octave.

Comparing to John Ks diagrams I see almost the same drop off between 40 and 20 Hz (or 30 and 15 Hz respectively).

So thank you for backing up Johns findings with your measurements so well. :D
 
Let's be clear about the diagrams on John's page. THEY ARE SIMULATIONS. I actually measured the small room response of a dipole.

I'll discuss John's Figure 1 for a moment, assuming it reflects reality.

We should not compare the bass level to the resonant peak at 28Hz, but rather the average level above. The average level above the fundamental for both simulated speakers is about -25 to -30db. The dipole is at this same level at 20Hz while the simulated monopole is up 15db. The dipole doesn't drop at all in the normal listening range. Checking the sub-bass to 5Hz, we find the dipole down 5-10 db, while the monopole is up nearly 30db. I don't think the problem lies with the dipole unless you want your speakers to implement ISO226.

Perhaps the simulation correctly takes into account the xmax limitation of the dipole, which would explain the drop at 5Hz? Keep in mind that the dipole was necessarily also simulated.

In my sweep measurement, the bass below resonance is 5-10 db down compared to the level from 120-160Hz. In the burst measurment, it's actually 5db above! There is nothing like the precipitous drop that John's simulation appears to show below the resonant peak at 28Hz in his Figure 1.

I think you and I have gone back and forth on this enough now. You are interpreting my data and John's simulation differently -- and incorrectly IMHO -- than I am.

- Eric
 
Hi Eric,
Thanks for publishing your experiment! I applaud everyone who actually tests first hand. Nothing beats testing it for yourself. Great work!

I'm not too surprised by the difference between the experimental and simulated results. Simulations always have to make assumptions about reality, and as such, rarely reflect it. They're a useful tool, but can never replace an actual measurement.

John K's theoretical work seems to stand up to scrutiny. In a sealed room (or one with a relatively small leak through a single doorway) where all (almost all) of the energy is maintained within its boundarys, pressurisation would seem the only available mechansim for creating a sound field below the room's fundamental resonance.

Eric, your experiment takes into account flexible boundarys that are able to store and re-transmit energy into the listening environment. I think this is the critical point where your experimental setup differs from the simulations. A pressure wave developed from one side of the diapragm will impinge upon a closely spaced wall and cause a sympathetic vibration from it irrespective of dipole cancellation in a small room, especially where baffle size is a significant fraction of the room's dimensions.

I suspect that a typical listening room will lie somewhere the simulations of John K and Eric's measurements, with an amplitude response that falls off below the lowest room mode, but not as quickly as simultions might suggest.
I know from my own measurements that dipole theory leaves something to be desired in real rooms with real drivers and real baffles. Although I haven't investigated response below the lowest room mode as I use a sealed sub.

It would be very interesting to see a direct comparison between a dipole and monopole in the same experiment.

Cheers, Ralph
 
Eric,

I too applaud your effort, but I have to agree with Ralph and Rudolph. You should be happy though, because you proved that part of the theory was incorrect, just not the one you intended.

The only way to get close to pure dipole radiation in our small rooms is with the speaker in the center of a symmetrical room. The flaw I see in the theory is that it's not the lowest room mode, but the on axis room mode that is important. Below this frequency it makes perfect sense that response will decrease rapidly, because as the wavelength increases, the front and rear wave (and their reflections) become closer to directly out of phase and the baffle size becomes of minor importance. It doesn't have anything to do with room pressurization, since there is none with dipole. What matters is the phase relationship of the front and rear wave like always when you're talking about dipole bass.

I read your results the same as Rudolph, the drop in SPL below 100hz proves what you were trying to disprove. It's the irregularities of the room shape, construction and speaker placement that resulted in the leveling off of response with construction the most likely culprit since the LF waves were going right through the walls and a higher percentage of the rear wave going through through the door than the front wave going through its wall of 1st reflection.

An easy test would be to use a relatively long narrow room. Put the dipole speaker in the center of the room and play a tone slightly above the frequency of the long dimension mode. Measure the output with the speaker axis along the long dimension and with the speaker turned to the other axis. I guarantee that the on axis output (not nearfield) will be lower along the shorter dimension.
 
> Below this frequency it makes perfect sense that response will
> decrease rapidly, because as the wavelength increases, the front and
> rear wave (and their reflections) become closer to directly out of
> phase and the baffle size becomes of minor importance.

I don't see the difference between what you described and the reason
dipoles roll off at 6db/octave. What does this have to do with room
size? Do you think a larger baffle would somehow make a difference?

> I read your results the same as Rudolph, the drop
> in SPL below 100hz proves what you were trying to disprove

The drop in SPL below the modal peak at room resonance -- to a
level commensurate with that above resonance -- shows that the theory
is wrong.

- Eric
 
Eric,

Imagine a speaker in 1ft room and nice long bass wave coming out of the front and back 180 degrees out of phase. You're going to have near perfect cancellation because at any point the waves and their reflections will be almost directly out of phase, netting to almost zero, and it has nothing to do with baffle size. It is room size dependant. Then just work back to where this effect starts to occur and it will be at the modal frequency.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.