MJK OB-bass study

[MJK]

A 1 m distance seems to be an established reference point for evaluating and comparing SPL response curves. In the OB case presented the baffle is only 20" wide, a max driver to edge distance of 10", so a 1 m distance should not exhibit the effects you have described. The listening position was set as a common reference from the previous article and is in fact more than 1 m away from the center of the driver. Any user of the MathCad worksheets can select any listening position, so a study of the effects you describe is possible.

 

[riff.ca]

Hi Martin,
In reading your U & H frames pdf I came across the following.

"The U and H frame designs used a common cross-sectional area and cavity depth. The internal cross-sectional area is 16” wide and 16” tall. The depth of the cavity was defined as 7.5”. The depth was selected to push the first quarter wavelength resonance above the desired crossover frequency of approximately 200 Hz.

Leffective = 7.5” + 0.6 x reffective
Leffective = 7.5” + 0.6 x 9.0”
Leffective = 12.9” = 0.328 m

f1/4 = c / (4 x Leffective)
f1/4 = 344 m/sec / (4 x 0.328 m)
f1/4 = 262 Hz"

I don't follow where the '0.6' and '9.0 inches' come from?

 

[skorpion]

Thanks a lot AJinFLA and Martin,

The link to Linkwitz webb is: http://www.linkwitzlab.com/

I followed Martin's instructions and got this result for 2 Eminence Alpha 15s in parallel in a H-baffle of double height and the same width and length as in Martin's paper (and all other parameters unchanged):

An externally hosted image should be here but it was not working when we last tested it.

/Erling

 

[john k...]

I have an article on dipole woofer and listening distance here:

http://www.musicanddesign.com/Dipole-axis.html

On a couple of the figures I have misslabled R4 = 8 instead of R4 = infinity.

It's probablyimportant to consider distance from the wall behind the speaker as well because as a dipole woofer system gets close to a wall the bass drops off. The rear wall reflections cancel the free field bass response.

 

[MJK]

Hi Martin,
In reading your U & H frames pdf I came across the following.

"The U and H frame designs used a common cross-sectional area and cavity depth. The internal cross-sectional area is 16” wide and 16” tall. The depth of the cavity was defined as 7.5”. The depth was selected to push the first quarter wavelength resonance above the desired crossover frequency of approximately 200 Hz.

Leffective = 7.5” + 0.6 x reffective
Leffective = 7.5” + 0.6 x 9.0”
Leffective = 12.9” = 0.328 m

f1/4 = c / (4 x Leffective)
f1/4 = 344 m/sec / (4 x 0.328 m)
f1/4 = 262 Hz"

I don't follow where the '0.6' and '9.0 inches' come from?

The 0.6 and 0.9 are a calculation of an end correction, similar to a port in a bass reflex or the open end of a transmission line. The end correction length is approximated by 0.6 x average radius. The average radius of a 16 x 16 square is about 9 inches.

 

[MJK]

It seems to me that a mic to driver distance that is fairly close to, or smaller than, the dimension of the baffle front will give an FR plot with a lot more bass than the room actually hears. The mic picks up a lot more direct sound from the driver, as opposed to the wrap around from the back which cancels the lower frequencies at distance.

An issue - or not?

I ran the OB worksheet when I got home increasing the listening distance from 1 m to 3m and then 10 m at a constant heigth of 32 inches above the floor. The OB SPL response curve maintained the same shape but just at a lower SPL due to the greater distance. I conclude that the effect you are describing is not present in the calculated SPL responses that are presented in the document.

 

[MJK]

The link to Linkwitz webb is: http://www.linkwitzlab.com/

I followed Martin's instructions and got this result for 2 Eminence Alpha 15s in parallel in a H-baffle of double height and the same width and length as in Martin's paper (and all other parameters unchanged)

Erling,

You got 93 dB down to about 30 Hz in a relatively small package!

Very nice,

 

[AJinFLA]

It's probablyimportant to consider distance from the wall behind the speaker as well because as a dipole woofer system gets close to a wall the bass drops off. The rear wall reflections cancel the free field bass response.

xlbaffle is a sim that does just that, although real rooms have lossy walls and most will angle their speakers inward....so caveats apply.
Still a useful tool. Free too.
(Hopefully this post is not heretical to church doctrine and thereby "constructive" 🙂 )

cheers,

AJ

 

[Pano]

I conclude that the effect you are describing is not present in the calculated SPL responses that are presented in the document.

