Almost 1500 replies to this thread, most useless to the topic.
-that's quite a long turd.
This is a shock to me.
Basically if frequency response is mathematically linked to transient response, then it doesn't matter what's producing the sound, if the FR is FLAT, then the sound if by definition transient perfect?
So then getting all worked up over some giant alnico magnet super driver makes no sense.
But this can't be right because people always talk about qts and damping and how a driver sounds different?
Unless that's all reflected in the frequency response? Maybe the cabinet (internally absorbing the backwave) and good dispersion and lack of diffraction and acceptable nonlinear distortion all combine to create a sense effortlessness and dynamics?
Any thoughts?
Last edited:
That's a bit harsh!Almost 1500 replies to this thread, most useless to the topic.
-that's quite a long turd.
Just because the thread has drifted from the original topic in the title doesn't make all the interesting discussion that followed a turd.
There are lots of threads on this forum with amazingly insightful discussion that bear little resemblance to what the thread started out to discuss - should they all be pruned or labelled as turds as well ?
What happened is that the original question posed was based on the implicit assumption that achieving ultra low distortion in drivers is desirable, and audible, thus the search to find a driver with really low distortion.
Looking back at the beginning of the thread in the very first page I see a post from Earl calling into question how important very low distortion actually is and whether it is even audible - that will have been the beginning of the thread drift from distortion to frequency response.
I tend to agree on this point - that it takes a lot more harmonic or intermod distortion than people think for it to become audible in music (as opposed to on pure tones) and that a lot of what people call "distortion" when they think they are hearing distortion, is not non-linear distortion but linear distortion like resonances.
I think this is particularly true on a speaker suffering from uncontrolled high frequency cone breakup. It sounds like "distortion" but it's really multiple high Q resonances that we are hearing.
I also see it written a lot that cone breakup "distortion" has some kind of threshold where it starts breaking up above a certain SPL from the driver, but I think the threshold is really in our hearing not in the speaker.
If there was some kind of breakup threshold in the speaker we should be able to measure a different frequency response through the breakup region at different SPL's - but when I've tried I don't see any difference between very low and relatively high levels!
I think it's a lot more likely that our ears just have an SPL threshold above which they become intolerant of high frequency resonances - below that threshold resonances don't bother us and we just hear the tonal imbalance they cause, above that threshold they start to get really annoying, really quickly, and make you reach for the volume control to turn it down!
I've certainly noticed that the flatter and smoother the frequency response is and the more free from high frequency resonances, the louder you can play it without any discomfort, "ear bleed" or perception of "distortion". And that's without taking any distortion measurements...
So the reason the thread drifted in the first place is that the quest for ultra low distortion drivers itself may be misguided and not identifying the real reason of why a speaker with flaws sounds bad to us at a higher SPL. I think that's quite an interesting discussion, especially if it challenges long held beliefs of the importance of achieving very low levels of non-linear distortion.
Last edited:
It's actually ARTA that I use not REW.
It sounds to me that in a long roundabout way that you are calling into question the validity of fast-fourier in general as a measurement technique ?
If so that's unfortunate, because that's what everyone is using these days, and I don't really see any reason to doubt it. If we can't trust our measurement tools to make valid measurements what can we trust ?
Not sure what relevance that has...
As someone who has been using ribbon tweeters for 15 years I think they do have a smooth effortless sound - at least the Aurum Cantus ones I have used. I can't speak for all ribbon tweeters.
But the reason why that is the case is shown in the measurements - a very smooth frequency response and clean CSD with little or nothing in the way of resonances. About the only resonance a closed ribbon tweeter usually has is a cavity resonance, but that is usually minor and easily corrected.
There's no magic pixie dust in a ribbon tweeter - it sounds good because it measures good - in the parameters that matter.
I strongly dispute the claim above - measurements just don't back that up, at least not in the tweeters I've measured.
The claim of "a number of side lobes" is incorrect. On a 50mm tall ribbon there are no significant side lobes in the vertical plane. I have a full set of vertical and horizontal off axis measurements I took of my tweeters years ago.
I just reviewed them to double check and there are no vertical side lobes at all over the majority of the tweeters range, and two minor vertical side lobes around 16Khz at 30 degrees off axis right down at about -12dB - so basically insignificant. See below:
If you used a much longer ribbon side lobes in the vertical plane might be an issue, but not in one that fits into the same size faceplate as a comparable dome tweeter.
