Qms is just a measure for the amount of friction that dampens cone movement. Low Qms=much friction. Now, there is no such thing as linear mechanical friction without stick, so the less mechanical friction, the better it is. Leaves open one other source of friction: voice coil induced. Why do you think there are stil VC's around with organic formers? The reason is to prevent the occurance of eddy currents, which not only cause friction, but also introduce non-linearities. Thereby, high Qms is a good indicator for the relative absence of distortion causing mechanisms.
vac
When it hits the target. That assumes that you've started out with a target- if you haven't, you'll never make any progress. There's an old expression that says, "In the development of a product, there is a point at which it is necessary to shoot the engineers and get it out the door."
Jacco, if you're listening to 2905's powered by ExtremA's, I can publicly admit I envy you.
Last edited:
You should be able to use an averaged FFT technique to do this.
Maybe pulsed and windowed tones to capture only primary energy. Hard at low frequencies, but I have always wanted to explore issues of linearity at low SPL's
Dejan,
Amen.
Earlier on I had a cute quip about the drivers used in your favourite speaker (the Audax HDA Drivers and the Ti Dome). They all for example have very badly behaved breakup and I can hear that and it bothers me. I have heard many speakers with Audax HDA drivers in the 90's (they where fashionable in the UK) but I could never get over the harshness and edginess of them, something NOT related to frequency response, but strictly to the cone material.
What I wrote in my self censored quip was that I was sure a very good speaker could be made with these drivers, but I simply did not wish to work that hard.
When designing a speaker I always listen to candidate drivers naked lying on the floor, no XO, no EQ, nothing. Anything that has obvious objectionable sonic flaws I exclude (which is most), simply because I know I will have to fight the driver, fight it's problems to get descent sound, instead of being able to work with the Drivers to maximse what they do well.
In those tests the Audax HDA Cones failed miserably, while the Audax doped paper cones and (to a little lesser degree) woven carbon fibre cones passed very well.
Ciao T
Amen.
Earlier on I had a cute quip about the drivers used in your favourite speaker (the Audax HDA Drivers and the Ti Dome). They all for example have very badly behaved breakup and I can hear that and it bothers me. I have heard many speakers with Audax HDA drivers in the 90's (they where fashionable in the UK) but I could never get over the harshness and edginess of them, something NOT related to frequency response, but strictly to the cone material.
What I wrote in my self censored quip was that I was sure a very good speaker could be made with these drivers, but I simply did not wish to work that hard.
When designing a speaker I always listen to candidate drivers naked lying on the floor, no XO, no EQ, nothing. Anything that has obvious objectionable sonic flaws I exclude (which is most), simply because I know I will have to fight the driver, fight it's problems to get descent sound, instead of being able to work with the Drivers to maximse what they do well.
In those tests the Audax HDA Cones failed miserably, while the Audax doped paper cones and (to a little lesser degree) woven carbon fibre cones passed very well.
Ciao T
Hi,
Hold it.
Qms is a measure of loss of the RESONANT system at the fundamental resonance. How this loss occurs is a totally different story.
True, but if we (just use one example) use a flow resistance integrated into the basket, our stiction is at the level of a single Nitrogen Atom...
Nope, it ain't actually, work out the proof yourself.
High Qms is only a good indicator of this if specific (poor) design techniques are employed. I am not suggesting that such options are a good idea to obtail low Qms.
However that does not invalidate my point that the mechanical resonant system should be in a linear fashion and and by mechanical means be critically damped, in favour of attempting to damp it electrically through the Amplifier.
Ciao T
Hold it.
Qms is a measure of loss of the RESONANT system at the fundamental resonance. How this loss occurs is a totally different story.
True, but if we (just use one example) use a flow resistance integrated into the basket, our stiction is at the level of a single Nitrogen Atom...
Nope, it ain't actually, work out the proof yourself.
High Qms is only a good indicator of this if specific (poor) design techniques are employed. I am not suggesting that such options are a good idea to obtail low Qms.
However that does not invalidate my point that the mechanical resonant system should be in a linear fashion and and by mechanical means be critically damped, in favour of attempting to damp it electrically through the Amplifier.
