I may be misreading - but a surround that somehow chokes low excursion would cause even worse problems at high excusion, I imagine.
Perhaps someone out there that has rebilt (or manufactures) these different "liked" drivers can comment on the properties of the individual suspension components? The few cones I've handled without a spider were very compliant, Fs would have been subsonic. Otherwise, a bending transducer would result (at any volume).
I've often thought that a really wide surround, with concentric rings (possibly in shapes other than perfect circles) of differing properties would be the best way to allow a variable compliance, stagger/shift/absorb resonances, etc. Surely someone has experimented in this area . . .
-- Mark
Last edited:
I think given the added complexity, not much beyond the "tried and true" formula occurs as far as production. More a consequence of cost/benefit ratio. Different types of corrugation (or none at all). Different types of materials and doping used. ..and that's pretty much it. Surround-less designs have been done before, and Fertin still makes some. Flat-damped surrounds can be found from several sources - Focal's tweeters, Audax/Phl/Beyma/B&C/B&W mid.s, etc.. After that pretty much everything falls-back to some form of basic corrugation with a variety of styles - even reverse roll's like some Audax and Fostex drivers. Some of the newer applications on corrugation include the "tangential" surrounds from Fostex and 18sound (..along with various auto-sound manufacturers that likely predate Fostex's and 18sound's products.)
At some point those surrounds will start restricting excursion again, but I think long before then the spider is the dominate limiting factor. (..obvious exceptions would include surroundless designs and flat damped surrounds.)
Lynn
I can't give you a lot of rigourous comparison info vs. prosound drivers as I have not lived with those sort of units against the 285GMFs. I have heard them against a hodge-podge of older drivers from 12 to 18" , at our regular Yorkshire meetings and have owned a pair for a while myself and used in open baffle and Onken enclosure. You don't really hear what they can do in an Onken, as I found out to my cost ( several hundred hours of work with 25mm birch ply ! ) , as the stored energy and resonances swamp the bass tone mostly in those .
In open baffle you can judge them , particulary in the ones ('Quasars') James Doddington has run for many years, made from 20mm acrylic . They are just a really engaging , natural-sounding driver with lots of low-level info that binds the music together . They have good tonal colour and a very pleasant midrange ( for a bass driver ) that allows them to work fine with first-order slopes like the series crossover James designed for the 'Quasar' OB speakers . The midrange and 1-5kHz region sounds better than the FR curves would suggest .
The cones are very light - quoted as 20g, but I think from work on Hornresp I would say they are more likely to be 28-30g based on the Qts values.
It would be great to hear them against something like Alnico 414's - I expect the Altecs would have the edge due to the Alnico . There are of course Alnico and field coil Supravoxes, which I would love to hear, although the T-S parameters are different and the field coils are out of my price league !
Mea Culpa
Not in my case: three very clear problems were found (mix fet and bjt circuit)
a) turn on and moving from linear to saturated mode provided widely varying "small signal" parameters - in particular open loop gain changed as devices turned on and off.
b) mosfet capacitance varies as a function of gate to src and drain plus instantaneous channel current.
c) variation on the above was charge storage effects - both in gates and in diodes used to control gates - which really messed with closed loop stability.
Coming back to conventional class A/B (and my amp was NOT- I think Hugh Dean has finally solved some of the problems I dashed myself against) if you have enough bias current that there is a real area of Class A operation with suitable gain and enough slew rate and good looking bode plot then I agree with you claim. However usually output stages are biased to the point where both devices aren't really ON. But the 1w/1kHz and full power 1kHZ figures look OK on paper...
Not in my case: three very clear problems were found (mix fet and bjt circuit)
a) turn on and moving from linear to saturated mode provided widely varying "small signal" parameters - in particular open loop gain changed as devices turned on and off.
b) mosfet capacitance varies as a function of gate to src and drain plus instantaneous channel current.
c) variation on the above was charge storage effects - both in gates and in diodes used to control gates - which really messed with closed loop stability.