Here's my thoughts on why ribbon tweeters can sound good, and this is specifically referring to closed back ribbons mounted in short waveguides, which are the kind I have used. An "open" ribbon without a waveguide will only have some of these properties and not others, and I have not used or measured that kind.
In something like a dome tweeter only the edge of the dome is driven, this applies a bending force on the diaphragm due to the unsupported mass in the middle of the dome.
As frequency goes up some point is reached where the centre no longer follows the edge and the dome starts to bend. This is analogous but opposite to cone break up in a cone driver where it is driven at the middle and the edge starts to do its own thing.
In a soft dome this breakup occurs in the audible range - typically starting as low as 6-8Khz, and the only thing you can do is try to damp the breakup - which is why soft domes are doped with a damping agent.
In my opinion this is not particularly successful for the same reasons that uniform damping around the edge of a cone driver is not - it just adds loss to the system but doesn't do anything to control the pattern of the standing waves on the dome, it just reduces their amplitude a bit.
The dome is also at a big disadvantage here compared to a cone driver - at least with a cone driver driven in the middle you have a transmission line from the centre to the edge of the cone which you could try to terminate at the edge with damping, with a dome driver you don't even have that theoretical advantage because you're driving it at the edge, and the bending wave travels towards the middle then keeps going and arrives at the edge again on the other side... So the only place you can try to damp it is through the cone itself.
A hard dome like beryllium on the other hand tries to push the breakup frequency beyond our audible frequency range - to about 24Khz. But because its an extremely stiff, undamped metal it resonates like crazy at those supersonic frequencies. A peak of 10dB or more is pretty common.
There is a lot of debate about whether this is audible in any way. For example could such a high SPL just outside our hearing still irritate us even if we can't hear it as a tone ? Could the strong resonance at that frequency cause intermod products that produces changes to the signal in the audible range ? Maybe.
The jury is still out on whether having a massive high Q resonance just out of our hearing range is benign or not, hence the continuing debate between soft domes and hard domes.
Personally I'm not happy with either compromise!
In a properly designed ribbon there are no "cone" resonances as the foil is both voice coil and sound producing diaphragm in one, this means the entire diaphragm is driven directly and uniformly by the magnetic field, meaning that there is no bending moment applied to the diaphragm. (This depends on the magnetic field being uniform of course, so that is part of the design challenge of making a good ribbon tweeter)
This means that done right, there just aren't any resonances in the diaphragm, either in-band or supersonic. This is a good starting point for the design of a clean driver.
Although you can make an open ribbon without a waveguide in my opinion unless you're making a long line array a ribbon really needs a small wave guide to be able to produce a high enough SPL at a low enough distortion, and have a high enough sensitivity to compete with or exceed the dynamic performance of a top of the line dome tweeter.
So what are the characteristics of a ribbon tweeter with an 8.5mm wide 50mm tall (visible) ribbon in a waveguide like the ones I use ?
In the horizontal plane the 8.5mm ribbon is a lot narrower than a 25mm dome, so the horizontal directivity is completely under the control of the waveguide design.
Horizontal directivity is kept between 120 degrees and 90 degrees from 2Khz up to about 15Khz, and it starts to narrow slightly above that. Below 2Khz pattern control completely collapses.
It looks very similar to a CD waveguide in the horizontal plane and its not surprising when you look at the profile of the sides of the waveguide which are basically straight +/-45 degrees chamfers with slight radiusing at the ends, similar to a CD waveguide.
The benefit of this is that in the horizontal plane out to about 45 degrees either side the frequency response is very uniform with almost none of the fall off of the top end that you would see from a 25mm wide dome tweeter. And when you go beyond +/-45 degrees the amplitude starts to drop away but it still remains relatively flat above 2Khz.
So any sidewall reflection is going to be spectrally balanced, which is considered to be good for imaging and spaciousness, whereas with a dome tweeter the top end of the treble will be rolled off in the sidewall reflection.
In the vertical plane the response is very different - it is as shown above, and is similar to what you'd see from a 50mm cone driver - a smooth uniform tapering off of the high frequency response the further you go off the vertical axis.
It's wide enough in the vertical plane to accommodate a variety of seating heights but narrow enough that if you stood up you will hear the top end of the treble roll off a little.
This is usually stated as a disadvantage for a ribbon tweeter, but I think it's actually an advantage, for two reasons.