Ciao T
Dejan,
When it "feels right".
This is basically an intuition based restatement of Sy "When it hit's the target".
I find that the the real world rarely if ever is a simple and clearcut as to provide such a "target". If it did every piece of audio gear could be designed according to such clear targets and at the same pricepoint and with the same thickness frontpanel they would all sound exactly the same and more cruciall, they would all sound great, basically the exact reverse of the real situation.
So one is left with employing that uniquely human facility, intuition or the ability to make mostly correct choices from insufficient data to allow such a choice to be made.
Some people have this facility to a degree (this seems as far as I can tell a Talent, rather something that can be learned) that allows them to look a given system they understand sufficiently (or sometimes not even that) and "see" what is wrong with it.
So I can for example often look at something and say "change this" or "change that" (this does not apply to audio alone - one of main applications of this skill is actually in the context of financial computer systems) and if you change it the way I said things that before did not work well suddenly do.
If you asked me to give you detailed technical justification of my suggested changes you would find me mostly at a loss. I can with enough time come up with suitable theories and explanations, sure, though they may not be the right ones.
Ciao T
When it "feels right".
This is basically an intuition based restatement of Sy "When it hit's the target".
I find that the the real world rarely if ever is a simple and clearcut as to provide such a "target". If it did every piece of audio gear could be designed according to such clear targets and at the same pricepoint and with the same thickness frontpanel they would all sound exactly the same and more cruciall, they would all sound great, basically the exact reverse of the real situation.
So one is left with employing that uniquely human facility, intuition or the ability to make mostly correct choices from insufficient data to allow such a choice to be made.
Some people have this facility to a degree (this seems as far as I can tell a Talent, rather something that can be learned) that allows them to look a given system they understand sufficiently (or sometimes not even that) and "see" what is wrong with it.
So I can for example often look at something and say "change this" or "change that" (this does not apply to audio alone - one of main applications of this skill is actually in the context of financial computer systems) and if you change it the way I said things that before did not work well suddenly do.
If you asked me to give you detailed technical justification of my suggested changes you would find me mostly at a loss. I can with enough time come up with suitable theories and explanations, sure, though they may not be the right ones.
Ciao T
Well, my "target" is often to do it as well as I am able.
But I know what you mean, even so one has that gut feeling that this is it, you can beat it with a sledge hammer, but it won't get any better, because the end is usually the end of my own know-how.
I also totally understand the saying you quoted - just when I think I'm done, another idea comes from somewhere, and I just HAVE to try it out, and ... well, you know the rest.
But in the real world, I find it's much better to let go at some point, let it live its life and jot down your observations, and the revisit a year from now.
Dejan,
When it "feels right".
This is basically an intuition based restatement of Sy "When it hit's the target".
I find that the the real world rarely if ever is a simple and clearcut as to provide such a "target". If it did every piece of audio gear could be designed according to such clear targets and at the same pricepoint and with the same thickness frontpanel they would all sound exactly the same and more cruciall, they would all sound great, basically the exact reverse of the real situation.
So one is left with employing that uniquely human facility, intuition or the ability to make mostly correct choices from insufficient data to allow such a choice to be made.
Some people have this facility to a degree (this seems as far as I can tell a Talent, rather something that can be learned) that allows them to look a given system they understand sufficiently (or sometimes not even that) and "see" what is wrong with it.
So I can for example often look at something and say "change this" or "change that" (this does not apply to audio alone - one of main applications of this skill is actually in the context of financial computer systems) and if you change it the way I said things that before did not work well suddenly do.
If you asked me to give you detailed technical justification of my suggested changes you would find me mostly at a loss. I can with enough time come up with suitable theories and explanations, sure, though they may not be the right ones.
Ciao T
Yes, I agree with the above. We all have that feeling that this IT, right now, I simply can't do any better, just perhaps different.
Also agreed on that feeling. At times, I look at something I've done, and I wonder why was my mind our grazing grass in the park, instead on my work, because under normal conditions, I'd NEVER do something like that. And it's just a glance.
