This is a great thread!
I was impressed by some Philips MFB speakers back in the late 70s but had other things on my mind then - don't ask - then revisited the idea in the 90s, unfortunately a colleague convinced me the idea was dead due to improvements in speakers.
Well somehow I feel that in the meantime the speaker designers and the amplifier/electronic designers have gone their separate ways and there could be a great deal to gain through improvements in speaker design combined with the advent of improved electronics.
I was impressed by some Philips MFB speakers back in the late 70s but had other things on my mind then - don't ask - then revisited the idea in the 90s, unfortunately a colleague convinced me the idea was dead due to improvements in speakers.
Well somehow I feel that in the meantime the speaker designers and the amplifier/electronic designers have gone their separate ways and there could be a great deal to gain through improvements in speaker design combined with the advent of improved electronics.
I known that this principle is called negative impedance or current drive, but how can you take it, put filter and equalizer before the input signal and call it something else. The only important thing is the correction of unwanted cone movement.
Only if you take another source for the signal, like a second voice coil, piezo or optical device, it is something else.
The huge advantage of negative impedance is the simplicity, if you use a 0.1 Ohm resistor and an op-amp, you already have the high tech version.
On the other hand, a little epoxy glue and some sub 5$ parts might give you a much more advanced feed back today.
I really don´t stick to the “old system”, if piezo or electret sensors give much better results, I would rather go that way.
In the end there will be the same costs with any system: The uncorrected power demand of the driver, the additional power to force it to lower frequency response than it´s Qtb predicts (like a Linkwitz transformation) and finally the energy needed to correct the cone movement, as detected by a sensor.
Only if you take another source for the signal, like a second voice coil, piezo or optical device, it is something else.
The huge advantage of negative impedance is the simplicity, if you use a 0.1 Ohm resistor and an op-amp, you already have the high tech version.
On the other hand, a little epoxy glue and some sub 5$ parts might give you a much more advanced feed back today.
I really don´t stick to the “old system”, if piezo or electret sensors give much better results, I would rather go that way.
In the end there will be the same costs with any system: The uncorrected power demand of the driver, the additional power to force it to lower frequency response than it´s Qtb predicts (like a Linkwitz transformation) and finally the energy needed to correct the cone movement, as detected by a sensor.
I'll avoid writing more crap. However I follow ...
Using dual voice coil for feedback was discussed a couple of months(?) ago in the Blowtorch thread
with your method you have a direct connection to the circuit. Fast reaction and dependability.
Piezo's and accelerometers are out of the circuit and measuring cone movement feeding a signal back into the drive circuit.
Both have their place however I am partial to some and direct.
Mr. Erath's circuit is like the negative current method only a bit more complex with extra parts and stages. Which makes it somewhat temperamental for others. Mr. Erath knew that circuit inside and out and could adjust it for any scenario.
Troy, can you point to Mr. Erath?
Negative impedance goes back to the late 50th, but I want to implement some DIYS error correction to 2019 material. So all filter stuff will be done by a DSP, which we both know costs about 15$. When I first read about MFB you had to etch your own PCB... so my idea is to take as much finished material, like DSP board and D-amp modules. Then you only have to build a tiny board to do the signal correction. Even there are cheap mini boards for any purpose.
Sure, the voice coil is the fastest gun for correction signals, but the frequency range up to 100Hz is so slow, compared to electronics, that I don´t think a microphone fixed at the cone or a piezo element glued to the dust cap will have any time problems.
Backes&Müller for decades build some the best active speakers available, for sensing the bass they used cheap electret microphones, glued to the cone at the voice coil radius. The trick was to position the diaphragm vertically to the cone. With this construction the microphone only detected sound pressure and the signal could be used for correction. The analog circuit was really simple and, as with all these old constructions, most parts where used for high and low pass filters. So take away what, today, the DSP can do better and the whole thing gets very simple.
There is a lot of talk about the old patents and circuits, but I don´t want to copy, but modernize this stuff.
Another problem with old constructions, they where often used to improve more or less simple, cheap drivers to sound good. Sure an idea loved by the industry. The whole BOSE name is build on this principle...
But I want to use good hardware, put it in decent housings and then get better performance than the passive version.
That´s why I keep on asking. Is there no one who shares this idea?
Chris and PirateLogic go half the way, they modernize the chassis and electronics, but they build with the WAF in mind, small and neat. I don´t care for size, my planing for the next home cinema / music sub has reserved 400 liters of space, for example. I don´t think I will ever build a sub with less than 70 liters. Simply does not make sense for me. A sub woofer has to rock the room...
Negative impedance goes back to the late 50th, but I want to implement some DIYS error correction to 2019 material. So all filter stuff will be done by a DSP, which we both know costs about 15$. When I first read about MFB you had to etch your own PCB... so my idea is to take as much finished material, like DSP board and D-amp modules. Then you only have to build a tiny board to do the signal correction. Even there are cheap mini boards for any purpose.
Sure, the voice coil is the fastest gun for correction signals, but the frequency range up to 100Hz is so slow, compared to electronics, that I don´t think a microphone fixed at the cone or a piezo element glued to the dust cap will have any time problems.
