The Manger Subsonice mentioned by dblab is actually a third-party product.
It looks as if it is based on an older Backes & Müller Patent.
Regards
Charles
It looks as if it is based on an older Backes & Müller Patent.
Regards
Charles
SSassen said:
Sorry, I've got nothing useful to add to this thread in the direction it's heading...
The idea you're talking about is called current feedback. But doesn't address all of creating a current source. One of the characteristics of a current source is a high source impedance.
Look up "transconductance" amplifiers for the kind of topology you are looking for. Basically, as typical voltage amplifiers have emitter-follower (or source-follower) output stages, a current amp is going to have some kind of Common Emitter/Drain or a current mirror kind of output stage.
--
Danny
forr said:
Thanks for the current feedback info Forr! I havent read anything on it,I think its about time 😀
Mike.e
An advantage of using a transconductance amp in a motional feedback arrangement is the cancelling of the lowpass pole that is generated by the voice-coil inductance. This makes stable feedback a bit easier to achieve.
Regards
Charles
Regards
Charles
A series notch filter in parallel to the voice coil tuned to the drivers fs might be a solution to flatten out the hump in the frequency response and will give some electrical dampening.
I dont want this thread to go to waste so here's another thought: use an amplifier that has a negative output resistance, this will effectively give the driver a very very strong motor.
Some sort of EQ will be neccesary to get a flat frequency response.
Some sort of EQ will be neccesary to get a flat frequency response.
I’m thinking about upgrade my sub with an accelerometer.
So today i did some measurements and I discovered that below that the reso freq. the acceleration is about the same as the input voltage.
But at the reso. freq and higher there’s always a phaseshift. That makes it impossible to use a voltage driven amp in the feedback system.
BUT, when I measure the current, the signal from the accelerometer almost matches the signal from the current.
my conclusion is that current driven amp can only be used when using a feedback and (very important) a circuit that prevents the woofer exceeding Xmax
So today i did some measurements and I discovered that below that the reso freq. the acceleration is about the same as the input voltage.
But at the reso. freq and higher there’s always a phaseshift. That makes it impossible to use a voltage driven amp in the feedback system.
BUT, when I measure the current, the signal from the accelerometer almost matches the signal from the current.
my conclusion is that current driven amp can only be used when using a feedback and (very important) a circuit that prevents the woofer exceeding Xmax
mike.e said:
This is a must read patent for servo or any feedback system. However, I do think negative output resistance approach has its own drawbacks. First, it makes the system even more suseptible to voice coil resistance change (such as thermal compression, or thermal memory effect). For instance, let us assume there is an 8ohm driver with 50% thermal variation. That is it can vary from 8ohm to 12ohm after heat-up. Let us say we use -6ohms to bring it down to 2ohm. If that -6ohm does not adjust, when the voice coil is at 12ohm, the net resistance is 12-6=6ohm and that is a 200% variation in the feedback system now (vs 50%). It is the pivotal effect. Second, because of this sensitivity issue, we often need to reserve some margin for any cancellation schemes to avoid going overboard under any operation condition, there is a limit on how much you can reduce. Third, keep in mind negative resistance is achieved by positive feedback. Any negative issues with positive feedback apply here.
Our patented sensing coil method completely elliminate the above issues. There is no positive feedback. In the close form analysis, the voice coil resistance is not even in the equation.
I do sell servo subs based on my patent and therefore my view can be biased.
Brian
Rythmik Audio
Re: Why Feedback?
This is a good paper in its own right (not just to the specific topic).. it is however no longer available except in cached form via servers like Google.
http://72.14.207.104/search?q=cache...celling+driver+nonlinearities+push+pull&hl=en[/url]
one of the interesting things it states is that a push-pull system theoretically reduces ODD-order distortion (not even-order as is commonly percieved AND stated in the Loudspeaker Cookbook.. of course Mr. Linkwitz has it stated correctly on his site whan last I checked).
