I did dap into compression drivers more the last years, and noticed a big difference in sound. The high rated 18Sound compression drivers sound very harsh to me while the Faital compression drivers sound soft and smooth. I can't pinpoint why that is. Mostly it's visible in specs and measurements why a driver sounds like this and that, but with compression drivers i seems to miss that factor.
I did build some systems with compression drivers the last years, not for me but for others. I did one with the NSD1095N in a XT1086 horn, and altough the client liked it, it sounded very harsh to me. I did mount a Faital HF 108R to the same horn and it sounded a lot better after i eq'd it (like i did with the ND1095ND). Both have little resonances in the top, both are fit for the cutoff frequency in that horn (crossover with minidsp 48dB/oct at 1800Hz) but still such a big difference. Amps are Ncore amps.
At the same time i did also build a higher power garden system (see pricture) with the Big Faital HF201 in a P-sound clone of the JBL 2380A horn and a Beyma 12BR70, and it got the same soft easy going sound as the HF108R. This one is crossed at 900Hz 48dB/Oct with a DBX PA2 (is for small gigs) dsp and crown amps. I did try a Beyma CP750Ti on it that i lended from a friend, but again it sounded relative harsh (not like the ND1095N altough).
Can someone explain me how this happens, what factors make a compression driver sound harsh or smooth. I know Faital makes compression drivers that I like a lot, but i would like to find more, without having to buy them first to test. My next project needs one from 1kHz to 20kHz (or at least high enough to cover all), and i could use the 108R in that XT1086 horn (works for hifi, i did test), crossing to a 10 woofer (for home hifi, not pa) but having other options (and know how to recognise them) would be nice...
I did build some systems with compression drivers the last years, not for me but for others. I did one with the NSD1095N in a XT1086 horn, and altough the client liked it, it sounded very harsh to me. I did mount a Faital HF 108R to the same horn and it sounded a lot better after i eq'd it (like i did with the ND1095ND). Both have little resonances in the top, both are fit for the cutoff frequency in that horn (crossover with minidsp 48dB/oct at 1800Hz) but still such a big difference. Amps are Ncore amps.
At the same time i did also build a higher power garden system (see pricture) with the Big Faital HF201 in a P-sound clone of the JBL 2380A horn and a Beyma 12BR70, and it got the same soft easy going sound as the HF108R. This one is crossed at 900Hz 48dB/Oct with a DBX PA2 (is for small gigs) dsp and crown amps. I did try a Beyma CP750Ti on it that i lended from a friend, but again it sounded relative harsh (not like the ND1095N altough).
Can someone explain me how this happens, what factors make a compression driver sound harsh or smooth. I know Faital makes compression drivers that I like a lot, but i would like to find more, without having to buy them first to test. My next project needs one from 1kHz to 20kHz (or at least high enough to cover all), and i could use the 108R in that XT1086 horn (works for hifi, i did test), crossing to a 10 woofer (for home hifi, not pa) but having other options (and know how to recognise them) would be nice...
i have the HF108 ( non R) and had the 18sounds NSD1095N Before and i found that dpends a lot with the horn used . also the NSD1095 needs more correction as it falls faster than the HF108 past 10K . used as tweeter from 5K with a small JMLC 2K horn i prefer the Faital HF108 , used with a ARAI480 Horn i prefer the NSD1095 as it can go lower with less distorsions.
i would also like to try the new cheap Beyma CD151 anular drivers as tweeter don't know wich model to chose between the polyester and the polymer but at low power with home use don't know if it will make any differences.
i would also like to try the new cheap Beyma CD151 anular drivers as tweeter don't know wich model to chose between the polyester and the polymer but at low power with home use don't know if it will make any differences.
Because frequency sweeps, tone burst and noise derived measurements only provide a response profile which is NOTHING like how a transducer will perform with complex musical content. What you’re hearing is both the diaphragm material timbre differences and the exit angle differences between the CDs and the different resonances they create within the horn throat and mouth.
I walked away from magical measurements club decades ago…..it was easy to expose the nonsense when you work with microphones…..same principles in reverse. A good engineer at least worth his weight in salt will try quite a few different mics before settling on the one to use based primarily on the instrument’s voice. The purpose here is to capture the timbre and the subtlety of dynamics as they are or intended. Secondary functions such as equalization which harm that capture can now be avoided…..the phase field of the tones in and around the mics field of view are not altered by the EQ process. Two mics can have near identical frequency responses, polar patterns and dynamic ranges and yet sound VASTLY different when compared using the same source.