Hmmm..... OK. Not quite sure what this means, but I'll think about it.

I certainly do see a loss of bass as the mic moves away from the baffle in actual measurements. Always thought it was because the mic was hearing a larger and larger proportion of the back wave vs. front wave and thus canceling. But heck, it may just be my living-room's lousy acoustics.

When I get a chance (maybe next week) I'll try it outside to see what comes up. Will post results.

 

[riff.ca]

The 0.6 and 0.9 are a calculation of an end correction, similar to a port in a bass reflex or the open end of a transmission line. The end correction length is approximated by 0.6 x average radius. The average radius of a 16 x 16 square is about 9 inches.

Thanks Martin, that just wasn't in my DNA 😉

If I understand this then for a U frame 24 x 31 with 11" wings we get...

Leffective = 11” + 0.6 x reffective
Leffective = 11” + 0.6 x 15.5”
Leffective = 20.3” = 0.515m

f1/4 = c / (4 x Leffective)
f1/4 = 344 m/sec / (4 x 0.515m)
f1/4 = 167 Hz

I've assumed that the average radius of a rectangle is the radius of a circle with the same area.

 

[MJK]

I agree with your calculations. That is the answer I would get for your geometry.

 

[riff.ca]

Thanks again Martin!

This mirrors what I thought I was hearing though I'd guessed the bump freq to be closer to 200 HZ. I'm going to remove one of the wings tonight and listen to what that does.

My XO is at 375 HZ so I doubt I can get the 1/4 wave resonance above that and still have a baffle. Does anyone have comments on angling the wings. Seems like a reasonable idea on the surface at least.

 

[MJK]

If you angle the wings out, small area at the driver and expanding to the open end of the U frame you will raise the fundamental frequency. So your 167 Hz will go up significantly. In the limit, a flat OB, the standing wave frequency will go to infinity and no resonances will exist.

 

[skorpion]

There are some things you could try if you would like to stay with your relatvely high crossover frequency. The simplest way which I have practised in two different OBs is just taking away the top of the U. MJK also has this solution in his Big OB with 2 Alpha15s and the Lowther. You wouldn't loose any efficiency.

My exemple is 'The Volks-OB' with dimensions for the bass-U 24" x 16" x 12" (HWD) here: http://www.audiocircle.com/circles/index.php?topic=46951.0.
There are measurements of this bass-baffle and some discussion in this thread: http://www.audiocircle.com/circles/index.php?topic=48896.20 .
I also had an discussion with MJK about resonances in this type of 'toppless U-baffle'.

Another way would be to try to notch out the resonances and crossover steep but then you have to be able to measure.

/Erling

 

[Lynn Olson]

One thing to keep in mind when comparing 1 meter vs 3~5 meters is the dispersion pattern of the drivers in their working ranges. If they match, the FR measurements will track as well. If they are dissimilar, voicing the system for one distance will lead to (large) errors at a different distance.

Many of the mathematical models assume idealized omni, or half-omni, radiation patterns over the working range of the driver. If the HF driver is a horn, this is no longer true - the power does not drop off in the expected inverse-square-law ratio with distance, but at a substantially slower rate, due to the higher directivity index, or putting it another way, a tighter dispersion pattern. (Think of a laser - almost no drop off with distance at all.) Since real-world drivers have frequency-dependent dispersion patterns, a system that is EQ'ed flat at 1 meter will almost certainly be wrong at larger distances.

In real rooms, this is complicated by the critical distance, where reverberant and direct-arrival energy have equal power. At distances larger than the critical distance, power drops more slowly than predicted by the directivity index (polar pattern), and eventually reaches a steady state. In domestic listening environments, though, the critical distance may not exist anywhere in the room - this applies mostly to large spaces like auditoriums and movie theaters.

This is why a series of measurements that are averaged over the intended listening area area a good idea - and a lot more accurate than a model that is much simpler than the real physical drivers and a real acoustical space.

 

[MJK]

Many of the mathematical models assume idealized omni, or half-omni, radiation patterns over the working range of the driver.

The modeling of drivers in my worksheets does not treat them as point sources radiating into half space. The driver is modeled as a collection of sources that can add constructively or destructively at the assigned listening position, the driver and openings are all directional. Also, the baffle size and shape is included in the calculated SPL response as well as floor and rear wall reflections. My ongoing goal for the MathCad models is to continue advancing the calculations while closing the gap between the simple models used in many of the speaker simulation programs and real world in room measurements.