Works for other people's work, as well. Just take a look and notice a detail which seems out of place, or even quite wrong. But I have learnt not to blurt anything out, to keep my cool and investigate slowly. On a couple of occasions, that approach saved me some embarassment, because it turned out the other guy had a valid reason to do it just that way.
People don't like to be shown up wrong. Me, I will never go any further with anything after the simulation stage before my friend Ollie has had a look and has okayed it - he thinks on a different plane from mine, and since he's a really a great person to have around at all times, I am 100% sure he will give it a long and hard look, and if anything seems out of place, he will let me know. We discuss it, we pit ideas, we drink 3 pints of beer each, and the problem is solved.
The influence of beer on audio is significant.
Well, things are moving along. I have now reached my „Mk.2“ version as a result of the discussions on the forum and Thorsten’s advice.
The input stages now consist of low noise BC 550/560 B trannies. All bipolar. Probably due to my inexperience with FETs, I could not make the initial FET version go down with THD to where I’d like to see it, meaning <0.2% open loop. I did this quite easily with the bipolars. Oddly enough, using a single ended input, I did manage to get it right. Still much to learn there.
I changed the double paralleled transistors in the second voltage gain stage for a classic cascode. Again, the prime reason was the difference in the 50 kHz distortion, where the cascode came out better – not significantly so, but better anyway.
After that, to the end of the amp, a few values have changed and a few transistors have changed, but nothing much.
Thorsten suggested I add a FET buffer; that sounded reasonable, so I did. And it worked rather well, I must say. The level equalizing pot is now before the FETs, a much better place for it, I think.
The measured simulator performance is:
Full power open loop THD at 20 Hz: 0.16%
Full power open loop THD at 1 kHz: 0.18%
Full power open loop THD at 20 kHz: 0.23%
Full power open loop THD at 50 kHz: 0.32%
Full power closed loop with NFB 20 dB THD at 20 Hz: 0.12%
Full power closed loop with NFB 20 dB THD at 1 kHz: 0.13%
Full power closed loop with NFB 20 dB THD at 20 kHz: 0.19%
Full power closed loop with NFB 20 dB THD at 50 kHz: 0.035%
Open loop bandwidth, ref. 28.3 Vrms/4 Ohms: 80 kHz -3 dB
Open loop bandwidth, ref. 28.3 Vrms/8 Ohms: 83 kHz -3 dB
Since the input stage is biased at 1.38 mA per transistor, and since the compensation capacitor is only 3 pF, that should give me a theoretical voltage slew rate of around 460 V/uS, an absurd value which will in real life be more like 100 V/uS. Anyway, way more than I am likely to need.
Q6 and Q20 are biased at 5.48 mA – could have been more, but I saw no point in overdoing it.
The output stage is biased at 122 mA per transistor, or 366 mA for the whole; not spectacular, but even on the simulator, more gave no observable benefits. So why fry them?
A novelty are resistors R10 and R56, at 330 kOhm. I saw that applied somewhere, not sure where but I think it might have been an HK model 879 power amp or some such. I tried with and without it, and found that it does help a very little buit at 50 kHz. I am still digesting it, but decided to leave place for it on the printed board.
The one and only surprise was the small difference between 8 and 4 Ohm loads. I expected a bit more, but I’m not complaining. At 20 kHz, 4 Ohms, THD was just 0.032%, and at 50 kHz, it was just 0.055%.
Perhaps wishful thinking on my part, but I took this to show rather good linearity overall, even if the global NFB was just 20.4 dB. In that, I remained true to my initial set of objectives, and that’s the only point over which Thorsten and I disagree, not on the priciple, but on the execution, since he plans to use about half that. Well, Germans are notoriously pig headed, but that’s only because you haven’t met any Bosnians yet, compared to which the Germans are all Mother Theresa. Even the housewife Merkel.
Oh yes, I considered very carefully my initial idea of electronically stabilizing the current gain stages and eventually gave it up as inpractical in my case. I don’t think the amp will ever be asked to deliver more than say 50W/8 Ohms, and I think I can safely ride that out with classic electrolytic capacitors, which I happen to have a LOT of.