Take sibilance for example…….a word lately so often misused by the sighted listening and measuring cabal on YouTube that it hurts my soul every time they speak it. If you’re hearing sibilance from your speaker, it’s because it’s baked into the recording and nothing else….. Your speakers response is highlighting that sibilance, NOT creating it…..nothing left to do but cut it out with EQ and kill the surrounding dynamics in the process. On the subjective end of evaluating a speaker with whatever $h!t recording this is?…….sighted listening screw up 101. Bottom line is there’s a talent and skill factor in all of this that make the great engineers who they are…….not unlike the legendary speaker systems that enjoy so much praise and yet measure so poorly.
Music is non linear analog and organic like us…..sighted evaluation won’t even scratch the surface of the complexity of a the performance which comprises the recording, the speaker, the acoustic environment, the source electronics and the diversity of the individual listener.
I walked away from magical measurements club decades ago…..it was easy to expose the nonsense when you work with microphones…..same principles in reverse. A good engineer at least worth his weight in salt will try quite a few different mics before settling on the one to use based primarily on the instrument’s voice. The purpose here is to capture the timbre and the subtlety of dynamics as they are or intended. Secondary functions such as equalization which harm that capture can now be avoided…..the phase field of the tones in and around the mics field of view are not altered by the EQ process. Two mics can have near identical frequency responses, polar patterns and dynamic ranges and yet sound VASTLY different when compared using the same source.
Take sibilance for example…….a word lately so often misused by the sighted listening and measuring cabal on YouTube that it hurts my soul every time they speak it. If you’re hearing sibilance from your speaker, it’s because it’s baked into the recording and nothing else….. Your speakers response is highlighting that sibilance, NOT creating it…..nothing left to do but cut it out with EQ and kill the surrounding dynamics in the process. On the subjective end of evaluating a speaker with whatever $h!t recording this is?…….sighted listening screw up 101. Bottom line is there’s a talent and skill factor in all of this that make the great engineers who they are…….not unlike the legendary speaker systems that enjoy so much praise and yet measure so poorly.
Music is non linear analog and organic like us…..sighted evaluation won’t even scratch the surface of the complexity of a the performance which comprises the recording, the speaker, the acoustic environment, the source electronics and the diversity of the individual listener.
I've found that driver diaphragm composition has a big influence on the sound. IME, titanium diaphragms almost always sound harsh, while phenolic and polyamide mids and mylar tweeters tend to have a smoother, less fatiguing tone. Horn resonances can also play a factor in the harshness, as can eq or lack thereof. Finally, crossover frequency can influence the sound, many drivers lose directivity at the low end of their range, which affects phase response and can beam at the upper end, both of which can create a sound that is hard to pinpoint but is ultimately harder to listen to over time.
The HF-201 has a titatnium dome diagram, but sounds smooth also. I don't think the material of the diagram is the only factor. Berillium is also very hard but does not sound always very harsh. Some of the Radian BE drivers still sound reasonable smooth, not like a Faital, but still not harsh. But they are expensive and quality control is not always as it should be I heared first hand (and seen myself).
But it's probally true that hard materials for diagrams have higher resonances and so IMD. But just like with cone speakers some manage to keep them under controll in the passband.
But it's probally true that hard materials for diagrams have higher resonances and so IMD. But just like with cone speakers some manage to keep them under controll in the passband.
Because some aspect is missed being measured or we cant fully interpret the current set measurements,...
//
//
You may find it easier to avoid harshness by limiting the frequency range covered by a horn and driver. Crossing over at about 6K to a good horn tweeter can avoid some problems if you are crossing your midrange horn around 1K. Many manufacturers will want to sell you a horn and driver that can cover 800 or so hz to 20K. It will just be hard to really enjoy the sound from them.
My observations, mostly from more than a decade ago when I built a handful of CD based systems:
- ones man's "harsh" is another man's "detailed", and one man's "soft and smooth" is another man's "dull and boring", also I found a large day-to-day variance in what I felt it sounded like
- sometimes the harshness seemed to literally explode at higher levels, which might indicate HOM problems (as per Dr.Geddes), or just my overloaded ears on a bad day?
