No problem with that - but the converse is that mass-market manufacturers have a vested interest in sliding substandard systems under the door. In the early days of digital, Sony was trying to prove that 15 kHz (and a 32 kHz sampling rate) was good enough for anyone - so they could rush to market the primitive systems they had at the time. As it was, there was a huge wrangle between the AES and Sony/Philips when the standards for CD were being discussed - the AES was holding out for 48 kHz sampling as a rock-bottom minimum with a strong industry preference for 60 kHz. Sony/Philips flatly told the AES committee that they had no power over Sony/Philips and they were going to introduce 44.1/16 whether liked it or not. In retaliation, the AES recommended 48/16 as a minimum professional standard, which is part of the reason we have the 44.1/48 divide today. It was all about politics and money, not technology. The consensus opinion was to wait a few years until 60 kHz was ready, and standardize on that.
Sony and Philips needed the return on investment as fast as possible, blanketed the US market with "Perfect Sound Forever" flackery, doubled the price of a record, and drove LPs off of the US market in 3 years. They won, and the AES (and the rest of us) lost. The 44.1/16 had no expansion capability built in, so unlike the steady progress of LP's over the decades, CD's were frozen into VHS quality with no upward path. In a similar way, Dolby Digital is hard-coded into the American HDTV system and DVD's. Nice if you're a Sony, Philips, or Dolby stockholder, maybe not so nice for the rest of us.
That was a very different process than all the parties sitting down at a table in Los Angeles in 1957 and deciding on a stereo-LP standard. The Westrex 45-45 system (going back to Blumlein) was obviously superior to the other alternatives (vertical/lateral, the Minter ultrasonic carrier, and a horrible steered-mono proposal), and it was chosen for the entire industry. The best system prevailed. But the industry was much smaller then, and engineers still had a voice. Nowadays engineers are commanded by the marketing department, so the overall quality of audio engineering has steadily declined. Most of the focus is doing more with less, as in the various lossy-compression schemes which are purported as being "CD quality" - which was badly compromised at the outset. This has reached a reductio ad absurdem with so-called "HD Radio", where a lossy-compression digital signal with a remarkable data rate of 96 ksamples/sec is being marketed as "high definition FM".
I'm all in favor of using psychological studies, but commercial interests can easily sway the results - and always in the same direction, things that are cheaper to build and faster to get to the market. The larger point is that as the industry gets larger and more financially concentrated, big marketing campaigns overshadow legitimate technical concerns - which gets us back (finally!) to sibilance. There are many reasons - mostly concerned with engineering oversights and omissions - for sibilance. Most of them go unknown for the reasons mentioned above - that XYZ technology is "good enough" and should go on the market as-is, which maximizes ROI, improves share price, and enhances the income of the board of directors and the CEO. In business terms, a "win-win" solution.
Sony and Philips needed the return on investment as fast as possible, blanketed the US market with "Perfect Sound Forever" flackery, doubled the price of a record, and drove LPs off of the US market in 3 years. They won, and the AES (and the rest of us) lost. The 44.1/16 had no expansion capability built in, so unlike the steady progress of LP's over the decades, CD's were frozen into VHS quality with no upward path. In a similar way, Dolby Digital is hard-coded into the American HDTV system and DVD's. Nice if you're a Sony, Philips, or Dolby stockholder, maybe not so nice for the rest of us.
That was a very different process than all the parties sitting down at a table in Los Angeles in 1957 and deciding on a stereo-LP standard. The Westrex 45-45 system (going back to Blumlein) was obviously superior to the other alternatives (vertical/lateral, the Minter ultrasonic carrier, and a horrible steered-mono proposal), and it was chosen for the entire industry. The best system prevailed. But the industry was much smaller then, and engineers still had a voice. Nowadays engineers are commanded by the marketing department, so the overall quality of audio engineering has steadily declined. Most of the focus is doing more with less, as in the various lossy-compression schemes which are purported as being "CD quality" - which was badly compromised at the outset. This has reached a reductio ad absurdem with so-called "HD Radio", where a lossy-compression digital signal with a remarkable data rate of 96 ksamples/sec is being marketed as "high definition FM".
I'm all in favor of using psychological studies, but commercial interests can easily sway the results - and always in the same direction, things that are cheaper to build and faster to get to the market. The larger point is that as the industry gets larger and more financially concentrated, big marketing campaigns overshadow legitimate technical concerns - which gets us back (finally!) to sibilance. There are many reasons - mostly concerned with engineering oversights and omissions - for sibilance. Most of them go unknown for the reasons mentioned above - that XYZ technology is "good enough" and should go on the market as-is, which maximizes ROI, improves share price, and enhances the income of the board of directors and the CEO. In business terms, a "win-win" solution.