 

[ucla88]

One thing to keep in mind when comparing 1 meter vs 3~5 meters is the dispersion pattern of the drivers in their working ranges. If they match, the FR measurements will track as well. If they are dissimilar, voicing the system for one distance will lead to (large) errors at a different distance.

Many of the mathematical models assume idealized omni, or half-omni, radiation patterns over the working range of the driver. If the HF driver is a horn, this is no longer true - the power does not drop off in the expected inverse-square-law ratio with distance, but at a substantially slower rate, due to the higher directivity index, or putting it another way, a tighter dispersion pattern. (Think of a laser - almost no drop off with distance at all.) Since real-world drivers have frequency-dependent dispersion patterns, a system that is EQ'ed flat at 1 meter will almost certainly be wrong at larger distances.

In real rooms, this is complicated by the critical distance, where reverberant and direct-arrival energy have equal power. At distances larger than the critical distance, power drops more slowly than predicted by the directivity index (polar pattern), and eventually reaches a steady state. In domestic listening environments, though, the critical distance may not exist anywhere in the room - this applies mostly to large spaces like auditoriums and movie theaters.

This is why a series of measurements that are averaged over the intended listening area area a good idea - and a lot more accurate than a model that is much simpler than the real physical drivers and a real acoustical space.

I'm not sure what the directionality gain is for horns. But monopole radiators in a typical home room have a fairly short reverberation distance. 3-4 feet wouldn't be unusual.

 

[sreten]

In domestic listening environments, though, the critical distance may not exist anywhere
in the room - this applies mostly to large spaces like auditoriums and movie theaters.

Hmmm.......

This is a new one on me. Unless you have line arrays / long ribbons
which due to being linear fall-off with distance rather than squared,
or the more extreme case a pair of monster front horn loaded
speakers which can be even more directional.

Generally speaking as a far as I'm aware and in my experience
(having had a pair of ribbons) the cd does does exist in nearly
all listening rooms and for many people more than the cd is
the preferred listening distance.

http://www.pispeakers.com/ssdm_99.pdf

Section 5 has a good introduction to the issue.

You must have a very dead room not to have a cd in a room.
(or highly directional speakers).

FWIW the ribbons did push back the cd - making the experience
more near field than one would expect - but the corollary of this
was that it made conversation uncomfortable when wanting to
have background music rather than a serious listening session.

🙂/sreten.

 

[Lynn Olson]

All of the previous posts are good points. One thing to keep in mind about measurements made beyond the critical distance is the measurement technique makes a difference: if it is a time-averaged measurement like 1/3 or 1 octave pink noise, the direct-arrival energy and total room energy are summed (time information is discarded). Conversely, an FFT or MLS measurement looks only at the first-arrival energy (time information is preserved), unless the window is opened to a very large value like more than a 100 mSec (which adds up direct-arrival and most of the room reflections).

The summing of direct-arrival and total room energy becomes significant for pink-noise measurements made beyond the critical distance, and affects the measured spectral balance, while by contrast a FFT or MLS measurement with a "normal" direct-arrival window of 0 to 5~10 mSec responds only to the direct-arrival energy and ignores the contribution of the room.

The controversy arises from the perceived spectral balance made by direct-energy, direct+room, and room spectral balances. I'm in the old-school camp that gives greatest weighting to the direct-energy contribution, while other people assign relatively more weight to room energy.

This is why different designers have different definitions of "flat" - they're using different measurement techniques, different weightings of direct vs room energy, and different measurement distances. Note the results from these measurement techniques (pink-noise vs FFT/MLS) diverge more strongly when the speaker has markedly different directivity indexes for different parts of the spectrum (as most speakers do).

The biggest variation in DI/frequency is probably a speaker with a closed-box woofer/midbass and traditional exponential/sectoral horn treble - the woofer is omnidirectional below 300 Hz or so, while the horn typically radiates across 90 degrees horizontally, and substantially less in the vertical direction. A speaker like this will have quite different balances depending on measurement distance, liveliness of the room, and measurement technique (pink-noise vs FFT/MLS). In a very real sense, there is no absolute "flat" measurement - the results all depend on the technique.