The input stages now consist of low noise BC 550/560 B trannies. All bipolar. Probably due to my inexperience with FETs, I could not make the initial FET version go down with THD to where I’d like to see it, meaning <0.2% open loop. I did this quite easily with the bipolars. Oddly enough, using a single ended input, I did manage to get it right. Still much to learn there.
I changed the double paralleled transistors in the second voltage gain stage for a classic cascode. Again, the prime reason was the difference in the 50 kHz distortion, where the cascode came out better – not significantly so, but better anyway.
After that, to the end of the amp, a few values have changed and a few transistors have changed, but nothing much.
Thorsten suggested I add a FET buffer; that sounded reasonable, so I did. And it worked rather well, I must say. The level equalizing pot is now before the FETs, a much better place for it, I think.
The measured simulator performance is:
Full power open loop THD at 20 Hz: 0.16%
Full power open loop THD at 1 kHz: 0.18%
Full power open loop THD at 20 kHz: 0.23%
Full power open loop THD at 50 kHz: 0.32%
Full power closed loop with NFB 20 dB THD at 20 Hz: 0.12%
Full power closed loop with NFB 20 dB THD at 1 kHz: 0.13%
Full power closed loop with NFB 20 dB THD at 20 kHz: 0.19%
Full power closed loop with NFB 20 dB THD at 50 kHz: 0.035%
Open loop bandwidth, ref. 28.3 Vrms/4 Ohms: 80 kHz -3 dB
Open loop bandwidth, ref. 28.3 Vrms/8 Ohms: 83 kHz -3 dB
Since the input stage is biased at 1.38 mA per transistor, and since the compensation capacitor is only 3 pF, that should give me a theoretical voltage slew rate of around 460 V/uS, an absurd value which will in real life be more like 100 V/uS. Anyway, way more than I am likely to need.
Q6 and Q20 are biased at 5.48 mA – could have been more, but I saw no point in overdoing it.
The output stage is biased at 122 mA per transistor, or 366 mA for the whole; not spectacular, but even on the simulator, more gave no observable benefits. So why fry them?
A novelty are resistors R10 and R56, at 330 kOhm. I saw that applied somewhere, not sure where but I think it might have been an HK model 879 power amp or some such. I tried with and without it, and found that it does help a very little buit at 50 kHz. I am still digesting it, but decided to leave place for it on the printed board.
The one and only surprise was the small difference between 8 and 4 Ohm loads. I expected a bit more, but I’m not complaining. At 20 kHz, 4 Ohms, THD was just 0.032%, and at 50 kHz, it was just 0.055%.
Perhaps wishful thinking on my part, but I took this to show rather good linearity overall, even if the global NFB was just 20.4 dB. In that, I remained true to my initial set of objectives, and that’s the only point over which Thorsten and I disagree, not on the priciple, but on the execution, since he plans to use about half that. Well, Germans are notoriously pig headed, but that’s only because you haven’t met any Bosnians yet, compared to which the Germans are all Mother Theresa. Even the housewife Merkel.
Oh yes, I considered very carefully my initial idea of electronically stabilizing the current gain stages and eventually gave it up as inpractical in my case. I don’t think the amp will ever be asked to deliver more than say 50W/8 Ohms, and I think I can safely ride that out with classic electrolytic capacitors, which I happen to have a LOT of.
Last edited:
Err, No , dont think so ...
Never said it meant only horns ...
What i said was HE Horn speakers , not HE speakers per se ....
Best to define what is Low /mid/HE then we can move on to apple vs apple discourse ..
Yes voicing is very important ....
Reality , paper sounds best IMO, of course I have had good Poly,FG and CF speakers that worked , but paper is still my favorite ..
Because Thorsten took out some selected sentences, taking out the original logic into which they where placed, I feel forced to repeat myself all over again in duplication for a second time:
(post to which Thorsten reacted)
Qms is just a measure for the amount of friction that dampens cone movement. Low Qms=much friction. Now, there is no such thing as linear mechanical friction without stick, so the less mechanical friction, the better it is. Leaves open one other source of friction: voice coil induced. Why do you think there are stil VC's around with organic formers? The reason is to prevent the occurance of eddy currents, which not only cause friction, but also introduce non-linearities. Thereby, high Qms is a good indicator for the relative absence of distortion causing mechanisms.