Last edited:
To the moderators - sorry about what nearly devolved into a rant, but there is a political and economic dimension to the history of the audio industry. Going back at least as far as the history of FM radio and NTSC color television, large manufacturers have used politics, friendly magazine reviewers, and nationwide marketing campaigns to push technological deficiencies aside. To omit this dimension makes much of the history of electronics and mass media unintelligible - to understand television, you have to know the rivalry between Farnsworth and Sarnoff, as well as the unsung contribution from Blumlein at EMI in England. Similarly, for broadcasting, what happened to one of the greatest engineers of all time, Major Armstrong, is important. The political and economic dimension shaped entire industries, and is an important part of the narrative. These wrangles continue to the present day, but are usually only visible to industry insiders (who are frequently bound by non-disclosure agreements, and are not free to publish).
Last edited:
Well, I was using a dome tweeter as a common example of a driver where cone breakup is significant in relation to the discussion of sibilance. I never said it was the only source of distortion, and I'm actually more concerned about the spectral decay than the non-linear distortion, although cone breakup introduces some of both.
With the exception of true ribbons or planar drivers, most drivers (including the compression drivers in horns) tend to experience cone breakup to some degree whether in band or just outside their passband since only part of the diaphragm, (either the middle or the edge) is driven. Horns are at a bit of an advantage over a direct radiator since a much smaller diaphragm can produce the necessary SPL's at a given frequency - so cone breakup onset is pushed higher in frequency. They're not immune to it though.
Well, the entire diaphragm is definitely driven in phase - at no frequency is one part of the diaphragm moving forwards when another part is moving backwards, which is what happens in cone breakup.
As for how symmetrical the drive is along the length of the ribbon - that depends entirely on the design. On my Aurum Cantus G2 they combat that by making the total ribbon significantly longer than the exposed portion. The length of foil in the magnetic gap between the top and bottom clamps is 70mm, but the aperture in the front plate is only 50mm high, so the 10mm at each end of the foil which may not be moving quite as much is not contributing to radiation.
As well as that the entire foil is corrugated so that it has a little bit of give at the ends to allow the visible part of the foil to move freely without being constrained by the ends. (Let's not forget that the lateral excursion is minuscule compared to the length of the ribbon, so it doesn't take much "give" in the corrugations to allow the visible portion of the foil to move more or less freely)
As for measuring it to see where most of the radiation is coming from, at very near close mic'ed measurement distances the vertical foil will act as a line source, and if you measure at one end of a line source you'll get less output even if the radiation from the line source is uniform right the way along, so that doesn't really tell you anything. To determine if the excursion is equal along its length you'd need to measure it optically.
The Decca ribbon is not even remotely the same design as the G2. For one thing the Decca is horn loaded with a moderately large horn, while the G2 uses a small waveguide that's only about 15 mm deep.
Compare:
Ribbon Tweeter Horn Loudspeaker: Decca London/Stanley Kelly - Efficiency, Size, Filter and Frequency Response.
Aurum cantus
You're completely misunderstanding the differences between horns and waveguides. Technically they are both similar devices, but as far as what parameters are optimized they're at the extreme opposite ends of the spectrum.
A horn's primary purpose is to act as an impedance transformer increasing efficiency and net power response - directivity is something of a side effect. A horn typically has a gradual progressive expansion profile. The long throat and gradual curvature tends to cause beaming of the high frequencies since the driver diaphragm is obscured from view past a certain angle, and only the lower frequencies will bend to follow the curvature of the mouth. The "horn loading" only exists down to a certain frequency depending on the length of the horn - to get it to work down to 2Khz it would need to be something of the order of 50-100mm long.
The Decca ribbon is an an example of a horn.
A waveguide is the opposite extreme - it's primary design purpose is to control the directivity not increase efficiency. There is still an increase of on axis SPL and therefore "sensitivity" but it's due almost entirely to increasing the directivity index not "horn loading", in fact in a constant directivity design you're trying to avoid horn loading.
A constant directivity waveguide has essentially a conical shape with a very short radius transition from the driver to the final angle - typically 45 degrees for a 90 degree CD waveguide. There is then usually a small radius at the mouth to minimize diffraction from the edge, but over 90% of the throat is straight, and due to the straight taper the driver diaphragm is visible right out to the design angle, which eliminates beaming at high frequencies, hence constant directivity.
Because you're not relying on horn loading, a CD waveguide doesn't have to be very deep to function...
The Aurum Cantus G2 (and a lot of other modern ribbons) are not horns because they're nowhere near deep enough for the frequency range they're covering - the foil is only about 15-20mm behind the front of the face plate on the G2, and if it had a horn taper it would only be effective down to about 8Khz if you were lucky.