Backes&Müller for decades build some the best active speakers available, for sensing the bass they used cheap electret microphones, glued to the cone at the voice coil radius. The trick was to position the diaphragm vertically to the cone. With this construction the microphone only detected sound pressure and the signal could be used for correction. The analog circuit was really simple and, as with all these old constructions, most parts where used for high and low pass filters. So take away what, today, the DSP can do better and the whole thing gets very simple.
There is a lot of talk about the old patents and circuits, but I don´t want to copy, but modernize this stuff.
Another problem with old constructions, they where often used to improve more or less simple, cheap drivers to sound good. Sure an idea loved by the industry. The whole BOSE name is build on this principle...
But I want to use good hardware, put it in decent housings and then get better performance than the passive version.
That´s why I keep on asking. Is there no one who shares this idea?
Chris and PirateLogic go half the way, they modernize the chassis and electronics, but they build with the WAF in mind, small and neat. I don´t care for size, my planing for the next home cinema / music sub has reserved 400 liters of space, for example. I don´t think I will ever build a sub with less than 70 liters. Simply does not make sense for me. A sub woofer has to rock the room...
Last edited:
I understand the idea, am actually talking with a company who is looking at putting feedback into a 32 inch equipped sub the size of a washing machine. No doubt it will work but is it the way forward for MFB ? Why throw away it's unique selling point - the ability to make small sound very very big - and build just another large box ?
If you insist on big let me just daisychain 8 of my Little/One's for you
I've tried modern adxl sensors and found them not only to be noisy, they are limited to putting out ones and zero's requiring lots of extra dsp chips that all add noise too. Piezo based analog accelerometers also generate noise which in my experience with small drivers remains pretty much unnoticed upto about 20db feedback. However if you combine digital room correction, 30dB feedback and large drivers into an audiophile setting with people putting their ears 15cm away from the dustdome it will probably become a point of discussion.
Current drive, or better, current feedback allows you to get status info back from the voicecoil whereas voltage drive just dumps energy into it without further bothering what state the coil is in. In an ideal world current feedback would give the same results as accelerometer feedback, however it's not an ideal world : as the voicecoil heats up it's characteristics change making the current info useless.
Dual voicecoils suffer from transformation effects between the driving and sensing coil, they also tend to become nonlinear as the coil travels towards xmax due to BL variations.
Here's a simple thought experiment I did to visualize what would happen if drive current was kept constant at all frequencies rather than drive voltage. Is there an error in my analysis?
Voltage Drive
Fr is the point the speaker is most efficient and has the highest impedance to current flow (Fr). Even when mounted in a box and the system Q is only, say, 0.7, an impedance peak still occurs at the system Fr. This impedance peak would naturally limit current in the voice coil around Fr and results in the smooth, damped acoustic frequency response characteristic we enjoy as predicted by programs like WinISD for the closed box alignment.
Current Drive
Unlike the constant voltage drive, constant current drive will produce more current in the voice coil around Fr where the impedance is high. This will result in the acoustic amplitude peak at Fr as described above by ds23man.
I understand the idea, am actually talking with a company who is looking at putting feedback into a 32 inch equipped sub the size of a washing machine. No doubt it will work but is it the way forward for MFB ? Why throw away it's unique selling point - the ability to make small sound very very big - and build just another large box ?
If you insist on big let me just daisychain 8 of my Little/One's for you
More speakers is one way around it!
There is no reason to limit the application of this technology to small speakers. Different requirements need different solutions. There are applications for everyone!
What MFB can do for small speakers with regard to distortion and frequency extension is amazing. But as pointed out by Turbowatch2, basic physics (i.e. loudspeaker displaced volume) greatly limits the ability of small speakers to produce very low frequencies at usable sound pressure levels. At really low frequencies like 20 Hz, small loudspeakers with MFB will really struggle to produce enough SPL to be useful.
During his time he made "The Ferrari of loudspeakers". (Not my quote, one of the major hi-fi rags of the time)
Yahoo! Groups
Yahoo group for LWE motional feedback loudspeakers made by Louis W. Erath & Acoustron | Audio Forums
Louis W. Erath....Rest In Peace | Audiokarma Home Audio Stereo Discussion Forums
Yahoo! Groups
Yahoo group for LWE motional feedback loudspeakers made by Louis W. Erath & Acoustron | Audio Forums
Louis W. Erath....Rest In Peace | Audiokarma Home Audio Stereo Discussion Forums
Last edited:
This is seductive.
Could we adjust the gain on the feedback signal, sensing the current in a reference resistor?
Attachments
Sounds like a valid analysis to me - the only reason why I'm still on voltage drive is because I'm using bridge tied load amplifiers and implementing current sensing in BTL setups is a pain.
Elliott wrote two very articles on this subject:
Variable Amplifier Impedance
You could eq the bump out, but:
Effects Of Source Impedance on Loudspeakers
The only way to correct the missing damping factor is MFB.
Variable Amplifier Impedance
You could eq the bump out, but:
Effects Of Source Impedance on Loudspeakers
The only way to correct the missing damping factor is MFB.
Looks very much like the 3way 15" McKenzie with ACH01 sensor, 6" Audax pr170m0 mid and JBL tweet I build back in 2012
