That is correct, yours is a different approach. What I suggested was looking at his method for the electronics portion up to the woofer. He didn't get a chance to patent the last servo used in the Trout but it was a full range unit that worked up to 20k.
Hi,
In #59 You wrote about possible problems of a Mic capsule as sensor. While it indeed reacts on 'external' sound sources, the membrane it will be connected to will dominate. A room mode may be one kind of an external source for the Mic, which the speaker would try to correct for also. B&M of Germany used Mics in their early 90s commercial speaker range successfully as acceleration sensor (mounted perpendicular to the axis of movement). An issue of Mics might rather be their transfer function which may be more complicated than that of an acceleration sensor. This might ask for some kind of compensation or lead to a reduced bandwidth.
Powersoft from Italy offer a system which looks simiilar to Yours at first glance. A high powered amplifier -class D in this case- together with a low latency DSP. They differ in that they use a differential pressure sensor which is to be mounted into the casings wall, thereby sensing/comparing the air pressure inside and outside of the casing. Their technique has been patented tgough and is named IPAL.
jauu
Calvin
In #59 You wrote about possible problems of a Mic capsule as sensor. While it indeed reacts on 'external' sound sources, the membrane it will be connected to will dominate. A room mode may be one kind of an external source for the Mic, which the speaker would try to correct for also. B&M of Germany used Mics in their early 90s commercial speaker range successfully as acceleration sensor (mounted perpendicular to the axis of movement). An issue of Mics might rather be their transfer function which may be more complicated than that of an acceleration sensor. This might ask for some kind of compensation or lead to a reduced bandwidth.
Powersoft from Italy offer a system which looks simiilar to Yours at first glance. A high powered amplifier -class D in this case- together with a low latency DSP. They differ in that they use a differential pressure sensor which is to be mounted into the casings wall, thereby sensing/comparing the air pressure inside and outside of the casing. Their technique has been patented tgough and is named IPAL.
jauu
Calvin
About the mystery of how MF systems react to room acoustics, I can't see how an accelerometer sensor would differ from a mic sensor or VC-bridge system, except as constrained by their feedback-loop parameters. (VC-bridge system works to the edge of DC.)
MF turns the familiar floppy cone into something that feels more like a solid wall (albeit one made from cardboard): when you push on a MF cone, it feels solid and resists moving (assuming the feedback-loop works down to very low frequencies). Very weird to experience but should be intellectually self-evident if you understand feedback. The driver "eats" the room acoustics and it can't distinguish ambient impingements from any other kind of driver "error" and I am not sure I see any reason why you'd want to bother except for some kind of misplaced purist urge. (You can possibly avoid "confusing" the MF system with room acoustic impingements but then you have to get into the sort of hyper-complexity Calvin describes in the previous post).
Mic-borne MF seems a non-starter. A mic sensor at, say, 3 inches from the dust cap will lead your feedback into convulsions due to phase problems (different wave lengths for different frequencies). Like anything, there may be complex electronic corrections possible for this physical reality obstacle, but not a route that seems promising to take.
BTW, don't experiment with MF unless you are prepared to blow out a few expensive cones along the way.
Ben
MF turns the familiar floppy cone into something that feels more like a solid wall (albeit one made from cardboard): when you push on a MF cone, it feels solid and resists moving (assuming the feedback-loop works down to very low frequencies). Very weird to experience but should be intellectually self-evident if you understand feedback. The driver "eats" the room acoustics and it can't distinguish ambient impingements from any other kind of driver "error" and I am not sure I see any reason why you'd want to bother except for some kind of misplaced purist urge. (You can possibly avoid "confusing" the MF system with room acoustic impingements but then you have to get into the sort of hyper-complexity Calvin describes in the previous post).
Mic-borne MF seems a non-starter. A mic sensor at, say, 3 inches from the dust cap will lead your feedback into convulsions due to phase problems (different wave lengths for different frequencies). Like anything, there may be complex electronic corrections possible for this physical reality obstacle, but not a route that seems promising to take.
BTW, don't experiment with MF unless you are prepared to blow out a few expensive cones along the way.
Ben
Last edited:
When I was experimenting with the pressure feedback system, I developed a low distortion high SPL pressure transducer capable of responding down to 1Hz at 160 dB which is the sort of capability needed when measuring the sound pressure close to the diaphragm. Otherwise the equivalent microphone capable of this performance would be in the 1000's of dollars which is just not economical.
Even with that I only had to blow air around the pickup and the speaker would pump wildly as it tried to correct for the external pressure disturbance. There were just to many problems so I decided to abandon it in favour of the accelerometer approach which is the traditional and most reliable way of doing it.
As for room correction, this system would attempt to only correct at the point it was located and would require power to do it. Also I'm not sure if there would be a net benefit with this type of system since the acoustic impedance at the speaker would not match the characteristic impedance of the free air and thus there would still be reflections at the room to speaker boundary. It's better to correct for room acoustics first using acoustic treatments followed by electronic means.