In fact it measures almost flat right down to about 1.5Khz, because it's acting (in the horizontal plane) as a CD waveguide with no horn loading. The horizontal off axis response is almost dead flat out to about 40 degrees then then entire range starts dropping off quickly - a CD characteristic. There is no significant horizontal off axis beaming of the high end - it stays steady from below 2Khz to 10Khz at about 90 degrees, then smoothly drops to about 60 degrees by 20Khz. Directivity index stays between about 2dB and 3dB across the entire operating range.
In short, a constant directivity waveguide has completely different characteristics to a horn, so don't confuse the two. There's a great article here that's well worth a read and explains horns vs waveguides:
http://www.pispeakers.com/Pi_Speakers_Info.pdf
Completely different driver for a different application, I'm not even sure why you're trying to compare them. A 50 mm aluminium foil waveguide ribbon is not trying to produce PA level SPL's...we're talking about a ribbon with a similar size front plate to a dome tweeter, maybe twice as deep, and in the same ballpark price wise.
Show me ANY dome tweeter that can produce 128dB SPL, and you might have a point to make
The SA ribbons are horn loaded, so will naturally have a higher maximum SPL, (but will suffter somewhat from high end beaming and non-flat response) and a kapton former diaphragm will have far greater power handling, but wont have nearly as good CSD as an aluminium foil. It will play loud, but it won't sound nearly as good at home hi-fi volume levels.Seriously though, who needs 128dB SPL capable tweeters in their home system ? Most music has an average treble spectral content that is a full 10-20dB lower than the midrange and bass, as well as most high SPL treble being of a very transient nature. (Hi-hat hits etc) A tweeter that can produce a continuous 102dB SPL without strain and transients up to around 112dB SPL is more than enough for even a large home system - and I like loud music...
So basically you've compared a PA driver with a home hi-fi driver...which is a bit silly. If I want something that will produce PA level volume I'm not going to be using a small foil ribbon - I'll be using a horn of some sort. (Or maybe a long line array of larger ribbons)
That depends whether you subscribe to the "wider the dispersion the better" philosophy. It seems to be almost taken as a given that the widest possible dispersion in the treble is something to be strived for and is somehow the pinnacle of reproduction. In the last few years an appreciation of the merits of controlled directivity, (especially constant directivity) is emerging so I think things are starting to tilt back in the other direction.
When I first started using ribbons I used them despite knowing that their vertical dispersion is considerably narrower than a dome, accepting it as one of the tradeoff's that I would have to make for their other desirable characteristics. It's taken me a while to realise, but what I thought was a disadvantage actually turns out to be an advantage, in so many ways...
You assume I haven't heard them outdoors ? What exactly do you expect me to hear ?I disagree.
Take ur ribbon outside, where the room reflections will not color your opinion and have a listen. Let me know what you hear?
True, but if I offer into the original discussion that a ribbon tweeter can "solve" the sibilance issue that many drivers have, and you say "yes but a ribbon has so many practical drawbacks", whether that is actually the case is vaguely relevant to the discussion.
I'm using a full range driver which goes to 15Khz but is pretty flat out to about 8Khz.
Even with 18dB/oct crossovers I still need a full octave overlap on both sides. (I might revisit a 24dB/oct crossover one day, I wasn't happy with one the first time around)
I'm fully in the "cross the tweeter over high and steep" camp, as I mentioned in an earlier post responding to Lynn. It's harder to find a driver that will go that high and do it well, but if you do I think there are a lot of benefits to be had, one among them being the IM distortion, power handling and SPL requirements of the tweeter are drastically reduced, and honestly I just think it sounds better with the crossover frequency above the presence region instead of right in the middle of it...
Last edited:
Very informative stuff Lynn...Question, what's your take on odd order xovers..?
regards,
I think I made my point earlier, if we can't get on the same page, so be it.
The question is degree - better drivers or ribbons will tend to be driven "in phase" across their surface.
So the energy behind and above and below the little hole that the ribbon peeks out of magically goes away with no effect whatsoever? Just checking on this.
you can see the effect by using a longer ribbon - obviously on a short short ribbon it tends to act more like an ideal ribbon...
Not all DECCA ribbons used a horn.
The effect of the horn was in the main on the LF end of the ribbon's output.
You're completely misunderstanding the differences between horns and waveguides. Technically they are both similar devices, but as far as what parameters are optimized they're at the extreme opposite ends of the spectrum.
A horn's primary purpose is to act as an impedance transformer increasing efficiency and net power response - directivity is something of a side effect. A horn typically has a gradual progressive expansion profile. The long throat and gradual curvature tends to cause beaming of the high frequencies since the driver diaphragm is obscured from view past a certain angle, and only the lower frequencies will bend to follow the curvature of the mouth. The "horn loading" only exists down to a certain frequency depending on the length of the horn - to get it to work down to 2Khz it would need to be something of the order of 50-100mm long.