Coming back to conventional class A/B (and my amp was NOT- I think Hugh Dean has finally solved some of the problems I dashed myself against) if you have enough bias current that there is a real area of Class A operation with suitable gain and enough slew rate and good looking bode plot then I agree with you claim. However usually output stages are biased to the point where both devices aren't really ON. But the 1w/1kHz and full power 1kHZ figures look OK on paper...
Thanks for drawing my attention to that, marco_gea; something I will pay more attention to in future.
I fell into the trap of many loudspeaker designers in thinking the JBL and GPA would sound similar, possibly even identical. Both had pretty much the same efficiency, and thanks to the 700 Hz crossover, both would operate in the flat-response piston-band range. There were differences from 1.5 kHz on up, but the 3rd-order lowpass filter really should address that successfully. Same cabinet, so no difference there.
But in reality they sounded quite different. Not in terms of unwanted tonal colorations like driver breakups or the usual things that trouble midbass drivers, but something subtler ... rendition of tone colors in the music. The JBL was flatter and grayer sounding, almost as if the music was played by a synthesizer, and the subtle dynamic shifts disappeared back into the mix. I'm not saying it's a bad driver; it sounded better than a lot of audiophile drivers in low-efficiency speakers.
But the GPA 416 Alnico sounded much more "alive" and dynamic, and was a lot better dynamic match for the AH425 & Radian 745 Neo. Tone colors were much more vivid; you could differences between pianos, and small dynamic shadings of the performers. You could now hear all the little fingerings going with an electric bass guitar, for example, which was not audible with the other driver. Put another way, the 416 Alnico has a lot of the dynamic shading and tonality of a horn-loaded Lowther, minus the annoying whizzer shout.
A few years ago at the Rocky Mountain show, my friend Alexander (of RAAL) walked me through gap-saturation in full-range and bass drivers. Modern high-quality drivers are intended for Theile/Small applications, with Qts falling in the familiar 0.32~0.38 range, while old drivers either had very low Qts (if designed for high efficiency), or high Qts (table-radios and TV speakers). The high-efficiency drivers, as a side effect of trying to cram as many field lines in the gap as possible, frequently has partially or fully-saturated pole pieces. The cheap drivers, as a result of very small pole pieces and voice coils, also frequently had partially saturated gaps.
As the gap approaches saturation, additional lines of force are excluded, similar to a current source in an amplifier. Back-EMF's created by voice-coil motion cannot push into the magnetic structure of the driver, since it is already saturated, and will not accept more, so they must all return to the amplifier. Put another way, saturating the gap makes the magnet system invisible to the back-EMF's, in the same way that a current source isolates the power supply from the audio circuit of an amplifier.
In practice, the magnetic circuits of real-world drivers are only partially saturated, with the saturated regions at the corners of the pole-piece ... but this is an area where drivers are quite different from each other, particularly old-school and modern drivers, which are optimized to use magnets as cost-efficiently as possible, and to hit the Theile/Small marks of Qts.
The Rms spec conceals a pretty nonlinear system which is only in the grossest sense "mechanical resistance". On average, yes, it kind of looks like resistance, but building a linear mechanical resistor is really, really hard, particularly if you care about trying to minimize "stiction" effect (jumping from one location to another). Filling the spider with a lossy goo is one way, along with damping the surround.
The spider and surround are one of the weak points of dynamic drivers. If you leave off one of them, the diaphragm is free to flop from side to side, which is one of the problems with large-diameter dome midranges. In practice, we need the spider and surround to work together to make sure that the diaphragm stays centered as the cone moves in and out.
The spider is typically responsible for the majority of the Cms figure: softer spider, and the driver has more compliance. The Rms figure is not easy to control in production, although it fortunately only has a small effect on the overall T/S alignment. Rms is one of the things that can change quite a lot as the driver is "broken in" during the first 100 hours of use. At a micro level, when the driver is broken in, a random percentage of the fibers in the spider are actually torn apart, which softens the spider.