The first is that while moderate amounts of sidewall reflections are generally considered to be "good" (at least for spaciousness) and can be separated by our ears from the direct signal, vertical reflections from the ceiling and floor are universally considered to be "bad", and our ears have difficulty separating them from the direct signal.
This can lead to blurring of imaging and perceived tonal imbalances. Some vertical directivity at high frequencies is beneficial to reducing the treble bouncing off the floor (if bare) and the ceiling.
I used to live in a house with an open plan living room and kitchen - including the kitchen the dimensions were 2.4 metre ceiling, 4 metres wide and 8 metres long with the speakers on the 4 metre wall. It was very apparent to me that the ribbon tweeters were able to maintain good stereo imaging focus to a much greater distance than dome tweeters used in their place on the same speakers, and I put that mostly down to a much reduced treble reflection from the ceiling in the long, low ceiling room.
The second reason reduced vertical dispersion is a good thing is to get a good overall power response balance in the room at high frequencies. A flat power response always sounds too bright, so its generally accepted that the power response should slope down in the treble a few dB to sound right in a normal listening room.
But how to achieve that ? If you had a driver with exceptionally wide dispersion you would have to make the direct sound slope down at high frequencies as well to achieve that goal, and that goes against something else that is generally accepted - which is that the direct response should ideally be neutral or flat in the treble.
So you have a contradiction that can only be solved by choosing the right directivity. If you use something like a dome tweeter that has the same horizontal and vertical directivity then in practice to be able to get both a flat on axis response and a suitably sloped down power response you have to choose a size of dome that means the sidewall reflection is also significantly sloped down, or if the tweeter is smaller you may have to slope down the on axis response a bit.
But with the different horizontal and vertical directivity of the ribbon tweeter you are able to have a flat, neutral on axis response, a flat neutral sidewall reflection, a greatly reduced treble reflection from floor and ceiling and get the desired roll off in overall room power response, all at the same time.
By reducing the contribution of the vertical off axis to the rooms high frequency power response you are able to increase the contribution of the horizontal off axis to the rooms high frequency power response.
So it's my opinion that having different vertical and horizontal directivity is actually beneficial, and I think this is a contributing factor to what can make a ribbon tweeter sound nice. We respond to reflections in the horizontal and vertical planes very differently - so why should a driver have identical directivity in both planes ?
Disadvantages of ribbon tweeters are they don't like to be driven low in frequency so invariably require a steep electrical filter (3rd or 4th order) and their harmonic distortion is a little bit higher than dome tweeters - something often pointed out by fans of dome tweeters, but if that slightly higher distortion is still below our thresholds of hearing (which I believe it is) that doesn't actually matter.
Attachments
Last edited:
By the way, just for laughs, here is the off axis response of a Peerless 811815 soft dome measured on the same baffle at the same time as the Aurum Cantus vertical dispersion measurement:
Pretty horrendous. It's not really the drivers fault, it shows what happens when a driver has no directivity control - all of the lumps and bumps in the polar response are due to the diffraction from the edge of the cabinet summing and cancelling at different angles. Notice how at higher frequencies (about 10Khz) where the dome becomes more directional the polar response becomes a lot cleaner and better shaped. But at lower frequencies it's a disaster.
The driver with directivity control of its own (waveguide) does not illuminate the cabinet edges and thus you don't get the same garbage looking polar response plots that are due to diffraction.
A good lesson for why making a tweeter have extremely wide dispersion is actually a bad idea. At least if it will be mounted on a conventional baffle...
Pretty horrendous.
It's not really the drivers fault, it shows what happens when a driver has no directivity control - all of the lumps and bumps in the polar response are due to the diffraction from the edge of the cabinet summing and cancelling at different angles. Notice how at higher frequencies (about 10Khz) where the dome becomes more directional the polar response becomes a lot cleaner and better shaped. But at lower frequencies it's a disaster.The driver with directivity control of its own (waveguide) does not illuminate the cabinet edges and thus you don't get the same garbage looking polar response plots that are due to diffraction.
A good lesson for why making a tweeter have extremely wide dispersion is actually a bad idea. At least if it will be mounted on a conventional baffle...
Attachments
Last edited:
Thanks Simon, great explanations. I"m still hung up a bit on the idea of transient perfection, iow if the frequency response of two very different drivers is the same then so is the transient response. IOW one with a light paper cone and giant magnet and one with a heavy metal one and small magnet would be the same in theory.