The Decca ribbon is an an example of a horn.[/quote]
No the Decca ribbon is an example of a ribbon driver.
The horn was added, and is/was optional.
I'm talking about the behavior of the ribbon, without regard to the horn.
Let's not argue this here, it's the wrong thread.
The Aurum Cantus G2 (and a lot of other modern ribbons) are not horns because they're nowhere near deep enough for the frequency range they're covering - the foil is only about 15-20mm behind the front of the face plate on the G2, and if it had a horn taper it would only be effective down to about 8Khz if you were lucky.
In fact it measures almost flat right down to about 1.5Khz, because it's acting (in the horizontal plane) as a CD waveguide with no horn loading.[/quote]
I tend to think it is doing next to nothing, and the polar response is entirely due to the shape and size of the ribbon WRT wavelength.
The horizontal off axis response is almost dead flat out to about 40 degrees then then entire range starts dropping off quickly - a CD characteristic. There is no significant horizontal off axis beaming of the high end - it stays steady from below 2Khz to 10Khz at about 90 degrees, then smoothly drops to about 60 degrees by 20Khz. Directivity index stays between about 2dB and 3dB across the entire operating range.
In short, a constant directivity waveguide has completely different characteristics to a horn, so don't confuse the two. There's a great article here that's well worth a read and explains horns vs waveguides:
http://www.pispeakers.com/Pi_Speakers_Info.pdf
Completely different driver for a different application, I'm not even sure why you're trying to compare them. A 50 mm aluminium foil waveguide ribbon is not trying to produce PA level SPL's...we're talking about a ribbon with a similar size front plate to a dome tweeter, maybe twice as deep, and in the same ballpark price wise.
Show me ANY dome tweeter that can produce 128dB SPL, and you might have a point to make
Unfortunately my points seem to have been missed?
Perhaps I was not clear.
First, the SA ribbon had an optional wavguide.
It was non-flat with the waveguide.
The CSD we can argue.
Actually you are completely wrong on the last point. It will sound way better due to vanishingly low distortion levels and zero compression on peaks.
Again, you miss the point.
You do not NEED to play it loud.
I do not.
I want LOWEST POSSIBLE DISTORTION.
Maximum linearity & headroom.
Highest possible sensitivity.
Look ma, I don't clip my amp(s) ever, you do.
So what I am saying is that my example was about the technology of a driver that exists with lower distortion by an order of magnitude or more than almost all other HF drivers, and a higher power handling + greater max SPL. So, used at normal "hi-fi" levels it reaches and matches the distortion levels of some amplifiers.
Why?
I bought them, and used them for a long time.
They can be used.
Anything out there can be used.
If the name of the game is to optimize the sound, and eliminate sibilance then how is this silly??
Not sure how this pertains...
When I first started using ribbons I used them despite knowing that their vertical dispersion is considerably narrower than a dome, accepting it as one of the tradeoff's that I would have to make for their other desirable characteristics. It's taken me a while to realise, but what I thought was a disadvantage actually turns out to be an advantage, in so many ways...
You assume I haven't heard them outdoors ? What exactly do you expect me to hear ?
Safe bet.
I expect that you would NOT hear something.
You would notice just how narrow your tweeters may be in polar response, once the reverberant field of your room is eliminated.
It's an easy test - and requires only the tweeter actually, and does not need to be a high quality source, just has to be outside and away from reflective surfaces. A grassy area is nice for this... dirt works...
Try it sometime. There is something to be learned.
I'll say that a ribbon can not in an of itself solve the sibilance issue.
It may offer an improvement IF the main source of sibilance is a driver with a distortion and/or breakup issues. But so can any "better" driver than the one before offer an improvement.
I mentioned the SA as an example of a driver that while flawed in the vertical polar response department offers a full order of magnitude better distortion spec than almost anything out there, your driver included.
It did not "eliminate" sibilance. It did offer improvements in that department, as things that were heard to be much more "obvious" became apparent once it was being used... the other sources of sibilance remained to be conquered.
In otherwords, sibilance does not have a single source or cause.
Anyhow, interesting discussion...
_-_-bear
PS. I have owned and used all of those Deccas that are shown on the page you linked!
Last edited:
Hmm ... kind of mixed. Back in my Audionics days, I used them along with 2nd-order, but my tastes and understanding have evolved since then. For example, the Ariel is electrically 2nd-order, but acoustically (and by measurement) 4th-order.
There is a certain kind of "rightness" to having the drivers phase-track each other through the crossover region. With a quadrature phase relationship (in 1st and 3rd-order crossovers), well, it sounds more phasey, which draws unwanted attention to the crossover region. Everyone goes on and on about polar patterns, but maybe because of all the time I spent developing quadraphonic decoders, I became very sensitive to "phasey" sensations, and can easily hear them from across the room - it just sounds unnatural, as if the sound can't decide where it's really coming from, unlike a single driver.