So there are several hidden variables that don't appear in FR curves, impulse response, or the Theile/Small specs. How much of the gap is saturated, and where? Is the voice coil overhung (like 99% of modern drivers), which means 30 to 50% of field lines cutting through the voice coil are outside the gap and not well-controlled? Is the voice-coil underhung (very unusual in modern loudspeakers), which means the field lines cutting through the voice coil are all in the gap? How are the spider and surround constructed, and what kind of damping (if any) is used?
Last edited:
I once tried a 18Sound 6ND410 (6" 102dB 120Hz Fs 0.24 Qts) on a 200Hz JMLC horn - FR modelled and measured well. I ran it in compression on a 2.75" throat with a small back chamber. But compared to a Lowther PM2A (both limited to same bandwidth) it quite obviously lacked all the subtlety and fine low level detail of the Lowther. I figured at the time it must be the nature of the suspension - run in could not make up for that much. I thought sensitivity was what it was all about, but kind of learnt there is more to it. Seems like there is a floor do downward resolution - The Pro driver is more sensitive, goes massively louder, but floor is higher. I know this is inadequate and subjective - not sure what a measure for low level resolution is.
martin
martin
I agree Martin. Low level coherence and expression is very important. Meyer Sound and Rudy Bozak both complained about the -40db down floor they were saddled with. John K, Lowther, EnABL, RAAL and I am sure others are all working this. Planet 10 Dave refers to it as downward dynamics and the only view I have seen, from typical max level portrayal of tests, comes from wavelet analysis.
Bud
Bud
Qms is the system Q-value considering only Rms as damping. The same way, Qes is the system Q-value considering only electrical damping.
Qms = 1/(2*pi*fs*Rms*Cms)
Qes = Re/(2*pi*fs*(Bl)^2*Cms)
A couple thoughts;
The Qms /Rms portion is the mechanical damping / Resistance connected to the radiator’s piston motion.
As a result, IF the driver only has piston motion, then these parts have a decreasing impact with increasing frequency.
In the 80’s when I was working on the servomotor driven subwoofers, I used to visit a local driver company called Heppner. Steve the engineer said something which captures the next part, “the best radiator material is the lightest, strongest, cheapest thing that has enough internal damping”
I think what you guys might mean is the internal damping of the cone body. This becomes increasingly important as the frequency climbs. Yes the edge has some damping, but ideally it wouldn't need to if you follow.
If you were to take the frequency response of each driver in an identical way for each and hopefully without room stuff (for this best on a large flat baffle), the difference between the drivers might be discernible in the breakup region.
The issue here is that damping adds mass and lowers efficiency and you get the best set of tradeoffs when the cone material itself has sufficient damping. That is why the stiffest / lightest materials are usually what “sounds best” .
At one point back then at intersonics I was developing a rotary full range driver and among the materials I tested for the radiator, “selected” balsa wood was an outstanding material. In several discussions with the two materials scientist I worked with, the ‘best guess” why was that wood in general came from plants that adapted to and damp wind loads coming from any direction and so they had Euler buckling at a similar load as tension failure.
A steel beam buckles well before snapping in tension while concrete snaps in tension far before buckling.
Anyway I am rambling, look at the response in the upper half of each drivers range and see if they look / act differently.
Best,
Tom Danley
The Qms /Rms portion is the mechanical damping / Resistance connected to the radiator’s piston motion.
As a result, IF the driver only has piston motion, then these parts have a decreasing impact with increasing frequency.
In the 80’s when I was working on the servomotor driven subwoofers, I used to visit a local driver company called Heppner. Steve the engineer said something which captures the next part, “the best radiator material is the lightest, strongest, cheapest thing that has enough internal damping”
I think what you guys might mean is the internal damping of the cone body. This becomes increasingly important as the frequency climbs. Yes the edge has some damping, but ideally it wouldn't need to if you follow.
If you were to take the frequency response of each driver in an identical way for each and hopefully without room stuff (for this best on a large flat baffle), the difference between the drivers might be discernible in the breakup region.
