? How are peaks over 100db considered average levels? The "average" levels being referred to there are as THX says: "85db with 20db of headroom" (yet still louder than is pleasant, even on compression drivers in my opinion).
Which music have you analyzed on a peakmeter which ever resembles a flat spectrum? The only thing I can think of is extremely compressed rock and pop mixed right up to the edge of digital clipping, and I hope you're not arguing that such are transient in nature. Even EDM music, which is artificial in nature, seems to be inherently associated with a HF rolloff in its peak spectrum. Or at least, do you know of an AES paper which makes the argument that musical transients resemble a flat spectrum?
Playing with an SPL meter on "fast" setting, and switching between unweighted and A-weighted tends to indicate the same to me. I don't design my speakers for artificial electronic tones with a flat sound power "peaks" because I've empirically observed any indication of their existence.
I use the Foo_uie_peakmeter_spectrum plugin with foobar2000 and it indicates that my assumptions are not only safe, but conservative. That doesn't however mean that "any direct radiating driver" can be used. I of course doubt the behringer tweeter has the excursion even with the waveguide, and likewise the Orion's Millenium tweeter on a flat baffle probably doesnt satisfy the above scenario.
Last edited:
Ribbons on waveguides sound excellent.
I think it's due to the superior CSD.
For instance a five kHz sine wave is .2ms long.
If a tweeter rings for five milliseconds, then it will not be able to resolve that sine wave accurately. (Because of the ringing.)
It's too bad they're so prone to failure.
I have never properly heard a horn loaded ribbon. Came over one yesterday, a SLS PRD500. It was broken....
http://www.slsloudspeakers.com/PRD500datasheet.doc.pdf
http://www.slsloudspeakers.com/PRD500datasheet.doc.pdf
I'd be curious to hear the ribbons in these:
SoulSonic Speakers
I've often wondered what a ribbon would sound like in a horn of a sort similar to some of these models - although some seem not to have curved surfaces, but some variant on a conical horn.
SoulSonic Speakers
I've often wondered what a ribbon would sound like in a horn of a sort similar to some of these models - although some seem not to have curved surfaces, but some variant on a conical horn.
It's a shame the website don't show any meaningful data or measurements. Never a good sign.
About a year ago I posted a bunch of data I'd collected from various people, to illustrate what happens. Here's a summary:
1) As noted in my previous post, you need a fast decay on the diaphragm for high frequencies. For instance, to reproduce 5khz the diahpragm would ideally settle in two tenths of a millisecond! (Because that's the length of 5khz.) And of course, this is complicated by the fact that music is complex, and a 2khz tone may be playing at the same time as 5khz. Basically, a fast decay is a good thing and ribbons decay very very quickly.
2) If you look at the CSD of various drivers, you'll notice a correlation between decay times and mms. Heavy diapragms take time to settle.
Based on points 1 and 2, you might think "it's time to stop using compression drivers, ribbons kick butt."
Unfortunately, it's not that simple. The ideal radiator for a waveguide is small. If the throat is larger than about 1", it becomes very difficult to get good polars and good high frequency extension, no matter how light the diaphragm is.
So we have a catch-22 here. Ribbons have great CSD, but even the smallest ones have a large diaphragm because their displacement is so small. Compression drivers have an MMS that's much higher than a ribbon, but their diaphragm is nice and small.
This is the reason I tend to use tiny compression drivers, like the CDX1-1425. It gives you a compromise between the two extremes.
If you want to dig up the data, dig around for some of my old posts on this, particular the ones with the measurements that I used from Brandon (Augerpro's) website.
Also, I think another solution which would work well is a ring radiator on a wave guide, like what Peregrine(sp?) Acoustics uses. He's posted lots of data too. He's also put ribbons on waveguides also.
Is it the weight of the diaphragm or the control of the motor that gives good CSD?
This question has been discussed thoroughly on subwoofer forums.
Also, I am a believer that larger diaphrams are better at all frequencies.
Less travel equals less IMD and doppler effect. (this is not opposing the horn thought theory) it's just an observation.
Lest not forget AMT's. Great surface area, lite and fast.