- drivers with smaller diaphragms and lower compression ratios tended to sound a bit better to me (with 3" or even 4" domes it's almost impossible that they don't break up just above 10kHz), but that could (partly) be expectation bias
- different amps also seemed to make a difference, at the time I tried to use Hyped UcD but found classic linear AB amps did sound better
- manufacturer's measurements were generally useless to estimate sound character
- I never came to try Faital's (or other annular ring radiator types, or types with non-metallic diaphragms) which might show much larger differences.
- OTOH, when very precisely EQ'd to the same target (to better than 0.1dB if possible) with a DRC/EQ toolbox like (((Acourate))), different drivers on the same horn started to sound almost the same, or at least the obvious differences mostly went away (I did that only once, so no statistical relevance). Note that a few parametric EQs might not get you there.
That last thing is what makes me think that CD+Horn sound specifics is simply frequency response for the most part. Nonlinear distortion, as well as HOMs and breakup play a role, but aren't dominant (except for the occasional corner case and very elevated levels, probably).
The interface of driver and horn by far make the most significant difference in character of sound.
I personally really like the NSD1095N on smaller horns with a higher xover. The treble (once leveled and contoured to suit the neighboring driver) has the most neutral and analytical quality, very close to a good dome tweeter. The plastic diaphragm drivers sound more veiled and become very glazed over sounding at high SPLs, but they're far more forgiving to less careful EQing.
If you listen to brass instruments alot and want the immediacy in dynamics with the harmonics and less veiled top end, a well designed metal diaphragm driver on a decent horn will sound more accurate, but it depends on how severe its breakup mode is. The breakup mode can really ruin the sound, especially at high volume levels. Suppressing it makes a huge difference as does whether the breakup mode falls in the audible spectrum. Even if BU is inaudible by itself, it can down modulate into the lower frequencies, increasing IMD.
Phase plugs also can perform differently on various horns. Its not only the driver exit angle which can be incompatible but also the phase plug design with the geometry of the horn throat. Impedance reflections inside the horn can severely ruin the clarity and raise distortion levels. Some drivers rely on high Q lower end rolloff slopes to enhance performance at lower cutoff. These can sound mouthy and shouty when the driver HP is pushed closer to the horn's lower unloading point, especially when the diaphragm loses control due to insufficient loading / reactance. EQing will help this but it won't totally get rid of it.
I personally really like the NSD1095N on smaller horns with a higher xover. The treble (once leveled and contoured to suit the neighboring driver) has the most neutral and analytical quality, very close to a good dome tweeter. The plastic diaphragm drivers sound more veiled and become very glazed over sounding at high SPLs, but they're far more forgiving to less careful EQing.
If you listen to brass instruments alot and want the immediacy in dynamics with the harmonics and less veiled top end, a well designed metal diaphragm driver on a decent horn will sound more accurate, but it depends on how severe its breakup mode is. The breakup mode can really ruin the sound, especially at high volume levels. Suppressing it makes a huge difference as does whether the breakup mode falls in the audible spectrum. Even if BU is inaudible by itself, it can down modulate into the lower frequencies, increasing IMD.
Phase plugs also can perform differently on various horns. Its not only the driver exit angle which can be incompatible but also the phase plug design with the geometry of the horn throat. Impedance reflections inside the horn can severely ruin the clarity and raise distortion levels. Some drivers rely on high Q lower end rolloff slopes to enhance performance at lower cutoff. These can sound mouthy and shouty when the driver HP is pushed closer to the horn's lower unloading point, especially when the diaphragm loses control due to insufficient loading / reactance. EQing will help this but it won't totally get rid of it.
Manufacturers datasheets never show the real behaviour of these resonances. Did you measure yourself with high resolution, how much is the difference?
At 1,8kHz crossover you should try DE360 - I use it in my living room cinema and it's far from any harshness.
Or maybe FaitalPRO HF107 which looks great on paper but I don't trust Faital measurements. But has the potential to be crossed lower and impedance looks clean up to 20kHz.
I would take a hard look at how the driver handles the secondary resonance and the controlled break-up in the surrounds used in many all metal diaphragms. This is used to fill in the last octave or so.
Another issue is how the drivers are used. In a passive system you attenuate the mid-band. In an SR system you don't want to waste SPL capability so you boost the last octaves. This amplifies any issues with break-up modes and any issues with the secondary resonance making them more audible.
Rob 🙂
Another issue is how the drivers are used. In a passive system you attenuate the mid-band. In an SR system you don't want to waste SPL capability so you boost the last octaves. This amplifies any issues with break-up modes and any issues with the secondary resonance making them more audible.