In reading your explanation about ribbons it seems that its true that soft domes operate in break up. This smears the signal in such a way that I don't know that it can be transient perfect. IOW transient perfection is impossible even with equalization. So this goes back to the issue of linearity and that not everything can be equalized. You would have to essentially stop the break up to get it flat.
Also narrow vertical dispersion has its advantages. But it seems that ribbons are often sold as being a light weight element with a powerful magnet and this may not be audible assuming that the transient is perfect with a flat frequency response. They probably would need less equalization than say a beryllium dome, which last I checked do have resonances in the audible range.
It seems safe to say that if you plan on using equalization then hard cones with high q resonances are better since they can be made flat. I think I was planning on this when I designed the speakers I use now, which are basically an 7" metal piston and a 1.7" metal piston. I think some of the metal "splashy sound" is actually those upper frequency resonances coming through, which are hard to perfectly equalize.
In reading your explanation about ribbons it seems that its true that soft domes operate in break up. This smears the signal in such a way that I don't know that it can be transient perfect. IOW transient perfection is impossible even with equalization. So this goes back to the issue of linearity and that not everything can be equalized. You would have to essentially stop the break up to get it flat.
Also narrow vertical dispersion has its advantages. But it seems that ribbons are often sold as being a light weight element with a powerful magnet and this may not be audible assuming that the transient is perfect with a flat frequency response. They probably would need less equalization than say a beryllium dome, which last I checked do have resonances in the audible range.
It seems safe to say that if you plan on using equalization then hard cones with high q resonances are better since they can be made flat. I think I was planning on this when I designed the speakers I use now, which are basically an 7" metal piston and a 1.7" metal piston. I think some of the metal "splashy sound" is actually those upper frequency resonances coming through, which are hard to perfectly equalize.
Last edited:
I have been at ribbon development for a number of years and can tell you this, so long as you are designing around a narrow ribbon width ( below about 15mm) its almost too easy to get good performance . At least above about 2 khz anyway.
Its when you want to use a wider diaphragm AND a lower than traditional crossover point that things get tricky.
narrow ribbons, like the one DBMandrake mentions are simply well behaved. Its very easy to get a clean decay. Widen that ribbon out and the fun begins with all sorts of tails creeping into the energy storage graphs. This is especially true of rigid corrugated designs. We see some exploiting the flatter diaphragm to get around this BUT they have other issues that limit them in the area that ribbons need to overcome most, low distortion , reliable operation to lower frequency.
On the issue of distortion I learned early on that low energy storage seems at least or more important to the ear than low distortion
Its when you want to use a wider diaphragm AND a lower than traditional crossover point that things get tricky.
narrow ribbons, like the one DBMandrake mentions are simply well behaved. Its very easy to get a clean decay. Widen that ribbon out and the fun begins with all sorts of tails creeping into the energy storage graphs. This is especially true of rigid corrugated designs. We see some exploiting the flatter diaphragm to get around this BUT they have other issues that limit them in the area that ribbons need to overcome most, low distortion , reliable operation to lower frequency.
On the issue of distortion I learned early on that low energy storage seems at least or more important to the ear than low distortion
Interesting. Is this due to the permanent magnetic field becoming non-uniform across the wide gap, or something else ? If the field is weaker in the middle of a wide ribbon then there would be some bending moment on the foil due to unequal force distribution and that would then potentially allow diaphragm breakup resonances to form. The more uneven the field the worse it would get.
I'm not sure I agree with really wide or really long ribbons for that matter, or trying to design them to go lower in frequency. It seems like trying to shoehorn them into an application that they're not really optimal for. I prefer a narrow ribbon and one that is not excessively long - something that can handle a similar crossover frequency to a typical dome tweeter.
I actually reduced the crossover frequency on mine from 4Khz 3rd order that I ran them at for many many years to 3Khz 4th order - for a number of reasons, although the main reason was because it let me get around a dip in the woofers response at 3Khz by taking advantage of shaping the low pass filter without adding additional components - something I couldn't have done if the crossover frequency had stayed at 4Khz.
I was a little bit worried that it might be pushing them too low but after a couple of months of listening I've come to the conclusion that they sound just fine going down to 3Khz, although I wouldn't want to take them any lower. Off axis response of the finished speaker at the crossover frequency is definitely improved as well, there seems to be a good match between the drivers now.