(my reaction to Thorsten's reaction, as numbered by me for convenience)
Ad 1.) Yes, that's how you measure Qms as a proxy for friction in the system.
Ad 2.) You mean like in a pierced pole filled with some dampening material, with the dust cap working as a piston? Has been tried, does not work. After all, high Qms is no problem anyways, since driver can (and should) be controlled by enclosure and electronics only.
Ad 3.) Don't remember where this refers to, but it is never a substantial argument.
Ad 4.) All forms of friction lead to non-linearities, and though some forms might be more benign than others, they all lead to distortion.
Ad 5.) Since Qes is always lower than Qms in even the worst speaker I ever measured, speakers are always mainly controlled by electronics and the box. What's wrong with that? It is rather the opposite. The more your electronics are able to control cone movement, the better it is. Apart from the distortion caused by friction, that is another reason you want to avoid it. It stands in the way of the original signal controlling the exact movement of the cone. In other words, for this reason alone, high Qms is a goal in itself.
vac
Hi,
Let me put it this way.
First, your logic is based on a false premise, namely that the current Amplifier-Speaker interface (Voltage) is defined correctly and is beneficial and should be used in preference to alternatives.
With that fundamental flaw and the use of dynamic speakers "contra naturam" the rest is pointless. You are debating details of putting the horse behind the cart with someone who is looking what it takes to put the horse where it belongs.
Second, your logic lacks any demonstration of the leaps and bounts it takes. In other words, you do not demonstrate logically that A follows from B and C follows from B, you claim so and incorrectly.
Third, my comments are not based in this context you use, but in one where Speakers are driven correctly, in accordance with the nature of the device. Moreover, I have experience with such systems as well as with speakers where the fundamental resonance is damped by mechanical means, so at least I know what I am talking about.
Now to some specific points:
Just because you measure damping does not mean it is based on friction in the suspension. There is no such direct link. All you can do is to measure Qm and state that the system contains X mechanical damping with Y Linearity. You cannot in fact determine from the measurement how this damping is applied.
No. I mean the way in which it has been applied in german Studio monitor Drivers since the mid 1930's up to 1992, when the final inheritor of the Eckmiller legacy in Berlin closed doors.
So, what is the level of non-linearity introduced by a flow resistance. Hint, in east Germany we used flow resistances also for making cardioid Speaker columns that where cardioids down to around 150Hz, so I know how non-linear they are...
Qes is a very poor way to not "control" the speaker. Plus, if you think Qms is nonlinear, try measuring Qes nonlinearity.
What Qe attempts and miserably fails is to correct a mechanical problem by simple electrical means. In the process it actually causes by far more gross distortion all across the bandwidth of the speaker, while still failing to critically and consistently damp the cone at all signal levels and frequencies.
The only problem is that your electronics cannot control cone movement through Qe and low source impedance.
Cone acceleration (and thus audio output) outside the frequency range around resonance is proportional strictly and only to voice coil current and magnet field/suspension linearity.
So Qe is not in the picture, only Re. And Re can be taken out of the picture (together with Qe) by current drive.
Now, lets look at the fundamental resonance frequency region (which is quite narrow in range). Does Qe control cone movement directly and in a linear fashion? Nope, it is in fact heavily signal envelope dependent and modulated by excursion (where usually larger excursions decrease damping).
Contrast this with damping by a flow resistance integrated into the basket, it has neither appreciable stiction nor non-linearity with level and driving the speaker with current.
In one fell swoop we have eliminated:
1) Thermal Compression (which is hotly debated but real neverless)
2) Eddy Current (in the pole piece) induced distortion (which is odd order and cubic law and leads to often observed preponderance of H3 over H3 in the midrange for many drivers)
3) Low Frequency non-linearity, which is caused by suspensions and electrical systems that are highly non-linear.
If we combine this with a magnet system that has a constant field strength for the range of cone travel of interest we will have made a Speaker system that has a dramatically improved performance in almost all relevant domains.