I'm making a modified eJMLC-1000 with an Airborne RT4001 AMT. Preliminary data suggests this is a workable match. The aperature of this tweeter is 1.5" (34mm) H x 1" (25mm) W, is made up of 5 horizontal slits stacked atop another. Using oil based clay sealed off the top and bottom slots. This left the aperature 1" wide x 3/4" high, which can then be rotated 90 degrees for better horizontal dispersion. The results of this tiny mod had little adverse effect.
I'm making a modified eJMLC-1000 with an Airborne RT4001 AMT. Preliminary data suggests this is a workable match. The aperature of this tweeter is 1.5" (34mm) H x 1" (25mm) W, is made up of 5 horizontal slits stacked atop another. Using oil based clay sealed off the top and bottom slots. This left the aperature 1" wide x 3/4" high, which can then be rotated 90 degrees for better horizontal dispersion. The results of this tiny mod had little adverse effect.
It would be both of course. You have a mass and a resistance in the form of a motor structure. The ratio of the two determines the Q and hence the ringing. This is more true of the lower modes, the higher tend to not be so strongly magnetically damped. But I would caution putting too much emphasis on CSD. There is not much evidence that it means anything - in fact Toole and Olive don't use it at all.
IMHO, Dan Wiggins did a good job of debunking the woofer speed argument.
The thing that's funny is that people obsess about woofer speed, but no one worries about tweeter "speed."
Also, I hate that term, what we're looking for is a fast decay.
The important thing here is that "fast decay" is frequency dependent. For instance, we would ideally want a 5khz signal to decay to in two tenths of a millisecond. (This is because 5khz is 0.2ms long.) As frequency gets lower, the window of time gets longer. At 500hz a decay of 2ms would be ideal.
So let's look at some data.
Here's the CSD of a B&C 8". As mentioned a moment ago, we'd like to see 500hz decay in 2ms or less. The B&C achieves that goal, for the most part. 500hz decays about 20dB in 2ms.
Here's a B&C DE250. We'd like to see 5khz decay in 0.2ms. Unfortunately, the DE250 can't do that. In two tenths of a millisecond, the signal at 5khz has only decayed about five decibels.
The Celestion CDX1-1425 has a diaphragm that's a small fraction of the DE250. And we can see that it improves the CSD. It's not as good as a ribbon, but it's better than the DE250.
And THAT is why the drivers with small light diaphragms sound "airier", IMHO. Basically fast decay is far more important for high frequencies than low, because the wavelengths are so much shorter. I think that what a lot of people interpret as "slow bass" is actually the reflected energy from the room itself, not the woofer. At low frequencies, wavelengths are long and therefore good CSD performance is easier to achieve.
measurements courtesy of John Krutke (zaphaudio) and Brandon (Augerpro)
Oh, another observation:
Note that at 2khz, the 8" woofer has better CSD than the B&C DE250 compression driver.
The CSD of direct radiators is generally superior than the CSD of drivers mounted on waveguides and horns. And that may explain why the direct radiating 8" driver has cleaner CSD at 2khz than the horn-mounted DE250.
CSD measurements may also be useful in detecting HOMs. For instance, let's say you have a distance of 5.67cm from the surface of a compression driver diaphragm to it's throat. And let's say that HOMs are generated at the throat, due to the impedance mismatch cause by the sudden change in expansion rate. If that happens, we would expect to see the CSD "elongated" by one half of a millisecond.
The additional 1/2 millisecond of delay is caused by the HOMs traveling 17cm. (Sound travels 34cm in one millisecond.)
Hope that makes sense.
Geddes reticulated foam should also clean up the CSD.
The tricky part with all of this is that it will require some very careful measurements to detect, because we're talking about differences measured in a few decibels, over a time frame of a thousandth of a second or less.
This may also explain why Harman doesn't bother with this stuff; you'd really need to know what you're looking for, and it would be easy to jump to invalid conclusions.
An anechoic chamber would help
Note that at 2khz, the 8" woofer has better CSD than the B&C DE250 compression driver.
The CSD of direct radiators is generally superior than the CSD of drivers mounted on waveguides and horns. And that may explain why the direct radiating 8" driver has cleaner CSD at 2khz than the horn-mounted DE250.