Another interesting note (more on-topic) is with regard to negative impeadance amplifiers as "servo systems" are actually in use by Yamaha. Note what Rod has to say about such amplifiers here:
http://www.sound.westhost.com/project56.htm
and more importantly here:
http://www.sound.westhost.com/z-effects.htm
(and both discuss "current" amplifiers at low freq.s as well - note that thermal related distortion can be FAR worse for drivers higher in freq.)
Roysyboy said:
This is a good paper in its own right (not just to the specific topic).. it is however no longer available except in cached form via servers like Google.
http://72.14.207.104/search?q=cache...celling+driver+nonlinearities+push+pull&hl=en[/url]
one of the interesting things it states is that a push-pull system theoretically reduces ODD-order distortion (not even-order as is commonly percieved AND stated in the Loudspeaker Cookbook.. of course Mr. Linkwitz has it stated correctly on his site whan last I checked).
Another interesting note (more on-topic) is with regard to negative impeadance amplifiers as "servo systems" are actually in use by Yamaha. Note what Rod has to say about such amplifiers here:
http://www.sound.westhost.com/project56.htm
and more importantly here:
http://www.sound.westhost.com/z-effects.htm
(and both discuss "current" amplifiers at low freq.s as well - note that thermal related distortion can be FAR worse for drivers higher in freq.)
ScottG said:
Thanks for not blasting me in the first place. I don't post here often for the same reason. And when I do, I try to stay on the technical aspect.
Distortion is one of the most complicated issues in transducers. If we compare the distortions on two subs operating on different principles, we normally don't see a clear winner. Case and point, one of the major source of 2nd order distortion is flux modulation. The flux modulation is lowest at resonace peak, where it draws the least amount of current. So if we compare an IB with resonance at 20hz vs another sealed box with resonace at 40hz, at 20hz the IB is better and at 40hz, the sealed box is better. In other words, IB has a sweet spot in terms of distortion around 20hz (where motional impedance remains high) and sealed box has a sweet spot around 40hz. Which one is more desirable? In my opinion, the distortion generated by 40hz is more audible than the distortion generated by 20hz (because of human hearing sensitity curve). Another extreme case is the sub configuration used by BagEnd, which they push the resonace into 80hz, right at the cross point.
That being said, another interesting aspect of servo subs is the transfer function and in particular the Qtc value of the system. The effect of our servo feedback is it cuts the Qtc by a factor of 3x. And that results in 3x improvement in how fast it dissipates the energy stored at the speaker components such as spider and surround. Take our sealed box servo configuration for example, without servo, the Qtc in a 2 cu ft box is 1.0. With servo feedback, the Qtc becomes 0.33. In general the lower the Qtc, the better control of the cone. To implement the same Qtc with conventional brute force approach, it will take 3x the magnet of what we currently use. Note that this Qtc is the intrinsic Qtc of the system. To achieve the overall bass extension of 14/20/28hz and Q=1.1/0.7/0.5, we put in a filter that implements this corner frequency roll-off charactersitic. This is very different from an LT equalized system which has a very high intrinsic Qtc (normally in the range of 1.0). In these systems, no matter how low we equalize the Q value to, the ability to control the cone remains the same as that without LT. In other words, it is not fair to compare transfer functions based on their face values, either. Remember the legendary transfer function challenge that Bob Carver took with Stereophile magzine? He had to iterally generate harmonic distortions and add output resistance to make his solid state "clone" amp matched the tube amp given by the magzine. Even though this is not going well with the idea of break-through or innovation (because who will come up with the prefect amp for him to clone?), but it did imply the transfer function is not a complete specification. In a pefectly linear world with every components being ideally linear, a transfer function can be a complete specification. But in real world, it is not.
Yet, another interesting aspect is the sensitivity analysis in the context of circuit analysis. Most people have completely ignored this important branch of circuit analysis. It's an indication of how robust a particular circuit system is. For instance, for a Qtc of 0.5 in IB, a 50% variation will bring the Qtc to 0.75, which is no longer a critically damped system. With the servo feedback, 50% variation of the Qtc brings it from 0.33 to 0.5 (which is still a critically damped system). You can imagine how this aspect applies to the Bob Carver's transfer function. The prototype unit and the units from the assembly line can sound differently.