The IPAL system seems to be an enhancement to the Stahl ACE BASS system with some improvements in the way it compensates for the environment. However like the original Stahl concept I don't believe there is much in the way of improvement in distortion and I can't see how it compensates for changes in the voice coil resistance and inductance.
regards
david
Last edited:
A wave that travels 3 inches to a mic is different in phase from one that travels 2 inches and an 80 Hz wave at 3 inches is different in phase (and hence in feedback) than a 100 Hz wave at 3 inches.
A mic at say 6 inches might be feeding back negative feedback at one frequency and positive feedback at another. Think you'd like that?
Ben
Thanks for great experimental information.
Like it or not, the driver does feed back room information that hits it (within its passband, of course)... like any other error signal.
There is power to do it because all MF systems are active systems. Depends on the feedback parameters.
When you were blowing on the mic, you were presenting monumental tsunami forces rather than the usual room level (unless a fan was on). No wonder the cone went wild.
It is very, very challenging to make MF work. You are working at the very edge of control of sensors, phases, outside influences (like a fan or bumping your speaker box...). Easy for store-bought systems to self-destruct - which is why there are few and, I bet, those who had any kind of commercial lifetime had very low feedback fractions. As far as I know, nobody ever successfully commercialized a generic black-box that could be used with different drivers.
Ben
Last edited:
I hope to change that all for everyone. 
Thousands of hours of design, simulation and testing has gone into this project
I am achieving 30 dB loop gain over a reasonable bandwidth on a crappy speaker without any instability using an accelerometer based servo system. With nested feedback I have achieved a peak distortion reduction of 43dB using the same speaker. This is unprecedented with a servo controlled speaker system !
Even when the amp is driven into clipping the system remains stable which is an issue with high gain servo feedback on a speaker system. Also some other accelerometer based servo systems have a problem in that when the speaker is moved the servo tries to maintain an inertial frame of reference and attempts to force the speaker cone in and out rather violently depending on the motion of the speaker cabinet. I'm happy to say that my system does not suffer from this problem and the speaker cone is rock steady when the speaker is moved much to the disappointment of one vendor that claims that this a real problem with accelerometer based servo systems
However, having said this there is still a lot more work to do to get it to a finished state but it's getting there.
regards
david
Thousands of hours of design, simulation and testing has gone into this project I am achieving 30 dB loop gain over a reasonable bandwidth on a crappy speaker without any instability using an accelerometer based servo system. With nested feedback I have achieved a peak distortion reduction of 43dB using the same speaker. This is unprecedented with a servo controlled speaker system !
Even when the amp is driven into clipping the system remains stable which is an issue with high gain servo feedback on a speaker system. Also some other accelerometer based servo systems have a problem in that when the speaker is moved the servo tries to maintain an inertial frame of reference and attempts to force the speaker cone in and out rather violently depending on the motion of the speaker cabinet. I'm happy to say that my system does not suffer from this problem and the speaker cone is rock steady when the speaker is moved much to the disappointment of one vendor that claims that this a real problem with accelerometer based servo systems
However, having said this there is still a lot more work to do to get it to a finished state but it's getting there.
regards
david
Last edited:
Whilst it might work at low frequencies the delay from the pickup to the speaker as well as the resonances inside the cabinet at high frequencies would probably make it impractical.
regards
david
Yet he was most famous for a flat response in a small enclosure. His speakers were called the Ferrari of their time.
Again, I only suggested that his work be reviewed since it is my "opinion" that it may help David in his work.
Mr Eraths original work is long out of patent and he didn't get to patent his latest which was significantly improved due to newer technology.
For BEN- No one addressed you. And I made a suggestion to the thread starter not you. If you don't know history that is your failure.
If David chooses to disregard my suggestion that is his right. As it is my right to be able to make that suggestion.
All-
Sorry for the off topic post to Ben. Some people have no life and look for conflict.
Ben, FWIW, Erath speakers were well-known in the 1960s, but due to the enormous size and cost, were more heard of than heard, legendary in the literal sense. I first read about them in the late, lamented Electronics Illustrated; I was in my teens at the time and seeing a picture of them, immediately fell in lust. I think there were also extensive write-ups in Popular Electronics. Never got a chance to hear them.
Like Bozak, it's a name that sort of disappeared with time.
Like Bozak, it's a name that sort of disappeared with time.
Thanks but I have seen a lot of his patents as you do when you research the subject but there is nothing there that I can use. I have approached the whole thing from a completely different perspective and developed my own system which is different from all of these other things.
regards
david
OK, is there anything about active motional feedback of any sort in Erath's work, patent, or products that anybody here can relate to us? Unlikely he used MF in the mega-driver big boxes, but did he use MF in the small boxes?