The issue here is that damping adds mass and lowers efficiency and you get the best set of tradeoffs when the cone material itself has sufficient damping. That is why the stiffest / lightest materials are usually what “sounds best” .
At one point back then at intersonics I was developing a rotary full range driver and among the materials I tested for the radiator, “selected” balsa wood was an outstanding material. In several discussions with the two materials scientist I worked with, the ‘best guess” why was that wood in general came from plants that adapted to and damp wind loads coming from any direction and so they had Euler buckling at a similar load as tension failure.
A steel beam buckles well before snapping in tension while concrete snaps in tension far before buckling.
Anyway I am rambling, look at the response in the upper half of each drivers range and see if they look / act differently.
Best,
Tom Danley
Tom,
In general I have to agree with you but at the same time I also disagree with the statement that the stiffest material is optimum. What I have found in my own cone research is that stiffness alone is not the answer as you allude to with the use of the balsa core cone you talk about. I started with my original cone material using only carbon fiber and resin matrix and found that the stiffest material also was the least damped material, in other words it would ring rather than damp or decay the signal over time. It was obvious in the waterfall plots of the differing materials that I tested at the time. I actually needed to lower the modulus and change the matrix resin to achieve the balance necessary for both a smooth frequency response and a clean decay across the time response. You only have to look at the decay properties of something like an aluminum cone to see this at its worst condition, the waterfall response plots will not look very nice. I agree that you want to optimize the cone before counting on the surround to finish the job of absorbing and reducing any reflective energy that can be reflected back down the cone body from a few points in the cone assembly. The first reflection will be at the glue joint between the cone and the surround. If this is a very stiff adhesive you will again see this in the decay rate and in a reflection returning to the cone. The surround itself is the second material that can again reflect any energy back to the cone and this is one reason that I avoid NBR material as they have a high energy return rate and are not very good at damping the outgoing wave. Next you will have the glue joint from the surround to the frame and again this can either help or hurt any absorption into the frame by again being a rigid material. It is the combination of all the materials in a series that all add together or damp together the outward traveling waveform.
The mass of the cone is important in setting the resonant frequency and must be balanced with the other components of the moving assembly. A very light cone and stiff suspension is going to move the fs up in the frequency range just as the opposite will move it down. It becomes the seesaw effects that must be balance for efficiency vs fs that is so critical to the end result that we want. As the cone becomes lighter and lighter we have to adjust the stiffness of the spider and the contribution of the surround to keep the balance that we are looking for in the device. A 6" speaker that we are tuning to work as a midrange is not going to have the same mass and stiffness as the same size cone that we want to produce bass frequencies. I know this is all rather simplistic but I know you get the point here. We can produce a cone that is so light that to make low frequency response the suspension has to be extremely soft and that becomes its own problem.
In general I have to agree with you but at the same time I also disagree with the statement that the stiffest material is optimum. What I have found in my own cone research is that stiffness alone is not the answer as you allude to with the use of the balsa core cone you talk about. I started with my original cone material using only carbon fiber and resin matrix and found that the stiffest material also was the least damped material, in other words it would ring rather than damp or decay the signal over time. It was obvious in the waterfall plots of the differing materials that I tested at the time. I actually needed to lower the modulus and change the matrix resin to achieve the balance necessary for both a smooth frequency response and a clean decay across the time response. You only have to look at the decay properties of something like an aluminum cone to see this at its worst condition, the waterfall response plots will not look very nice. I agree that you want to optimize the cone before counting on the surround to finish the job of absorbing and reducing any reflective energy that can be reflected back down the cone body from a few points in the cone assembly. The first reflection will be at the glue joint between the cone and the surround. If this is a very stiff adhesive you will again see this in the decay rate and in a reflection returning to the cone. The surround itself is the second material that can again reflect any energy back to the cone and this is one reason that I avoid NBR material as they have a high energy return rate and are not very good at damping the outgoing wave. Next you will have the glue joint from the surround to the frame and again this can either help or hurt any absorption into the frame by again being a rigid material. It is the combination of all the materials in a series that all add together or damp together the outward traveling waveform.