The illusion I want to create is a solid, locked-in impression of one driver that is constant as you wriggle in your seat, get up, and move around the room. Loud or soft, it always sounds like one driver, not two. The brute-force method is of course to actually use a single driver, but there are penalties to be paid in dynamics, HF extension, and possible big trouble from whizzer cones.
Oddly enough many coaxials sound surprisingly disjointed - not spatially, but sonically. These things are surprisingly hard to get right. The only ones I've heard that were even close are the classic large-diameter Alnico-magnet Tannoys. But I don't think even a vintage Tannoy, good as they are, can match a well-realized 2-way system with a simple vertical stack and careful attention paid to time and phase relations.
If the acoustical crossover is an inphase type (acoustical 2nd or 4th-order), and attention is paid to cabinet diffraction, the locked-in single-driver impression can be as strong as a physical single driver. You can walk right up to it and it still sounds like a single driver. At least my speakers turn out that way, and other folks seem to be able to do this as well. In a way, it mirrors the requirements for a good mono speaker - don't annoy the listener with spurious spatial impressions.
I don't doubt acoustical 3rd-order crossovers can be done well, but I haven't had as much luck with them - managing the spatial impression is more difficult, and small ripples in the response seem to stand out more. A possible reason for the subjective impression might be the inter-driver phase angle is already at 90 degrees, and small departures from the ideal curve spread it into the danger region of 120 degrees, where unpleasant out-of-phase impressions start.
By contrast, with an inphase crossover, a small departure from the ideal curve only results in a 30-degree phase spread, which doesn't sound all that different than zero degrees. When the phase spread goes from 90 to 120 degrees, though, you can actually start to hear odd flanging and acoustic-null sounds. I recognize them immediately as what they are, but to a listener that doesn't have a background in phase-shift networks and what they sound like, it sounds like a hard-to-pin-down coloration that doesn't have an obvious tonal center. Switching from music to wideband pink-noise gives the game away; pink-noise immediately reveals phasiness for what it is. With music the coloration comes and goes, and is very difficult to pin down. If the spectrum is sparse, you might not hear it all.
I suspect much of the audiophile longing for single-driver and 1st-order crossover loudspeakers is to get away from these weird colorations. But I've heard plenty of speakers with 1st-order crossovers that were quite incoherent and "sloppy" sounding in the crossover region, along with noticeable mismatch in driver coloration and character. My impression is that 1st-order speakers are very, very difficult to design, and put nearly impossible demands on the drivers. The only one that comes to mind as a complete success was the original Quad ESL57, which is a rather different animal than a multiway dynamic system.
Last edited:
Here's a charming - and illuminating - interview with Wilma Cozart Fine of Mercury Records. She was directly involved in the process of remastering for CD, but regrettably, was not part of the SACD remastering. Even more remarkably, the 3 channels from the original mastertapes were mixed-on-fly directly into the 2-channel ADC, in the same way they were mixed-on-the-fly into the Scully cutting lathe. From the interview:
RAB: I admire so much that you have mixed the three tracks manually towards two channels during the cutting of the record. I myself made audiovisuals and had to mix the tape with speech and the tape with music manually and synchronize it with the images. I had to start a fade-in at the right moment and finish the transition to the following image on preselected moments in the music, sometimes slowly and at times very quickly. If I made a mistake I had to start the whole program of 20 minutes all over again. I can imagine how intense your work must have been.
WCF: Yes, we never mixed the channels to a master tape, but always directly to the cutting lathe. A tape would introduce extra hiss and we did not want that.
RAB: As the transfer was in progress you could change the stereo-balance. Did you do that?
WCF: I had to prepare myself well and I had to note beforehand which passages needed small adjustments. Too much correction is not good. That would disturb the natural sound balance of the orchestra.
RAB: You had to mix the three channels once again for the CD-releases. How did you go about it?
WCF: First I listened to the LP's, many times, so my memory came back. You know that as a writer, that when you are absorbed by the writing process you cannot be reached by other people. It takes a lot of concentration. I listened of course to the sound balance and the stereo-image on the LP. I wanted that the CD did not differ from the LP.
RAB: What was the most difficult mix?
WCF: That was the Balalaika-recording. It was difficult to get the naturalness of the plucking of the strings right.
RAB: CD is a completely different medium. How do you perceive it? Do you think it has its restrictions?
WCF: Yes, my husband always said that the sampling rate was too low. But the CD's are closer to the masters, the original tapes, than the LP's.
