This is why being able to "read" polar maps is so useful. If the aberation you are trying to correct is constant along a vertical line, like a mechanical reonance in the driver, as you suggest, then EQ is effective. But if it is diffraction or acoustical modal related then it will be different at every angle and correcting one angle will inherently make all the others worse.
You have to understand what it is that you are trying to correct and "black box" global solutions can't do that.
There are some aberations in my speakers for which EQ can correct - such as a spider resonance - and others for which EQ has no hope of correcting, like the axial response dip from the mouth diffraction.
My own waveguided Heil AMT
http://www.diyaudio.com/forums/planars-exotics/171441-hornloading-heil-amt1-14.html#post2561630
http://www.diyaudio.com/forums/planars-exotics/171441-hornloading-heil-amt1-14.html#post2561630
Talk about directivity and CSD, here is a different data set measured at 0 and 30 deg.
Not really sure to what measurement the eq was based on. But the match of 0 and 30deg SPL is quite unexpected.
An externally hosted image should be here but it was not working when we last tested it.
Not really sure to what measurement the eq was based on. But the match of 0 and 30deg SPL is quite unexpected.
I don't understand. Both are meant for sound reproduction.
Nor do I get the analogy; which is which?
Is the one with UE supposed to be better?
From what I can see, the resonances are somewhat narrower but with longer decay times.
From what I can see, the resonances are somewhat narrower but with longer decay times.
That's where you have to make a decision by listening. Measurements are sometimes effected by things we do not have total control of. I was asked to turn off UE during the show to see what happens, and the comment was "it still sounds good". For me, the UEed one sounded better. Is it the ultimate end? Not by a long range. The fun is in the journey to see what one can actually do to make changes that show up as improvements both technically and through listening.
They look like inaudible decay though, when the FR is 80dB and the first ripply is below 50dB Im not going to be concerned and over 18Khz is meaningless to me too. The 1Khz to 5Khz region is what I see has a huge improvement, the 2nd one (EQed, linear phase) looks to me like the "cleanest" sounding tweeter possible. I like that sort of sound (hence why I like Ribbons).
Last edited:
Actually, it's a small full range driver. Trying to get that 20kHz breakup under control took great efforts, and work is still going on to see if there is any way to redesign the driver for better performance.
The fast initial decay gets critical when loud playing occurs. For example, if the music is at 85db and you can get it to drop below 60db very fast, then this part of the system has little effect in sound coloration, and the room becomes dominant.
The fast initial decay gets critical when loud playing occurs. For example, if the music is at 85db and you can get it to drop below 60db very fast, then this part of the system has little effect in sound coloration, and the room becomes dominant.
I guess I wasn't looking at them correctly.
I was looking at the tail end timewise, where the mounds look bigger, failing to note the fast initial decay.
I find these difficult to read; shouldn't the color correspond to dB?
You can count the lines to figure out the time, but it's hard to tell the level because of the 3Dness.
I was looking at the tail end timewise, where the mounds look bigger, failing to note the fast initial decay.
I find these difficult to read; shouldn't the color correspond to dB?
You can count the lines to figure out the time, but it's hard to tell the level because of the 3Dness.
I generally find CSD's easy to read.. BUT very few are presented in the same manner.
For instance you really have to look closely to see that the scaling and "widowing" is NOT the same on the two graphs.
Yeah, I guess there are many ways we can look at them. I tried to look for the greatest separation of the actually response vs any decay response.
More or less the more dark red shown the better
The "raw" driver starts at about 420 Hz, the EQ driver starts at about 310 Hz.
Additionally the pressure of the "raw" driver is about -3db down from the average at 420Hz where it starts "relative" to the EQ driver. Of course the difference is more prominent than that because the "raw" driver is likely even lower in pressure near the 310 Hz of the EQ driver.
The real difference between the two (assuming they are identical except for equalization), is that the "raw" version displays additional decay relative to the average due to it's undulating freq. response (i.e. it's not "flat"). The EQ version by contrast is displaying the driver's "actual" linear decay character (..mechanical resonant "stored-energy" behavior), for a single axis (presumably the frontal 0 degree axis).
I personally think you need both the "eq" and "non-eq" versions to get a "grip" on the overall behavior of the driver's stored energy. Plus, I think the CSD should be derived for multiple axis's. (..assuming it's eq.ed, only that one axis will have a "flat" behavior at higher freq.s..)
I'd think there's only be minor variations with angle, since (apart from Fs) non-ideal CSD is dominated by the driver's mechanical vibration behavior.
Any-time the angle changes the response to "non-flat" - that shows up as additional energy in the CSD plot's decay. Of course if it's really only a minor change then any change in the decay will be minor.
Why's that?
If anything I'd expect the opposite - sound radiated by axisymmetric anomalies like cone breakup and surround resonances would be in phase and sum to their maxima on axis.
True.
I don't fully understand what does what, but up to now, I tend to feel that EQing to near field measurements provide for more accurate reproduction. I guess this means EQing the power response. Recall that Earl put lots of emphasis on power response.
I also posted EQed CSD of 0deg and 30deg from a different test session on a different speaker. Probably those were EQed from near field measurements. Since now I got some interconnects tuned to where it seems optimum, I'll be going back to listen at louder levels to see what listening impressions I get.
Only if you make sufficient measurements to properly account for the polar response.
I presume your thinking is that with farfield response reflections integrate to give the power response, but that's not really the case unless the boundaries have equal reflectivity at all freq.
The freq. non-linearity for any given axis imposes itself as an additional corresponding delay. Of course at the extremes (literally) there is usually a "slope" in response that creates most of the delay. ..Basically the inter-relationship of amplitude, phase, and time.
As far as mechanical resonances - yeah, I'd expect that off-axis you'll have certain problem areas that are altered (and perhaps lowered for a narrow "q").
If mass gets really low for a driver's Sd with insufficient damping for mechanical resonance, then the resonances will actually shift with a complex signal at higher spl's. (..this results in substantial loss of subjective clarity.) Whizer's are particularly problematic (..with no edge damping.) This leads to the classic statement of: "the music just falls apart at higher levels". ..pretty much the opposite of a good waveguide and compression driver combo.