CSD measurements may also be useful in detecting HOMs. For instance, let's say you have a distance of 5.67cm from the surface of a compression driver diaphragm to it's throat. And let's say that HOMs are generated at the throat, due to the impedance mismatch cause by the sudden change in expansion rate. If that happens, we would expect to see the CSD "elongated" by one half of a millisecond.
The additional 1/2 millisecond of delay is caused by the HOMs traveling 17cm. (Sound travels 34cm in one millisecond.)
Hope that makes sense.
Geddes reticulated foam should also clean up the CSD.
The tricky part with all of this is that it will require some very careful measurements to detect, because we're talking about differences measured in a few decibels, over a time frame of a thousandth of a second or less.
This may also explain why Harman doesn't bother with this stuff; you'd really need to know what you're looking for, and it would be easy to jump to invalid conclusions.
An anechoic chamber would help
That and Toole/Olive pooh pooh the whole idea. All I am saying is that there is nothing scientific that points to CSD telling us anything new. And there are lots of skeptics. Just because you can measure something does not make it important.
Also, HOM will likely not appear as you say. They should appear as later arriving signals separate from the main signal. I have seen some CSD done by Marshall Buck which I believe clearly showed HOMs.
meh.. did not read the whole thread.
just wanna point out that blind tests have nothing to do with stuff instruments can measure.
it is subjective. and not objective.
there are many very cheap speakers that for example i preferably find far better than professionally made ones.
i find OB to sound poor, and dislike tubes and transistors.
and i hate flat response curves. takes life away from speakers.
as to why did the "imp" get out as number 1, no exact clue.
one of the best ever speakers i did have chance to listen to was a cube with 5 coaxial car speakers. has something to do with them reflections.
ideal point source radiator thingy theory, not bad,
if that point source has a spherical radiation pattern equal in every direction.
but there is no such thing.
btw, i can't think of any instrument that will radiate as point source, how on earth is it expected to be accurately re-created by a point source ? right.
never will be good enough.
for some people horns, ob, bandpass, blablabla, everyone has a preference.
blind tests are not that honest if you think of it.
sure, they don't know if the speaker is a horn or not.
or ob or not.
or..will they?
how on earth is anyone going to tell, if they can or can't tell sub-concious ?
they will surely not know they are biased, and nor will you ever tell.
main problem is that there is no definition of a good speaker.
some definitions do apply, but those are generic stuff, not at all hard to manage.
other specs include some exotic things that can be measured, but not sure if they make a difference or not.
just wanna point out that blind tests have nothing to do with stuff instruments can measure.
it is subjective. and not objective.
there are many very cheap speakers that for example i preferably find far better than professionally made ones.
i find OB to sound poor, and dislike tubes and transistors.
and i hate flat response curves. takes life away from speakers.
as to why did the "imp" get out as number 1, no exact clue.
one of the best ever speakers i did have chance to listen to was a cube with 5 coaxial car speakers. has something to do with them reflections.
ideal point source radiator thingy theory, not bad,
if that point source has a spherical radiation pattern equal in every direction.
but there is no such thing.
btw, i can't think of any instrument that will radiate as point source, how on earth is it expected to be accurately re-created by a point source ? right.
never will be good enough.
for some people horns, ob, bandpass, blablabla, everyone has a preference.
blind tests are not that honest if you think of it.
sure, they don't know if the speaker is a horn or not.
or ob or not.
or..will they?
how on earth is anyone going to tell, if they can or can't tell sub-concious ?
they will surely not know they are biased, and nor will you ever tell.
main problem is that there is no definition of a good speaker.
some definitions do apply, but those are generic stuff, not at all hard to manage.
other specs include some exotic things that can be measured, but not sure if they make a difference or not.
The Behringer dcx2496 does appear to have significant analog circuit design problems. See my discussion at:
Bob's Behringer mod
I think the reason open baffles or di-poles in general sound better, when they do, is because the inevitable inter-aural crosstalk ruins the imaging cues below about 1kHZ. Di-poles or open baffles (same thing generally) recreate a false but largely effective set of timing cues in the lower midrange that can compliment the amplitude imaging cues of the upper midrange, and give us an impression of full bandwidth good imaging and thereby more sense of depth. When you consider the limitations of how recordings are made, you realize that artificial enhancement during playback may not always be as bad as we'd like to believe, if it's done right. Ultimately, if it's more fun to listen to, isn't that what we all want?