The mass of the cone is important in setting the resonant frequency and must be balanced with the other components of the moving assembly. A very light cone and stiff suspension is going to move the fs up in the frequency range just as the opposite will move it down. It becomes the seesaw effects that must be balance for efficiency vs fs that is so critical to the end result that we want. As the cone becomes lighter and lighter we have to adjust the stiffness of the spider and the contribution of the surround to keep the balance that we are looking for in the device. A 6" speaker that we are tuning to work as a midrange is not going to have the same mass and stiffness as the same size cone that we want to produce bass frequencies. I know this is all rather simplistic but I know you get the point here. We can produce a cone that is so light that to make low frequency response the suspension has to be extremely soft and that becomes its own problem.
Hi
I agree with all of that personally but his point (being at a company that made lots of different drivers) also had the word “cheapest” in it.
That is why most cones are still paper fiber based, they aren’t always “the last word” in a given aspect but they can be very good and is why most cones are still fiber based.
Also, it is worth considering “where” lowering the Qm with mechanical damping begins to effect the response.
With an efficient horn driver (where Qm was a broad band drag)I remember in one case that a Qm of less than about 4 was a problem so far as the efficiency and output.
I would hesitate to say “put goo on the edge suspension” to arrive at damping because it’s damping changes so much with time, heat and very recent motion. All the resistances (even mechanical damping) dissipate energy in the form of heat produced internally.
Best,
Tom
I agree with all of that personally but his point (being at a company that made lots of different drivers) also had the word “cheapest” in it.
That is why most cones are still paper fiber based, they aren’t always “the last word” in a given aspect but they can be very good and is why most cones are still fiber based.
Also, it is worth considering “where” lowering the Qm with mechanical damping begins to effect the response.
With an efficient horn driver (where Qm was a broad band drag)I remember in one case that a Qm of less than about 4 was a problem so far as the efficiency and output.
I would hesitate to say “put goo on the edge suspension” to arrive at damping because it’s damping changes so much with time, heat and very recent motion. All the resistances (even mechanical damping) dissipate energy in the form of heat produced internally.
Best,
Tom
as far as the fane 8m goes, i think you have very well measured and research founded results from JLH. for sure the driver goes much lower than 150hz. the real T/S parameters are as follows:
http://www.prodance.cz/protokoly/studio_8M.pdf
Interesting observation, and not the first time I've seen the same theory proposed, I have a question though, Rms is being measured at the bass resonance of the driver, but are you saying that the superior "low level detail" of a driver with low Rms is occurring throughout a wide frequency range ? Or just that the low level detail is better in the bass region where the Rms is measured ?
It seems a bit simplistic to extrapolate the Rms of the driver at bass frequencies to performance at higher frequencies, for one thing the Rms is unlikely to be the same at much higher frequencies, the other is that as the cone enters the breakup region things get a whole lot more complex.
Out of interest I went back and had a look at the Rms figure of a couple of drivers I use and have used.
Measured T/S specs of the two drivers can be found in the following post:
http://www.diyaudio.com/forums/multi-way/182688-help-understanding-physics-vas.html#post2461984
One is a Visaton W300S 12" woofer with a relatively heavy and damped butyl rubber surround, sensitivity about 92dB.
Qms is low at 2.19, and Rms is quite high at 5.2.
The second is a Coral Flat 8 II, an 8" full range driver with a cloth surround, a sensitivity of about 94dB (so not a lot higher) a Qms of about 7 (at least at low signal levels! See the discussion in that post) and an Rms of 0.43 - more than 10x lower than the first driver.
Interesting that I consider this driver to have extremely good low level detail compared to many others that I've heard, to find out the Rms is so low.
As for the Visaton - it sounds fine as a woofer to me, I never used it above 300Hz so I don't know what sort of low level detail it might have in the midrange, but I bet its nowhere near as good, for multiple reasons.