Rob 🙂
This is a very un-researched area which deserves alot more attention, but it gets overlooked as the most considered issues which manufacturers concentrate on are FR linearity, basic HD and coverage uniformity.
A poor choice of horn + driver combination is one of the primary causes of odd order HD IMO. High compression ratios coupled with excessive SPL, without notching/EQing out the breakup range frequencies, will also generate high levels of odd order HD across the entire midrange frequencies.
Excess HF diaphragm breakup feeds into this with resulting IMD being down-modulated. It can become extreme, producing various combinations of undefined multi mode distortion, essentially resulting in the generation of noise, which is no longer harmonically associated with the music being reproduced. We used to call this phenomenon "diaphragm scream" back in my PA sound days.
The various trigger levels for these severe modes of breakup to occur aren't linear or gradual and therefor can't be easily EQed or notched out.
Larger diaphragms do better, playing cleaner at higher SPLs, but these can produce some of the harshest breakup modes, especially in the audible range.
The larger available diaphragm surface area enables a wider range of breakup modes, which can add together higher in amplitude and be triggered more abruptly than those of smaller diaphragms.
The biggest variable in generating uncontrolled severe midrange breakup are the combinations of modes occurring at the same time at very close frequencies ie. multiple modes of phase/time shifted breakup taking place in different areas of the diaphragm. This is all of the available diaphragm surface area going out of control simultaneously in unsynchronised, unrelated fashion. You can have radial, axial and tangential modes all at once without any harmonic relationship if the diaphragm is large enough. It can be so severe that the diaphragm ruptures or fractures, especially undampened aluminum or beryllium diaphragms. Titanium is very tough and doesn't usually fail mechanically, even when the diaphragm bottoms out hard on the phase plug.
There is one thing which can significantly reduce the chance of a diaphragm going severely out of control due to multi mode resonance and especially with HF breakup. The use of dampening material on the rear chamber walls usually helps significantly, as does spraying the diaphragm with a light, even coating of PTFE dry film. I've seen Aquaplas coating being used successfully on the larger JBL 4" and 3" diaphragms, but this is very hard to apply in an even fashion using a spray gun.
Because its very difficult to find a good WG and CD combo which will also behave somewhat predictably at all sorts of levels, scenarios and program material, I take great care to put horn and driver combinations in completed speakers to great lengths of final listening tests.
The listening is done after all the design work is complete
using simulation and measurement. It takes a good amount of time with attentive (multiple) sets of ears to determine whether you got it right. If there are even the smallest, odd sounding artifacts or any objectionable harshness, the whole process usually starts over again, sometimes with different combinations of parts. Sometimes is just a matter of optimizing the filters. Sometimes the parts just don't work together harmoniously. You can't always force things just because they should work together on paper.
High efficiency drivers ie. CDs, WGs, are very difficult to use for all out high end hifi reproduction. They generate alot of their own (mis)behavior and are therfore difficult to tame without taking all the delicacy and liveliness out of them. The people who are talented at getting the right desired character and performance from these have very good hearing perception and talent.
The use of computers and measurement is essential at various stages of development of a successful horn based system, but it has to pass the final difficult test, which is the music playback test. It doesn't matter if it measures excellent on the best gear but sounds wrong to well educated ears.
Thats often overlooked when implementing the HF driver and trying to "hammer" the top octave into shape with EQ. I've found that boosting top end definitely aggravates some of the smaller abnormalities in the HF which are more bearable with passive shelf EQ using a bypass cap and resistor parallel, connected in series with the driver. It depends where in series those shelf EQ elements are placed and how much level reduction takes place up top. With alot of series resistance, the driver becomes more dominant in defining its overall FR, as the VC inductance takes over. This will also reduce any dampening from tighter driver coupling (from the lower insertion resistance) to the amplifier.
Many times, just like with soft dome tweeters, the upper end of the HF is supplementally carried by gradual diaphragm breakup. This is only possible due to the ear being less sensitive to higher frequency HD compared to the midrange. Too much of this breakup used as a crutch will make the top end sizzle and sound fuzzy, with an artificially wide stereo image.
Once you go over 5k or so, ears can't really discerne harmonics that well associated with these higher fundamental frequencies. Because of this, distortion becomes less dominant in the perceived audio quality of an HF driver above 8k or so (half of your upper hearing limit cutoff). Anything above 10k will be perceived as "air" and fluff. FR linearity will also be less important, being harder to decipher by the brain. The main thing critical here is the CSD, but not so much as in the midrange.
