Do you honestly think this has any relevance to an economically feasible solution?
Harking back to wines for a sec, this is for Scott ... http://www.wsetglobal.com/documents/wineaustralia2012.pdf, 
.
.For a test or a product? For a test, since almost all the parts are available - yes. For a product? I dunno. why don't you ask the hard drive and wafer fab guys.
I'll spring for the transamp
I'm reminded of Back to the Future on the erbium, when Dr. Brown says "I'm sure that in 1985, plutonium is available in every corner drug store, but in 1955 it's a little hard to come by".
Bionic ear could harness brain's 'octopus cells' to improve sound › News in Science (ABC Science)
Would it be accurate to say that certain stimuli cause the hearing system to become very sensitive at certain points in time?
Would it be accurate to say that certain stimuli cause the hearing system to become very sensitive at certain points in time?
There are speaker drivers with an addition coil on the former, providing velocity feedback for control. There are also drivers with an extra static coil in the gap which can be used to sense and control changes in the gap magnetic field - I've heard some of these and they are extremely impressive.
Control is all about cone velocity and acceleration, knowing the position doesn't help as much as knowing how the cone is moving versus how it is commanded.
Force on VC is simply proportional to current. How the cone then moves depends on mechanical impedance, a combination of mass/spring/damping provided by both the suspension and the enclosure. If it were not for mech imp, in response to a constant applied voltage, the cone would simply accelerate indefinitely. The motion of the cone in terms of position is not simply the applied programme waveform !
Control is all about cone velocity and acceleration, knowing the position doesn't help as much as knowing how the cone is moving versus how it is commanded.
Force on VC is simply proportional to current. How the cone then moves depends on mechanical impedance, a combination of mass/spring/damping provided by both the suspension and the enclosure. If it were not for mech imp, in response to a constant applied voltage, the cone would simply accelerate indefinitely. The motion of the cone in terms of position is not simply the applied programme waveform !
I'll spring for the transamp
I'm reminded of Back to the Future on the erbium, when Dr. Brown says "I'm sure that in 1985, plutonium is available in every corner drug store, but in 1955 it's a little hard to come by".
I'll add some ironic humor. I saw a movie about guys trying to make nukes
and they're smuggling plutonium around the world, etc. Russian gangsters
get involved and hijack the load. Cut to the final scene the two Russians are
congratulating themselves and the lavish lifestyles they will lead while they
lay a small vile of plutonium out on the table and start snorting it like coke.
I really don't know what you would do once you have the relevant information about speed/ acceleration or position of the voice coil.
1) on the low end, motional feedback can be used like Philips did to go extend the frequency response. However, the low end roll-off conforms to well understood physical laws and can be corrected by for example a Linkwitz transform in sealed enclosure. No need for real time measurements of cone behavior here.
2) If it is to correct for thermal compression, I don't really think you would want to do that in abundance. The increase of resistance of the VC as a result of heating is a protection mechanism. Thermal runaway will rear its ugly head when you try to compensate for it.
3) This would imo only leave distortion reduction as a possible reason to measure what the voice coil is doing in real time. First problem is like Lucky sketched in #70151. Stability will be a bitch. Second is that the most distortion products will be outside the usable frequency band of the driver. This means that the driver will be too slow to react to correction signals for most of the distortion products and especially for the more audible higher order distortions.
1) on the low end, motional feedback can be used like Philips did to go extend the frequency response. However, the low end roll-off conforms to well understood physical laws and can be corrected by for example a Linkwitz transform in sealed enclosure. No need for real time measurements of cone behavior here.
2) If it is to correct for thermal compression, I don't really think you would want to do that in abundance. The increase of resistance of the VC as a result of heating is a protection mechanism. Thermal runaway will rear its ugly head when you try to compensate for it.
3) This would imo only leave distortion reduction as a possible reason to measure what the voice coil is doing in real time. First problem is like Lucky sketched in #70151. Stability will be a bitch. Second is that the most distortion products will be outside the usable frequency band of the driver. This means that the driver will be too slow to react to correction signals for most of the distortion products and especially for the more audible higher order distortions.
The problem, omho, is some can always turn around a patent by modifying one of its aspects.
Around 1970, Clement had deposed a patent for its servo-ed radial tangential strait arm turntable. It was all about moving the arm. So I filed a patent for the company I was working for where the arm was mounted on a fixed point and the plate moving under-it.
Gego (France) produced a servoed enclosure, before 1968, where the speaker had its normal voice coil in the back and a one with a little magnet assembly in the front of the cone. I made one at this time, on my side, with this speaker.
Actually, this is a standard technique used for interferometers used in spectroscopy, where accurate knowledge and control of position and velocity are critical, so the problems were worked out decades ago for low frequencies (<1kHz). Look inside any FTIR spectrometer.
Barleywater is my newly appointed hero and he should be yours too.
http://www.diyaudio.com/forums/mult...m-monopole-dipole-speakers-2.html#post4361879
In a leisurely afternoon, I measured a small two way system I have with HolmImpulse, exported the impulse response into rePhase, turned back the phase shift above 20 Hz, and exported the resulting filter kernel into JRiver and Windows Media Player.
So, for the price of some free software and a bit of cable wear, I turned a fine set of speakers from minimum into linear phase.
Some results: latency is .34 seconds. This makes synchronization with video sources mandatory. It also means it is compromised for live monitoring.
Bass appears unchanged, but in the midrange there is a bit more immediacy and delineation. It is like sharpening in Photoshop, and makes the sound different enough to not need the X in AB-ing.
As to pre-ringing, no violations of causality have come to my attention.
http://www.diyaudio.com/forums/mult...m-monopole-dipole-speakers-2.html#post4361879
In a leisurely afternoon, I measured a small two way system I have with HolmImpulse, exported the impulse response into rePhase, turned back the phase shift above 20 Hz, and exported the resulting filter kernel into JRiver and Windows Media Player.