Sy, as much as I have the highest respect for your thoughts, I don't think I would put a recognized early hero of HiFi like Bozak in the same class as LWE. But you are right that both are mostly forgotten.
There has been an energetic body of study of MF over the last 50 years. Alas, all the miracle breakthroughs, like in this thread, are with very modest drivers - MF always helps them a lot.
The next step is to apply MF to good drivers and see the results. And then to let take the R&D out of your garage. My experimenting at Bell Labs was with drivers that were reasonably good for their time and my data was satisfying for my purposes, at least with the crude pulses and tone bursts of 1967 and profoundly obvious when A-B'd by ear.
Ben
Sy, as much as I have the highest respect for your thoughts, I don't think I would put a recognized early hero of HiFi like Bozak in the same class as LWE. But you are right that both are mostly forgotten.
There has been an energetic body of study of MF over the last 50 years. Alas, all the miracle breakthroughs, like in this thread, are with very modest drivers - MF always helps them a lot.
The next step is to apply MF to good drivers and see the results. And then to let take the R&D out of your garage. My experimenting at Bell Labs was with drivers that were reasonably good for their time and my data was satisfying for my purposes, at least with the crude pulses and tone bursts of 1967 and profoundly obvious when A-B'd by ear.
Ben
Last edited:
Hi,
regarding acceleration sensors.
Freescale offers some models which might be interesting as well as lowcost.
The Micro machined MMA1200-series (+-250g, 2%FSO nonlinearity, 110µV/Hz^1/2) and the MMA2300-series (+-250g, 1%FSO nonlinearity, 110µV/Hz^1/2). Both coming in a SOIC16 casing and low output impedance, rendering the requirement for special cables obsolete. Listed at Digikey.
And how about a digital output sensor?
Freescale offers the Xtrinsic MMA26xx-series with DSI2.5 compatible digital output (Welcome to Freescale - Freescale Semiconductor).
Piezoresistive sensors as well as Sensors after IEPE (with AC-coupled voltage output) may suffice.
See 7264 and 7264B4 and 2256A, 2250A of Endevco-Meggit (www.endevco.com) or the KS91B and KS94B, KS94L of MMF (Beschleunigungsaufnehmer und Schwingungsmesstechnik von Metra)
jauu
Calvin
regarding acceleration sensors.
Freescale offers some models which might be interesting as well as lowcost.
The Micro machined MMA1200-series (+-250g, 2%FSO nonlinearity, 110µV/Hz^1/2) and the MMA2300-series (+-250g, 1%FSO nonlinearity, 110µV/Hz^1/2). Both coming in a SOIC16 casing and low output impedance, rendering the requirement for special cables obsolete. Listed at Digikey.
And how about a digital output sensor?
Freescale offers the Xtrinsic MMA26xx-series with DSI2.5 compatible digital output (Welcome to Freescale - Freescale Semiconductor).
Piezoresistive sensors as well as Sensors after IEPE (with AC-coupled voltage output) may suffice.
See 7264 and 7264B4 and 2256A, 2250A of Endevco-Meggit (www.endevco.com) or the KS91B and KS94B, KS94L of MMF (Beschleunigungsaufnehmer und Schwingungsmesstechnik von Metra)
jauu
Calvin
Hi,
The mentioned acceleration sensors are all > +-250g, as I think that +-150g of the ACH1-03 could be marginal or too low for longthrow drivers.
The peak acceleration value may be easily calculated as â = 4*pi²*f²*Xmax.
With f = chosen upper bandwidth limit in Hz (or 1/s), for example 100Hz for a Subwoofer and Xmax = maximum excursion in m, for example 20mmp-p = 10mm = 0.01m.
This would calculate to 3948m/s² or 402g!
The 150g limit of the ACH01-03 would be reached for 20mmp-p excursion at 61Hz. Could be a bit low?
Or is the overhead of the ACH01-03 large enough (accepting increased nonlinearity)?
David, have You made provisions to implement other acceleration sensors with greater g-range?
jauu
Calvin
The mentioned acceleration sensors are all > +-250g, as I think that +-150g of the ACH1-03 could be marginal or too low for longthrow drivers.
The peak acceleration value may be easily calculated as â = 4*pi²*f²*Xmax.
With f = chosen upper bandwidth limit in Hz (or 1/s), for example 100Hz for a Subwoofer and Xmax = maximum excursion in m, for example 20mmp-p = 10mm = 0.01m.
This would calculate to 3948m/s² or 402g!
The 150g limit of the ACH01-03 would be reached for 20mmp-p excursion at 61Hz. Could be a bit low?
Or is the overhead of the ACH01-03 large enough (accepting increased nonlinearity)?
David, have You made provisions to implement other acceleration sensors with greater g-range?
jauu
Calvin
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.