To answer anther question of feedback vs feedforward system, let me play a devil's advocate as I have posted elsewhere, the worst scenario of feedforward system is it has both the natural distortion and corrective (or synthetic) distortion components in the system. The resulting system can sound differently. But is it worse or is it better? Do we actually hear the affect of the natural distortion being corrected? Or we actually hear both distortions? It is very difficult to answer based on how inconclusive we have come to understand the effect of distortion on sound quality.
To summarize, I don't think distortion by itself is the key to true life-like sound reproduction. I have heard several so-called low distortion subs and their sound can be best described as non-offensive, or bland. Instead, the key is a little bit of everthing: distortion, transfer function, intrinsic Q value, sensitivity to variation, plus a lot more I am still trying to find out.
Brian
Rythmik Audio
Re: Re: Why Feedback?
I have to disagree with what the paper said. First, intuitively, isn't the definition of odd or even harmonic distortion based on the multiplicative number of the frequency? In that case, one can think of fundamental as a special case of odd harmonics. That means the paper implied even the fundamental is elliminated. In this case, we know what they have done, they basically had the push and pull operation and didn't reverse the polarity of one of them (which was a mistake).
Second, let us look at symmetry and what the authors said about it:
The surround and spider suffer from similar problems. For both, although some finite resistance is acceptable and simply lowers efficiency, their resistance increases as the cone travels further from the rest position. Also, inconsistencies in construction lead to different resistances for forwards and backwards motion. Like the magnetic field’s inconsistency, the formerleads to odd-order harmonic distortion and the latter to even-order distortion.
So symmetry means odd order harmonics and asymmetry means even order harmonics (which are correct). However, if it is symmetrical, how does it matter if it is mounted inside out or the other way? The odd order harmonics are always in-phase and they are not going to be cancelled.
Brian
Rythmik Audio
ScottG said:
I have to disagree with what the paper said. First, intuitively, isn't the definition of odd or even harmonic distortion based on the multiplicative number of the frequency? In that case, one can think of fundamental as a special case of odd harmonics. That means the paper implied even the fundamental is elliminated. In this case, we know what they have done, they basically had the push and pull operation and didn't reverse the polarity of one of them (which was a mistake).
Second, let us look at symmetry and what the authors said about it:
The surround and spider suffer from similar problems. For both, although some finite resistance is acceptable and simply lowers efficiency, their resistance increases as the cone travels further from the rest position. Also, inconsistencies in construction lead to different resistances for forwards and backwards motion. Like the magnetic field’s inconsistency, the formerleads to odd-order harmonic distortion and the latter to even-order distortion.
So symmetry means odd order harmonics and asymmetry means even order harmonics (which are correct). However, if it is symmetrical, how does it matter if it is mounted inside out or the other way? The odd order harmonics are always in-phase and they are not going to be cancelled.
Brian
Rythmik Audio
rythmikaudio said:
..I try not to even imply harsh tones unless someone makes a personal attack..
of course if something is exceptionally obvious I can get a "little" sarcastic and become a bit of a 
Ah, you started to respond after I killed it! Of course I killed it because I started thinking about just how many variables a "system" would provide..
In any event I do think its important (*VERY*) because:
1. A comparison between reasonably like attributes is still meanigfull to a prospective purchaser (and the only comparison I've seen was with limited drivers you have compared - i.e. Adire, AND more importantly it lacks the comparitive value for reason 2 below).
2. Although our hearing is generally less sensetive to distortion as freq. decreases - it is however MORE important in the context of music as a reproduced event. (..and yes I realize this directly contradicts much of your premis above, but no insult is intended.) Why would I come to such a conclusion? Its not only my subjective response to a decrease in harmonic distortion at very low freq.s, but also the response of others as well (..those with me when "observing", and as evidenced in the "reviews" of a variety of publications, and the occasional responses of members here). (..so I'm pretty sure it isn't a mass halucination.)
As to why this occurs - well I believe it has to do with what such low freq. sound is actually providing for reproduced music. Most available recorded music simply does not have event related direct sound (vocal or instumentation) that extends lower then 40 Hz, (though occasionally extending to aproximatly 31 Hz). So what does that leave us with below this limit in a recording? Ambiant Noise.