So, for the price of some free software and a bit of cable wear, I turned a fine set of speakers from minimum into linear phase.
Some results: latency is .34 seconds. This makes synchronization with video sources mandatory. It also means it is compromised for live monitoring.
Bass appears unchanged, but in the midrange there is a bit more immediacy and delineation. It is like sharpening in Photoshop, and makes the sound different enough to not need the X in AB-ing.
As to pre-ringing, no violations of causality have come to my attention.
As far as I could figure out accurate knowledge of the position is critical View attachment GAC2.pdf
Not a word about using this knowledge to control position and velocity in a feedback loop. I don´t see why this would be necessary, since it is the knowledge of the position which is critical for the measurement, not how it came to that position.Velocity variations are irrelevant as long as position can be correlated to measurements. So, close, but no cigar.
You might want to actually look into the spectrometers and look at papers from specialists in that area (these are guys who are rock specialists, not FTIR designers). My first project when I worked at Nicolet was specifically directed at velocity control of the interferometer, since we were designing a top-of-the-line unit that could measure dynamic phenomena (it was commercially released as the 60SX, and may still be the best FTIR ever made).
why position?
Consider a transducer from which we are attempting to extract sound covering a wide frequency range that is being driven in whatever conventional manner. At higher frequencies there is progressively less voice coil movement associated with a flat frequency response, at low frequencies more. Suppose the signal consists of a superposition of two tones, one high, one low, for example. As the VC moves to reproduce the low frequency tone, it moves in accordance with some law (ignore for the moment the changes in resistance from self-heating). For an excursion outward or inward, depending on the suspension, volume of air, motor characteristics, at any point within that excursion the high frequency sensitivity varies in accord with a law---ideally this would be a constant, but it never is. As a result, driving open loop we have IM distortion based on the superposition of the two tones. If we can determine the law(s) we can compensate by knowing the instantaneous position and adjusting the drive to predistort in a compensatory way. As part of that we can get velocity and acceleration just by successive time differentiation of the positional signal, just as some systems do this in reverse by successive integration of acceleration signals.
Note that a good position transduction technique will be immune to suspension changes, vc temp, motor nonidealities of other sorts, etc. However the sensitivity law applied to high frequencies in particular may well be affected, so strategies beyond just the use of the positional data will be needed. Current drive may help, real-time determination of VC resistance, etc. Protection from VC overheating is needed and readily implemented.
With a lot of signal processing and horsing around thus we can reduce nonlinear distortions, both harmonic and IM (and even the closely-related "Doppler") but as pointed out this doesn't do everything, especially if the cone is itself non-pistonic. And none of this will compensate for a poor directivity index for the overall loudspeaker.
I got interested in this circa 1995 because of Clayton Williamson's early designs of teeny transducers for laptops and small desktop powered loudspeakers (those ubiquitous concave aluminum dome ones for example, which Harman named Odysseys), which tend to have a very large maximum excursion relative to their diameters, and suffer from a whole lot of IM distortion, while producing remarkable amounts of low frequency energy for their size, and covering a rather large range of frequencies. One could push the technique to larger diameters using some high-performance cones, but clearly at some point it's time to move to two-way and three-way systems.
Geddes and others have studied the audibility of nonlinear distortions, and there is undoubtedly more to be done. I will say though that the effect of reproducing midbass simultaneously with pitched transient sources like piano can be very audible indeed and quite unpleasant.
Consider a transducer from which we are attempting to extract sound covering a wide frequency range that is being driven in whatever conventional manner. At higher frequencies there is progressively less voice coil movement associated with a flat frequency response, at low frequencies more. Suppose the signal consists of a superposition of two tones, one high, one low, for example. As the VC moves to reproduce the low frequency tone, it moves in accordance with some law (ignore for the moment the changes in resistance from self-heating). For an excursion outward or inward, depending on the suspension, volume of air, motor characteristics, at any point within that excursion the high frequency sensitivity varies in accord with a law---ideally this would be a constant, but it never is. As a result, driving open loop we have IM distortion based on the superposition of the two tones. If we can determine the law(s) we can compensate by knowing the instantaneous position and adjusting the drive to predistort in a compensatory way. As part of that we can get velocity and acceleration just by successive time differentiation of the positional signal, just as some systems do this in reverse by successive integration of acceleration signals.
Note that a good position transduction technique will be immune to suspension changes, vc temp, motor nonidealities of other sorts, etc. However the sensitivity law applied to high frequencies in particular may well be affected, so strategies beyond just the use of the positional data will be needed. Current drive may help, real-time determination of VC resistance, etc. Protection from VC overheating is needed and readily implemented.
With a lot of signal processing and horsing around thus we can reduce nonlinear distortions, both harmonic and IM (and even the closely-related "Doppler") but as pointed out this doesn't do everything, especially if the cone is itself non-pistonic. And none of this will compensate for a poor directivity index for the overall loudspeaker.
I got interested in this circa 1995 because of Clayton Williamson's early designs of teeny transducers for laptops and small desktop powered loudspeakers (those ubiquitous concave aluminum dome ones for example, which Harman named Odysseys), which tend to have a very large maximum excursion relative to their diameters, and suffer from a whole lot of IM distortion, while producing remarkable amounts of low frequency energy for their size, and covering a rather large range of frequencies. One could push the technique to larger diameters using some high-performance cones, but clearly at some point it's time to move to two-way and three-way systems.
Geddes and others have studied the audibility of nonlinear distortions, and there is undoubtedly more to be done. I will say though that the effect of reproducing midbass simultaneously with pitched transient sources like piano can be very audible indeed and quite unpleasant.
- Status
- Not open for further replies.