Therfor a decent CD and horn combo needs to perform well up to 10k in the most important areas, not so much above that point. The addition of a super tweeter is far less critical in the mid to far field listening area.
To my ears, larger diaphragm CDs sound better than smaller ones, mainly due to the lower distortion and cleaner performance in the lower mids, especially at higher levels, where smaller drivers tend to struggle alot.
The artificial help designed into most smaller drivers and horns used to extend the lower mid cutoff point hurts more than it helps. IOW quantity over quality is chosen by many manufacturers. Once you've heard a really good large format CD/WG combo, you'll have a hard time going back to the less capable smaller drivers and horns.
Thank you, that is usefull info. Those that i like all have well damped resonances, or easy to controll resonances. And mismatch of horns is also true, but you would espect that 18Sound match their drivers to their horns, and still the Faital 108 sounded better on the XT1086 compared the the NSD1095 that i tried also recently. When comparing them i crossed higher (2kHz), as the 1095 does not go that low as a 108, i'm looking to cross a 1" fairly low (1K to 1.3K) for the moment.
I guess you're right on the resonances thing, it will surely something i'll look at more if i got the right data...
I guess you're right on the resonances thing, it will surely something i'll look at more if i got the right data...
I guess the only thing left after all those factoid opions is to get your hands moving and make something.you will likely never get it right the first time.thats the beauty of diy.you can evolve your creation to suit what your ears like.nobody can tell you what sound you like.you have to listen. And thrn adjust and adjust again.when you get the sound right you will know it.
The HF108 definitely goes lower on the XT1086 with less drama than the NSD1095N. If you cross higher towards 2k+, the NSD1095N wins in terms of clarity and definition. I prefer the smaller WGs with the NSD1095N mainly because the top end is this driver's strength. It doesn't have much output as the HF108 up top, but what's there is alot better. You just have to use a small -5 dB notch at 18k to remove some sizzle if you're trying to get the best from the combo - below are some measurements from another member of the two together.
Another option is the Eminence WG10 horn (not to be confused with the line source WG they sell under the same part number) which is also a shallow horn like the XT1086 without as much diffraction. Its more like the revered QSC horn people love (for good reasons) plus it also sounds very good with HF108, loading down to about 700 hz. You'd obviously want to cross closer to 900 hz for enough overlapping response to get the phase transition right. The HF108 and WG10 combo are really smooth. They're perfect for use with a 8" - 12" midbass, ie. B&C 12PE32 or 8PE21. It would also work good in an MTM with 2 of the 8NDL51.
Another option is the Eminence WG10 horn (not to be confused with the line source WG they sell under the same part number) which is also a shallow horn like the XT1086 without as much diffraction. Its more like the revered QSC horn people love (for good reasons) plus it also sounds very good with HF108, loading down to about 700 hz. You'd obviously want to cross closer to 900 hz for enough overlapping response to get the phase transition right. The HF108 and WG10 combo are really smooth. They're perfect for use with a 8" - 12" midbass, ie. B&C 12PE32 or 8PE21. It would also work good in an MTM with 2 of the 8NDL51.
Attachments
With a passive network It depends on the driver and horn/waveguide response without the high pass attenuation and EQ. You are assuming that all "eq" is in the form of a series shelf where in many cases its much more nuanced.
With typical CD horn's/waveguides when you do a passive comp there is little to no attenuation in the upper octaves, maybe some shaping.
Rob 🙂
I have used NSD1095N now for a couple of years and tweaked and have had various success with nice results. But there was always something in this system that was not quite right - a bit gritty, sibilance, hardness... I have heard the same reports from others here and some have given up on the driver.
Yesterday I hooked up a USB isolator between my Mac mini computer and the DAC (slightly tweaked smsl SU-1) and for the first time I hear no ill will, no hardness, no sibilance - just a very well behaved upper range - zero stress on ear all of a sudden - wonderful. (Xo@1,6kLR4)
Perhaps most of the negative experience of this driver has to do with the system in front of it? I'm amazed over what comes out of it now. Really...!
//
Yesterday I hooked up a USB isolator between my Mac mini computer and the DAC (slightly tweaked smsl SU-1) and for the first time I hear no ill will, no hardness, no sibilance - just a very well behaved upper range - zero stress on ear all of a sudden - wonderful. (Xo@1,6kLR4)
Perhaps most of the negative experience of this driver has to do with the system in front of it? I'm amazed over what comes out of it now. Really...!
//