Of course, this system would be incompatible with common amplifiers, however, unlike the situation three or more decades ago when amplifiers and speakers where separate, active speakers are becoming more common.
However this debate about how to optimise the Amplifier/Speaker system using unconventional, if superior means is already present in several threads on that topic here. including this more recent one:
http://www.diyaudio.com/forums/vendors-bazaar/157787-secret-tube-amplifiers-revealed-much-more.html
So no need to drag it in here, where generally conventional systems are debated, or to then take my comments on Driver Parameters in threads discussing current drive out of context to make it seem as if there is a contradiction and to them complain when I address not your flawed logic, but deal with the individual contentions you have raised.
BTW, my noted to the hysteresis effects in low efficiency systems in this thread here referred mainly to the system behaviour outside resonance region and to the results of attempting to damp cone resonance with the rubber surround...
Ciao T
Let me put it this way.
First, your logic is based on a false premise, namely that the current Amplifier-Speaker interface (Voltage) is defined correctly and is beneficial and should be used in preference to alternatives.
With that fundamental flaw and the use of dynamic speakers "contra naturam" the rest is pointless. You are debating details of putting the horse behind the cart with someone who is looking what it takes to put the horse where it belongs.
Second, your logic lacks any demonstration of the leaps and bounts it takes. In other words, you do not demonstrate logically that A follows from B and C follows from B, you claim so and incorrectly.
Third, my comments are not based in this context you use, but in one where Speakers are driven correctly, in accordance with the nature of the device. Moreover, I have experience with such systems as well as with speakers where the fundamental resonance is damped by mechanical means, so at least I know what I am talking about.
Now to some specific points:
Just because you measure damping does not mean it is based on friction in the suspension. There is no such direct link. All you can do is to measure Qm and state that the system contains X mechanical damping with Y Linearity. You cannot in fact determine from the measurement how this damping is applied.
No. I mean the way in which it has been applied in german Studio monitor Drivers since the mid 1930's up to 1992, when the final inheritor of the Eckmiller legacy in Berlin closed doors.
So, what is the level of non-linearity introduced by a flow resistance. Hint, in east Germany we used flow resistances also for making cardioid Speaker columns that where cardioids down to around 150Hz, so I know how non-linear they are...
Qes is a very poor way to not "control" the speaker. Plus, if you think Qms is nonlinear, try measuring Qes nonlinearity.
What Qe attempts and miserably fails is to correct a mechanical problem by simple electrical means. In the process it actually causes by far more gross distortion all across the bandwidth of the speaker, while still failing to critically and consistently damp the cone at all signal levels and frequencies.
The only problem is that your electronics cannot control cone movement through Qe and low source impedance.
Cone acceleration (and thus audio output) outside the frequency range around resonance is proportional strictly and only to voice coil current and magnet field/suspension linearity.
So Qe is not in the picture, only Re. And Re can be taken out of the picture (together with Qe) by current drive.
Now, lets look at the fundamental resonance frequency region (which is quite narrow in range). Does Qe control cone movement directly and in a linear fashion? Nope, it is in fact heavily signal envelope dependent and modulated by excursion (where usually larger excursions decrease damping).
Contrast this with damping by a flow resistance integrated into the basket, it has neither appreciable stiction nor non-linearity with level and driving the speaker with current.
In one fell swoop we have eliminated:
1) Thermal Compression (which is hotly debated but real neverless)
2) Eddy Current (in the pole piece) induced distortion (which is odd order and cubic law and leads to often observed preponderance of H3 over H3 in the midrange for many drivers)
3) Low Frequency non-linearity, which is caused by suspensions and electrical systems that are highly non-linear.
If we combine this with a magnet system that has a constant field strength for the range of cone travel of interest we will have made a Speaker system that has a dramatically improved performance in almost all relevant domains.
Of course, this system would be incompatible with common amplifiers, however, unlike the situation three or more decades ago when amplifiers and speakers where separate, active speakers are becoming more common.