Now ambiant noise at these lower freq.s has 2 distinct audible traits:
1. Freq. balance (or the subjective interaction of overall spl's between higher and lower freq.s)
2. Hall sound (or the reflected and "grouped" freq. specific spl's as they interact with the real or processed acoustic of the event).
I can't think of a significant reason why number 1 would be related - so that leaves us with Hall sound.
Hall sound (at very low freq.s) helps to generate the space of a real or virtual venue via low freq. grouping near boundries (again real or virtual). (i.e. bass "builds-up" near boundries like floors-to-walls.) Because it is not direct sound it is a distortion from an event perspective. Furthermore it is an "imaging" cue with regard to spatial localization (essentially defining the boundries/walls of a recorded venue which not only provides this aspect but also makes it easier for the listener to localize direct sound).
Now direct sound can be distorted harmonically (in a conssonate or dissonante fashion relating to the harmonics) - BUT for the purpose of imaging/localization such a distortion has little detrimental effect. This however does NOT seem to be the case with non-direct event sound. In otherwords a harmonic distortion of what is an event distortion is quite detrimental to that distortion with respect to imaging/localization. Subjectivly stated: an increase in harmonic distortion at very low freq.s makes the "sound space" more difficult to identify and additionally makes direct sound "image" localization more difficult. As to the reason? I can only think that it has its basis in the difference between our ability to process direct sound vs. non-direct sound (..the actual reason I'm sure is well beyond my understanding).
So the bottom-line is reducing harmonic distortion at very low freq.s is more important (in most systems) than reducing harmonic distortion (in a subwoofer) at higher freq.s - therefor information on this aspect of a driver is not just generally informative, but critical to selection - ESPECIALLY to DIY'ers creating subwoofers that are actually SUBwoofers. (and yes, I realize that the comercial reality is that most people constructing a so-called subwoofer are in fact looking to make a woofer system.. mores the pitty.)
Re: Re: Re: Why Feedback?
I could be wrong, but I think that the notion of phase relationship as a DIRECT factor towards a reduction in harmonics is wrong. In otherwords I don't think phase cancellation is the goal, (like in a push-pull amplifier - where I believe the notion of 2nd order "cancelation" arose), otherwise you would be correct that it simply reduces spl depending on the "nullity".
Instead I think that the reference is to "limiting" driver/VC motional behavior that results in a degree of cancelation.
Hmm, there used to be a member here that worked for KEF during their Isobarik days.. I wonder if he would "step-in" here to further explain the concept and its implementation w/ effect (..prob. too much to hope for).
rythmikaudio said:
I could be wrong, but I think that the notion of phase relationship as a DIRECT factor towards a reduction in harmonics is wrong. In otherwords I don't think phase cancellation is the goal, (like in a push-pull amplifier - where I believe the notion of 2nd order "cancelation" arose), otherwise you would be correct that it simply reduces spl depending on the "nullity".
Instead I think that the reference is to "limiting" driver/VC motional behavior that results in a degree of cancelation.
Hmm, there used to be a member here that worked for KEF during their Isobarik days.. I wonder if he would "step-in" here to further explain the concept and its implementation w/ effect (..prob. too much to hope for).
Re: Re: Re: Re: Why Feedback?
Stahl published a paper "Synthesis of Loudspeaker Mechanical Parameters by Electrical Means:..." in J. Audio Eng. Soc. Vol 29, No 9, 1981 Sept pp 587-596. Here is the exact quote from the paper pp 593:
Fig. 22 shows how second-harmonic distortion increases when the loudspeakers are mounted in a normal manner instead of the push-pull fashion. This would thus represent a good equalized bass-reflex system.
Of course "this" in the last sentence means the push-pull fashion.
In terms of the harmonic distortion comparision, I would agree its merit if the comparison is not done at one of the resonance frequencies. As a matter of fact, I already provided the number on 10hz (at lower SPL albeit). My concern is people misread the results (which already happened). One poster on the other thread think that implies my driver is only capable of 1/4" p-p excursion. However, my real intent is to demonstrate how it reduces the spider distortion which is the most prominent memory-related distortion. To do that I have to control the condition that other source of nonlinearity does not mask the results.