RAB: I admire so much that you have mixed the three tracks manually towards two channels during the cutting of the record. I myself made audiovisuals and had to mix the tape with speech and the tape with music manually and synchronize it with the images. I had to start a fade-in at the right moment and finish the transition to the following image on preselected moments in the music, sometimes slowly and at times very quickly. If I made a mistake I had to start the whole program of 20 minutes all over again. I can imagine how intense your work must have been.
WCF: Yes, we never mixed the channels to a master tape, but always directly to the cutting lathe. A tape would introduce extra hiss and we did not want that.
RAB: As the transfer was in progress you could change the stereo-balance. Did you do that?
WCF: I had to prepare myself well and I had to note beforehand which passages needed small adjustments. Too much correction is not good. That would disturb the natural sound balance of the orchestra.
RAB: You had to mix the three channels once again for the CD-releases. How did you go about it?
WCF: First I listened to the LP's, many times, so my memory came back. You know that as a writer, that when you are absorbed by the writing process you cannot be reached by other people. It takes a lot of concentration. I listened of course to the sound balance and the stereo-image on the LP. I wanted that the CD did not differ from the LP.
RAB: What was the most difficult mix?
WCF: That was the Balalaika-recording. It was difficult to get the naturalness of the plucking of the strings right.
RAB: CD is a completely different medium. How do you perceive it? Do you think it has its restrictions?
WCF: Yes, my husband always said that the sampling rate was too low. But the CD's are closer to the masters, the original tapes, than the LP's.
Last edited:
grumble grumble I got inspired and I played around with the simulator last night and got a nice 3rd order acoustic slope on my tweeter rolloff It integrated quite well with the midbass (the butterworth hump somewhat filling the hole in the midbass response curve) too.... I suspect it would be an improvement, but back to the drawing board
I wonder if Joe Dappolitos change to 4th order (from the original 3rd order he used for MTM) was for similar reasons....
Tony.
Last edited:
Agree very much with this ! In the past i have been able to achieve that one driver sound with odd order xovers...
Agree, i get the same impression when listening to them ...
Again agree with most of that Lynn, except I have found speakers that get there 1st order xover right, in regards to the midrange driver(s) tend to be better at achieving a certain "rightness" with regards to integration and sound than there 2nd and or 4th order counterparts .......
There is an Sluggishness of sound IMO with 4th order xovers, which is very apparent when you listen to or compare them to a 1st order system, similar i guess to when you listen to FR drivers.
I guess an FR is a bit too much of nothing and a 4th too much of.......
Regards,
Last edited:
Yikes, considering we're getting well off topic in this thread, there is a ton of stuff that we'll just never agree on, so I'll just respond to a few points.
Analysing the response of a pure ribbon hanging in free space doesn't really tell you anything about what it will behave like once in it's horn/waveguide, so there's really no point in going down that path.
In the horizontal plane the ribbon element is only 8.5mm wide, so you would expect it to have extremely wide horizontal dispersion in the 2-3Khz range, right ? If it didn't have a waveguide it would, however if you check the first image I've attached you'll see that the -6dB directivity is held to about 90 degrees (+/-45) right down to 2Khz. Below that it loses control and gets much wider as the waveguide becomes ineffective.
There is some apparent bloom in directivity around 5-6Khz but this is mostly a measurement artefact due to a 1dB diffraction dip in the on axis response, and the graph being normalized to the on axis response. (Thus boosting the apparent off axis response) There is some actual narrowing above 10Khz.
I've attached the vertical response as well. Yes it is a lot narrower, obviously.
Where the polar pattern and directivity control matter is in a room, not outdoors. Indoors, controlling the directivity is all about reducing the reverberant field in the room relative to the direct field, which is beneficial for good imaging at a distance among other things.
This is why I think sibilance is not a tonal balance issue but a resonances / spectral decay problem - the tonal balance of the recording and speakers just makes it more or less obvious, but a specific frequency response is not a cure.
Intermodulation and Harmonic distortion may be involved as well, (thanks to Lynn for starting me thinking along those lines) as it would cause sibilance region resonances to be excited by unrelated frequencies in the source material - for example the 3rd harmonic of 2Khz program material would excite a slow decaying 6Khz cone breakup resonance even though there is no 6Khz signal in the recording, making the distortion that much more obvious due to being stretched out in time. Likewise IM products from dense spectra would do the same - exciting the treble region resonances a lot more than the frequencies that belong in that region alone would.
Reducing distortion of the driver will reduce this tendency but only elimination of the resonances can cure it completely IMHO. Both would be ideal of course.
A horn or waveguide is an integral part of a drivers design - whether it's physically optional or removable is not really the point, when it's in place you effectively have a very different driver with different sensitivity, directivity, distortion, frequency response, power response, the lot.
Analysing the response of a pure ribbon hanging in free space doesn't really tell you anything about what it will behave like once in it's horn/waveguide, so there's really no point in going down that path.