Another possible factor to consider regarding low level detail is the change in Rms (and other parameters) with drive level, which change with all drivers, but change a lot more on some drivers than on others.
The post linked to was specifically exploring the large changes in T/S parameters with excurison of the Coral driver, although the Visaton driver also showed the same trends to a lesser degree. I was looking at Fs in particular which falls a lot with increased excursion, however Qms and Rms also change, Qms reducing and Rms increasing with increased excursion.
Rms on the Coral at high drive levels (1/2 Xmax) is 0.55 but falls to 0.43 at 20dB lower drive levels...
myhrrhleine,
I would wonder how much of the original properties of balsa would be retained after it was turned into a slurry that would be used to make paper? Whether the low density would be retained would be a question mark after destroying the original grain structure of the wood. What is little understood by many about paper cone manufacture is that there is often other fibers added to the mix besides cellulose when creating the cones. Acrylic fibers, cotton and other fibers are often added to high performance paper cones to give a combination of properties. It isn't always a simple paper formulation that is used in speaker cones.
I would think without knowing the particulars that in Tom Danley's application that the balsa was used as a core material between two outer layers of some other material. This is a common method of light weight composite construction and the grain orientation has a lot to do with the final results. Whether the grain is using the grain in compression or in a tensile condition will give very different end results. The majority of the balsa core that I have seen used in composite construction was used with the end grain across the thickness and not planar to the surfaces. It also make a major difference in how the material will react if it is used only in its pistonic region or has to be used above the pistonic range where you have to consider breakup modes of any cone. This is one of the reasons that I found that a so called extremely rigid material can cause problems when you have the material working above this range, the cone needs to deform at some point when producing the higher frequencies and this is where the rigid materials can start to sound rather harsh if the bending modes create very high Q resonant problem. It will show up as a peak in the FR and also can cause some serious problems in the waterfall plots with extremely long decay times in a very narrow band of frequency.
I would wonder how much of the original properties of balsa would be retained after it was turned into a slurry that would be used to make paper? Whether the low density would be retained would be a question mark after destroying the original grain structure of the wood. What is little understood by many about paper cone manufacture is that there is often other fibers added to the mix besides cellulose when creating the cones. Acrylic fibers, cotton and other fibers are often added to high performance paper cones to give a combination of properties. It isn't always a simple paper formulation that is used in speaker cones.
I would think without knowing the particulars that in Tom Danley's application that the balsa was used as a core material between two outer layers of some other material. This is a common method of light weight composite construction and the grain orientation has a lot to do with the final results. Whether the grain is using the grain in compression or in a tensile condition will give very different end results. The majority of the balsa core that I have seen used in composite construction was used with the end grain across the thickness and not planar to the surfaces. It also make a major difference in how the material will react if it is used only in its pistonic region or has to be used above the pistonic range where you have to consider breakup modes of any cone. This is one of the reasons that I found that a so called extremely rigid material can cause problems when you have the material working above this range, the cone needs to deform at some point when producing the higher frequencies and this is where the rigid materials can start to sound rather harsh if the bending modes create very high Q resonant problem. It will show up as a peak in the FR and also can cause some serious problems in the waterfall plots with extremely long decay times in a very narrow band of frequency.
Use the Search, Luke
See pages 51,57,58,62,63,313,384,866 for starters (Search for "tone tubby beyond ariel")
See pages 51,57,58,62,63,313,384,866 for starters (Search for "tone tubby beyond ariel")
In a nutshell, yes.
While it is true that Rms is measured at resonance, and that it is non-linear with frequency and level, I think it is still reasonable to assume that a low-Rms woofer will still exhibit a lower 'acoustic resistance' (and hence better low-level detail retrieval capacity) throughout a wide frequency range than a higher-Rms one.
Simplistic, I agree. But in lack of a more complex model, I find that it works as an approximate predictor of low-level detail retrieval capacity, at least at frequencies where the behaviour of the woofer is still mostly pistonic.
Marco