However this debate about how to optimise the Amplifier/Speaker system using unconventional, if superior means is already present in several threads on that topic here. including this more recent one:
http://www.diyaudio.com/forums/vendors-bazaar/157787-secret-tube-amplifiers-revealed-much-more.html
So no need to drag it in here, where generally conventional systems are debated, or to then take my comments on Driver Parameters in threads discussing current drive out of context to make it seem as if there is a contradiction and to them complain when I address not your flawed logic, but deal with the individual contentions you have raised.
BTW, my noted to the hysteresis effects in low efficiency systems in this thread here referred mainly to the system behaviour outside resonance region and to the results of attempting to damp cone resonance with the rubber surround...
Ciao T
Funny, you are one of many who think so, including me. It's odd that so many people should agree on a point, yet the industry remains deaf and dumb, probably wondering why is it going down.
I would imagine that in real life, all these space technologies in fact cost less than traditional paper cone, although we are made to think exactly the opposite. Wouldn't be the first time.
I still remember only too well my ancient AR5 speakers - their bass quality, with their 10" woofers, was honestly second to none in those days. In all honesty, I can claim that my present, aerogel based 10" woofer is a bit better, but that bit is way too small for the 40 years of development in between.
In most cases, I still prefer a doped paper cone to all these fancy materials.
Hi,
In fact, another datapoint. Many of the really ultra-cheap simple paper cones, with fairly thin uncoated paper and significant measurable breakup nevertheless sound surprisingly natural (including my "open air" test), while a similar plastic cone drivers, despite having much superior response flatness and all sounds like a duck quacking and very obviously and unnaturally coloured!?
Incidentally the (justly) much maligned Yamaha NS-10 also used a paper cone, probably (together with fairly low distortion and good impulse response) the only saving graces...
Then again, the Speakers I listen to now have drivers that I had some small design input into (except treble which is OTS) and are surprisingly high tech/space age for such an old reprobate as myself. The actual design was done by someone rather justly famous for his work with metal cone speakers, but possibly a bit too grounded in the traditional approaches.
The woofers have a low loss "bicycle tire" surround, a rather stiff spider a partial sandwich cone with a single piece ceramic oxide coated alu inverted dome covered in the centre 2/3rd of the cone diameter with carbon fibre reinforced paper cone layered and the pure alu/oxide for the outer parts. it's pretty much flat to 1KHz before it rolls off smoothly. They are more Sub's than woofers but do woofer duties quite well.
My main contribution was to ask for a vented overhanging pole piece, extra venting around the voice coil (as much as possible), a big voice coil diameter (I normally prefer 3 -4" for woofers, but can accept 2", barely) and a low Mms/BL ratio (<5), plus extended higher frequencies...
The midrange are actually 50Hz-15KHz 5"Diameter fullrange drivers with a very lathe voicecoil, neodymium magnet and alu/oxide cone with seperate alu/oxide dome.
The driver designer had never before designed full range drivers and early samples sounded promising in terms of dynamic range and bass output (for a small driver) but had a rather nasty and edgy tone to them making them u nsellable as "Full Range".
By shifting the unavoidable breakup resonances around (this was for example a case in Audio where I just pointed and said "change this and that in this way" and it was much better afterwards) the driver ended up sounding very natural, if a touch polite (I dislike stuff that jumps up to bile my ears off). Waterfall still looks quite hashy at higher frequencies (> ~5Khz) but does not sound the way it looks.
These drivers where quite expensive, more expensive than paper cones and are either sold out and will not be manufactured again or will not be restocked once sold out...
Given the use in a speaker that uses first order electrical crossovers (series, no conjugates or EQ used) the characteristics of the drivers to all be able to run "crossoverless" without sounding bad make a major contribution to the end result. Most more common "HiFi" drivers would have made quite bad sound using such a design... ;-)
I just wish for 6dB or so more efficiency, over the 89dB/1W/1m I alread have.
The issue is mostly one of consistency. If you make cheap stamped frame < 1 USD per pc in 1KU 4" Drivers for cheap speakers you can accept quite a high level of tolerance (wide tolerance).
But if you are making 50 USD per pc in 1KU "high end" 6.5" Driver you need quite tight tolerances, so you end up rejecting more cones than can be put into speakers and still have 30%+ rejection rate on the final drivers. If you instead use some "High Tech" material with great claimed cost and benefits (all of which are BS of course, sorry, marketing I meant to say) you probably get a 100% acceptance rate of the cones you buy in and > 90% on final drivers.