Perhaps some actual measurements will tend to make you think differently.
In the horizontal plane the ribbon element is only 8.5mm wide, so you would expect it to have extremely wide horizontal dispersion in the 2-3Khz range, right ? If it didn't have a waveguide it would, however if you check the first image I've attached you'll see that the -6dB directivity is held to about 90 degrees (+/-45) right down to 2Khz. Below that it loses control and gets much wider as the waveguide becomes ineffective.There is some apparent bloom in directivity around 5-6Khz but this is mostly a measurement artefact due to a 1dB diffraction dip in the on axis response, and the graph being normalized to the on axis response. (Thus boosting the apparent off axis response) There is some actual narrowing above 10Khz.
I've attached the vertical response as well. Yes it is a lot narrower, obviously.
You seem to have gone a bit off the rails here - (a) I have listened to them outside (along with many other speakers over the years) and they sound just fine thanks, and (b) why wouldn't they ? Outdoors where there are almost no reflections you're listening to the on axis response of the speakers anyway - what does it matter what their polar pattern is then ? It makes no difference.
Where the polar pattern and directivity control matter is in a room, not outdoors. Indoors, controlling the directivity is all about reducing the reverberant field in the room relative to the direct field, which is beneficial for good imaging at a distance among other things.
After listening to ribbons for 8 years on a wide variety of music I would say that most but not all sibilance issues are speakers, not source material. Very rarely do I hear a recording that I would consider to have sibilance issues. A lot of modern music is mastered very hot in the treble (which I don't like) but usually even then doesn't manage to sound sibilant, it just sounds very hot.
This is why I think sibilance is not a tonal balance issue but a resonances / spectral decay problem - the tonal balance of the recording and speakers just makes it more or less obvious, but a specific frequency response is not a cure.
Intermodulation and Harmonic distortion may be involved as well, (thanks to Lynn for starting me thinking along those lines) as it would cause sibilance region resonances to be excited by unrelated frequencies in the source material - for example the 3rd harmonic of 2Khz program material would excite a slow decaying 6Khz cone breakup resonance even though there is no 6Khz signal in the recording, making the distortion that much more obvious due to being stretched out in time. Likewise IM products from dense spectra would do the same - exciting the treble region resonances a lot more than the frequencies that belong in that region alone would.
Reducing distortion of the driver will reduce this tendency but only elimination of the resonances can cure it completely IMHO. Both would be ideal of course.
Doesn't it suggest that there is something about the characteristics of the driver that causes it to still have sibilance ? Do you have a way to measure their CSD ? That would be interesting to see. I've attached a CSD plot of the G2 for comparison. The "resonance" at 15Khz is actually a capsule resonance on my microphone which shows up on every driver I measure, but otherwise I think it's reasonably accurate. The gate markers were set so decay begins at 0.4ms.
Attachments
Last edited:
grumble grumble I got inspired and I played around with the simulator last night and got a nice 3rd order acoustic slope on my tweeter rolloff It integrated quite well with the midbass (the butterworth hump somewhat filling the hole in the midbass response curve) too.... I suspect it would be an improvement, but back to the drawing board
I wonder if Joe Dappolitos change to 4th order (from the original 3rd order he used for MTM) was for similar reasons....
Tony.
Not so fast! First, invert the phase of the tweeter (compared the "normal" connection) and compare the overall FR to the "normal" condition. If there's no difference, then the drivers are in phase quadrature - 90 degrees relative to each other. If the output actually goes up in the crossover region when it is inverted compared to "normal", that's a bad sign - it means the phase angle of the drivers is significantly greater than 90 degrees.
This is bad not just at the crossover frequency, but anywhere close to it, even an octave away. Drivers with high phase angles relative to each other will sound phasey and disjointed - always a bad thing. It will mimic driver colorations, but will be apparent at all levels, unlike many driver colorations.
It is common in many crossovers, particularly when one or more drivers have sharp deviations in response close to the crossover frequency, for portions of the overall response curve to increase in the "non-normal" test condition.
This is a sign of trouble - the narrow frequency range where you see that increase will draw attention to itself, even though it looks like a harmless little dip in the overall response when reverted to the "normal" connection. It is nothing of the sort: a dip in driver response is benign and not very audible, but two (or more) drivers going out of phase with each other do not sound good at all.
This is why it is so important to tightly control the slopes in the region around the crossover, and why many 1st-order loudspeakers get in trouble - ripples in the response of one driver lead to small regions of excessive phase spread between the drivers. (Many designers keep forgetting that it is the vector sum of the electrical and acoustic responses that controls phase - neither one by itself is significant, but the sum is what you hear.) And if the drivers are going out of phase for a portion of the audio spectrum, yes, that certainly will be audible, even though it is hardly visible on the overall response curve.