Last edited:
Presumably cone rigidity is advantageous, are there any cone drivers that are equally driven at the centre and the edge?
I don't know if there are, but even if there were you still have an unsupported undriven bit of the cone between the inner and outer voice coils, so it still doesn't solve the fundamental problem of only driving a cone over part of its surface that a ribbon tweeter does solve. And you have made your standing wave pattern even more complex...
uniform field can only do so much in the real world. Even with a near perfectly uniform field (tested this btw) the practical diaphragm is not perfect and in theory and practice there will be resonances.
Also the field in all the commercial offerings are not uniform. There can be quite a rise in field strength as you get close to magnet face.
The reason the narrow width corrugated ribbons so easily avoid the bending mode issues is that they just push the diaphragm resonance up high enough to be a non issue. From there you can play with the exact corrugation size shape till u get good control of these modes. other factors figure in too like the air load etc.
These designs however can be very responsive to these resonances and sometimes if the corrugations are over or under tensioned you will see their stiffness across ribbon change enough to upset the FR and CSD. This will likely happen above 10khz however in these narrow ribbons so may not be noticed by some
I agree. Some basic design need to be improved before eq of any kind is applied.
This is actually where a combination of stiffness, shape, mass distribution, need to be finely designed to integrate. Probably the toughest part in cone design.
When I replaced a ribbon on one of mine I did indeed notice that the tension on the foil (and therefore the number of corrugations visible in the gap) was quite important to get an exact frequency response match with the other tweeter.
When I initially changed the foil I made sure the number of corrugations visible was identical to the other tweeter, and while the frequency response was identical the sensitivity of the driver was about half a dB lower and impedance (at high frequencies) about 8 ohms instead of the original 7.5 ohms.
I assume the replacement foils I had are ever so slightly thinner than the originals. So I stretched it out slightly to get the same impedance and sensitivity - big mistake!
Now the sensitivity and impedance matched but the frequency response of the two tweeters diverged by nearly 1dB from about 10Khz up with the more stretched foil having reduced output from 10-20Khz...and the imbalance was audible on music. So no doubt the effect you are talking about in the change of stiffness of the corrogations.
So I threw that foil away and fitted a second one, putting it back to the original correct number of visible corrugations, matching the unaltered tweeter and just accepted the small difference in impedance and sensitivity, which I just compensate for by using a slightly different L-Pad for left and right tweeters in the crossovers.
The main thing is that their frequency responses are identical. I guess if I ever replace the ribbon in the other unit that they will be matched again (assuming my remaining spare foil is the same as the first two) and I will then have to change the L-Pad accordingly, but I don't see any reason to replace a foil when the tweeter is working perfectly.
Last edited:
When I replaced a ribbon on one of mine I did indeed notice that the tension on the foil (and therefore the number of corrugations visible in the gap) was quite important to get an exact frequency response match with the other tweeter.
When I initially changed the foil I made sure the number of corrugations visible was identical to the other tweeter, and while the frequency response was identical the sensitivity of the driver was about half a dB lower and impedance (at high frequencies) about 8 ohms instead of the original 7.5 ohms.
I assume the replacement foils I had are ever so slightly thinner than the originals. So I stretched it out slightly to get the same impedance and sensitivity - big mistake!
Now the sensitivity and impedance matched but the frequency response of the two tweeters diverged by nearly 1dB from about 10Khz up with the more stretched foil having reduced output from 10-20Khz...and the imbalance was audible on music. So no doubt the effect you are talking about in the change of stiffness of the corrogations.
So I threw that foil away and fitted a second one, putting it back to the original correct number of visible corrugations, matching the unaltered tweeter and just accepted the small difference in impedance and sensitivity, which I just compensate for by using a slightly different L-Pad for left and right tweeters in the crossovers.
The main thing is that their frequency responses are identical. I guess if I ever replace the ribbon in the other unit that they will be matched again (assuming my remaining spare foil is the same as the first two) and I will then have to change the L-Pad accordingly, but I don't see any reason to replace a foil when the tweeter is working perfectly.
YEP you have experienced how sensative the FR and sensativity can be to minor changes in the altering of tension on corrugated designs. As the corrugations size shape is changed from tension, the ribbons flex character can easily alter energy storage at different freqs., and even the smallest change can make a real difference especially when it messes with the balance between reference region and 10k. Critical for proper tonal balance. I usually like about 2 db down at 10k.