So in the end the single paper cone as such may be cheaper than the space age cone, HOWEVER, the final accumulated cost per Driver is often much higher than for artificial materials.
Amen.
And often even an uncoated paper cones and non-lossy surround (corrugated cloth) at that...
Ciao T
In fact, another datapoint. Many of the really ultra-cheap simple paper cones, with fairly thin uncoated paper and significant measurable breakup nevertheless sound surprisingly natural (including my "open air" test), while a similar plastic cone drivers, despite having much superior response flatness and all sounds like a duck quacking and very obviously and unnaturally coloured!?
Incidentally the (justly) much maligned Yamaha NS-10 also used a paper cone, probably (together with fairly low distortion and good impulse response) the only saving graces...
Then again, the Speakers I listen to now have drivers that I had some small design input into (except treble which is OTS) and are surprisingly high tech/space age for such an old reprobate as myself. The actual design was done by someone rather justly famous for his work with metal cone speakers, but possibly a bit too grounded in the traditional approaches.
The woofers have a low loss "bicycle tire" surround, a rather stiff spider a partial sandwich cone with a single piece ceramic oxide coated alu inverted dome covered in the centre 2/3rd of the cone diameter with carbon fibre reinforced paper cone layered and the pure alu/oxide for the outer parts. it's pretty much flat to 1KHz before it rolls off smoothly. They are more Sub's than woofers but do woofer duties quite well.
My main contribution was to ask for a vented overhanging pole piece, extra venting around the voice coil (as much as possible), a big voice coil diameter (I normally prefer 3 -4" for woofers, but can accept 2", barely) and a low Mms/BL ratio (<5), plus extended higher frequencies...
The midrange are actually 50Hz-15KHz 5"Diameter fullrange drivers with a very lathe voicecoil, neodymium magnet and alu/oxide cone with seperate alu/oxide dome.
The driver designer had never before designed full range drivers and early samples sounded promising in terms of dynamic range and bass output (for a small driver) but had a rather nasty and edgy tone to them making them u nsellable as "Full Range".
By shifting the unavoidable breakup resonances around (this was for example a case in Audio where I just pointed and said "change this and that in this way" and it was much better afterwards) the driver ended up sounding very natural, if a touch polite (I dislike stuff that jumps up to bile my ears off). Waterfall still looks quite hashy at higher frequencies (> ~5Khz) but does not sound the way it looks.
These drivers where quite expensive, more expensive than paper cones and are either sold out and will not be manufactured again or will not be restocked once sold out...
Given the use in a speaker that uses first order electrical crossovers (series, no conjugates or EQ used) the characteristics of the drivers to all be able to run "crossoverless" without sounding bad make a major contribution to the end result. Most more common "HiFi" drivers would have made quite bad sound using such a design... ;-)
I just wish for 6dB or so more efficiency, over the 89dB/1W/1m I alread have.
The issue is mostly one of consistency. If you make cheap stamped frame < 1 USD per pc in 1KU 4" Drivers for cheap speakers you can accept quite a high level of tolerance (wide tolerance).
But if you are making 50 USD per pc in 1KU "high end" 6.5" Driver you need quite tight tolerances, so you end up rejecting more cones than can be put into speakers and still have 30%+ rejection rate on the final drivers. If you instead use some "High Tech" material with great claimed cost and benefits (all of which are BS of course, sorry, marketing I meant to say) you probably get a 100% acceptance rate of the cones you buy in and > 90% on final drivers.
So in the end the single paper cone as such may be cheaper than the space age cone, HOWEVER, the final accumulated cost per Driver is often much higher than for artificial materials.
Amen.
And often even an uncoated paper cones and non-lossy surround (corrugated cloth) at that...
Ciao T
Vifa has a relatively recent line of drivers out that uses paper cones in a very advanced magnet motor design, with excellent specs. They sound well too. Have unfortunately not had time to build with them, but for those who like paper cones, it would be a serious option to consider for a new building project.
vac
- Status
- Not open for further replies.