To my ear, phase spread is more audible than small (1 dB or less) response deviations, so I'm willing to tolerate minor response variations in return for a better-controlled crossover region. I'm in a minority here, but not everyone is as sensitive to phase spread as I am. Going out on a limb, this is the most common and annoying fault I hear at hifi shows, especially with the WWMTMWW monstrosities that have become so common. They are extremely phasey and disjointed-sounding, and give multiway systems a bad name. Just getting a 2-way right is hard enough.
Last edited:
On more than a few occasions i have traced sibilance to the source material..
Just saying ...
I would agree that most source material does not suffer from sibilance, assuming the engineer knows how to use a De-esser. Unfortunately, many don't so there are plenty of lispy singers not properly corrected.After listening to ribbons for 8 years on a wide variety of music I would say that most but not all sibilance issues are speakers, not source material. Very rarely do I hear a recording that I would consider to have sibilance issues. A lot of modern music is mastered very hot in the treble (which I don't like) but usually even then doesn't manage to sound sibilant, it just sounds very hot.
This is why I think sibilance is not a tonal balance issue but a resonances / spectral decay problem - the tonal balance of the recording and speakers just makes it more or less obvious, but a specific frequency response is not a cure.
Intermodulation and Harmonic distortion may be involved as well, (thanks to Lynn for starting me thinking along those lines) as it would cause sibilance region resonances to be excited by unrelated frequencies in the source material - for example the 3rd harmonic of 2Khz program material would excite a slow decaying 6Khz cone breakup resonance even though there is no 6Khz signal in the recording, making the distortion that much more obvious due to being stretched out in time. Likewise IM products from dense spectra would do the same - exciting the treble region resonances a lot more than the frequencies that belong in that region alone would.
Reducing distortion of the driver will reduce this tendency but only elimination of the resonances can cure it completely IMHO. Both would be ideal of course.
The harmonic distortion and resonant breakup of speakers can easily make them sound sibilant, this can be revealed easily with a sine wave tone, looking at and listening to the harmonics above.
I still recall a test back in the 1980's, using a 6K tone into a 2445 driver resulted in a 12K tone of almost equal amplitude, "hashy" or "sibilant" .
Misalignment or debris in the gap (compression drivers or dome tweeters) can make for some sounds that depending on the program material may sound "sibilant"
Some drivers add harmonics at almost any drive levels, others increase harmonics with drive level. Some only seem to "cry" when the right blend of frequencies are present.
Back to the "can you have sparkling treble but without sibilance" question, it all depends on the speaker.
Art
DBM,
One of us is missing the bus. Maybe me.
I look at ur plot of ur tweeter and I see it narrowing with frequency.
Vertical and horizontal.
The Decca was NOT designed WITH the horn per se.
It was sold without a horn as well.
Afaik, it followed the Romagna design, which was integrated with a cast exponential horn (as seen in the Decca version in your link) which again was intended to permit the ribbon to be used lower in frequency, not for dispersion control.
If you don't hear any difference in your tweeters outside compared to inside either your inside is an anechoic room or else you didn't listen off axis??
_-_-bear
One of us is missing the bus. Maybe me.
I look at ur plot of ur tweeter and I see it narrowing with frequency.
Vertical and horizontal.
The Decca was NOT designed WITH the horn per se.
It was sold without a horn as well.
Afaik, it followed the Romagna design, which was integrated with a cast exponential horn (as seen in the Decca version in your link) which again was intended to permit the ribbon to be used lower in frequency, not for dispersion control.
If you don't hear any difference in your tweeters outside compared to inside either your inside is an anechoic room or else you didn't listen off axis??
_-_-bear
I think I need to buy some more caps, and some magnet wire to wind some coils, and start experimenting
I tried to get a 4th order acoustic slope last night, not a problem for the tweeter, worked a charm, but totally failed to get something satisfactory with the midbass... part of the problem being that Speaker Workshop seems to model resonant peaks from interactions in the crossover components and the driver resulting in a FR curve with 6db or higher peaks that aren't in the original response....... I will check this again tonight... I definitely tried flipping the tweeter polarity and was rather bemused that it seemed to make NO difference whatsoever, what I don't remember was whether this was the 3rd order butterworth, or 4th order bessel... I'd have thought at least the 4th order should show a null when 180 deg out of phase... if it was the 3rd order, then it sounds like it is on the right track
Something that is happening at the moment with my second order (ie quite good phase tracking)
at least something is reasonably right Tell me about it!
Tony.
If I recall correctly there were three different models of the Decca Ribbon Tweeter. The first was the Decca Kelly DK30, then there was a larger Decca London Ribbon, and finally a Decca Super Tweeter. Only the latter was hornless, and was released in the 1970s because so many people were removing the horns from the earlier versions to use them for supertweeters.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.