Brian
Rythmik Audio
ScottG said:
Stahl published a paper "Synthesis of Loudspeaker Mechanical Parameters by Electrical Means:..." in J. Audio Eng. Soc. Vol 29, No 9, 1981 Sept pp 587-596. Here is the exact quote from the paper pp 593:
Fig. 22 shows how second-harmonic distortion increases when the loudspeakers are mounted in a normal manner instead of the push-pull fashion. This would thus represent a good equalized bass-reflex system.
Of course "this" in the last sentence means the push-pull fashion.
In terms of the harmonic distortion comparision, I would agree its merit if the comparison is not done at one of the resonance frequencies. As a matter of fact, I already provided the number on 10hz (at lower SPL albeit). My concern is people misread the results (which already happened). One poster on the other thread think that implies my driver is only capable of 1/4" p-p excursion. However, my real intent is to demonstrate how it reduces the spider distortion which is the most prominent memory-related distortion. To do that I have to control the condition that other source of nonlinearity does not mask the results.
Brian
Rythmik Audio
I just had another though, if you use a microphone as an air pressure sensor on the front of the woofer, the side that faces the room, your should also have some correction for room modes.
I wonder how big the correction is, it could be negligible.
I wonder how big the correction is, it could be negligible.
Compare the DD18 to the other non servo subs !The servo mechanism appears very effective. Il have to study this topic further!
This is achieved with an accelerometer. The idea of using one of the DVcoils as a pickup has merit?(just like the wireless world articles)
http://www.guidetohometheater.com/features/1004way/index5.html
vs
This is achieved with an accelerometer. The idea of using one of the DVcoils as a pickup has merit?(just like the wireless world articles)
http://www.guidetohometheater.com/features/1004way/index5.html
An externally hosted image should be here but it was not working when we last tested it.
vs
An externally hosted image should be here but it was not working when we last tested it.
mike.e said:
The result certainly looks very good. On the other hand, I learned a lesson reading a review on one of the older Velodyne servo sub. Does anyone know if DD18 has the same limiter that other ULD servo subs have? That is, it automatically limits the frequency response when the excursion exceeds certain limit. The limiter is pretty tricky in the sense that it alters frequency response, instead of soft-clipping the waveform (so that they can claim the distortion is always less than..., no matter how hard you push it). At first, the roll off is at 20hz, then as the output comes close to excursion limit, it rolls off at 25hz, then 30hz, then 40hz. The person who did the test apparently didn't notice this and still report the distortion at the level he thought the output is (without knowing the output has been "linearly" attenuated, resulting in a very low distortion number). However, in the same review when plotting the frequency response at various output level, it clearly showed how the limiter changed the frequency response at various output levels, and reviewer must have overlooked the significance of that.
Ps: Oops. I didn't first check the link you have provided. Apparently you and I refered to the same article.
-Brian
Rythmik Audio
Hi Brian - This dynamic limiting seems very popular. I dont like the idea of it. It increases SPL by altering FR - sort of unethical to me!
mike.e said:Hi Brian - This dynamic limiting seems very popular. I dont like the idea of it. It increases SPL by altering FR - sort of unethical to me!
I agree with you. The way that velodyne implement it is essentially a form of AGC, or should we say automatic extension control. The objective is so that they can still claim low distortion. They achieved that goal, on paper (or what we should say, with static measurement). However, AGC has real world response time. When it begin to interfere with real life signal, it produces such an unnatural sound. I knew of a customer playing organ music on one of those subs. That AGC circuit did exactly that to him.
But, I think most importantly in this story is that "it showed us how easily the static distortion measurement can be fooled". You can imagine how bad the AGC will perform if the distortion measurement is based on a true dynamic signal such as music. I know I have been against any thoughts on digital feedforward techniques. Most of the implementation I have seen are still based on steady state distortion. In statistical term, that is what we call training sequence. Wait till the real sequences come. And there are zillions of them.
Brian
Rythmik Audio
- Status
- Not open for further replies.