Now these dynamic speaker mechanisms may not be fully understood, but with a voltage drive you get variable current and thus variable force on the cone, this is both understandable and desireable, as it is a mirror of the varying dynamic forces on the speaker VC, through a cycle the spring like forces changes in both magnitude and direction, the speaker driver is also a microphone and it creates its own current which is then summed to the driving current creating the measured distortion.
Current drive is just as bad (actually a lot worse) as it will be like constant force but there's still the varying load, this will create a lot of acoustical distortion as that will be the only physical possibility.
This truly highlights some of the issues with speakers and also underline the reachearch work we do at Raidho aimed at reducing speaker induced dynamic unliniarities, this is primarily done by reducing VC inductance (-75%), removing VC material hysteresis and by changing cabinet damping from being a resonat spring (sealed and refex) to a resistive flow structure, this reduces the varying forces on the membrane quite a lot and thus also reduces the speaker driver error currents.
Current drive is just as bad (actually a lot worse) as it will be like constant force but there's still the varying load, this will create a lot of acoustical distortion as that will be the only physical possibility.
This truly highlights some of the issues with speakers and also underline the reachearch work we do at Raidho aimed at reducing speaker induced dynamic unliniarities, this is primarily done by reducing VC inductance (-75%), removing VC material hysteresis and by changing cabinet damping from being a resonat spring (sealed and refex) to a resistive flow structure, this reduces the varying forces on the membrane quite a lot and thus also reduces the speaker driver error currents.
Last edited:
Can you explain more kind sir?
Because of PMA's fantastic real data, I was just getting out my ol' current drive chipamps and old BUF full ranger and was gonna play with Nelson's circuits:
from:
http://www.firstwatt.com/pdf/art_cs_amps.pdf
http://www.firstwatt.com/pdf/art_cs_xvrs.pdf
Hey Nelson, when you were playing with current drive, did you do any distortion measurements or did you stick to frequency response?
Thanks and cheers,
Jeff
Current drive does not change the physics of speakers. constant current will make the force on the cone constant, but as the load is not constant but very depending where in the sine cycle the driver is. Sometimes cabinet pressure is in the same direction as cone movement then when motion crosses the midpoint the force changes direction and cyclic magnitude, same with the springs from the surround and the spider in the driver. These forces will sum with the current driving force and cause the speaker motion to be a sum of that with quite a large acoustic compromise.
Voltage drive is in that respect self correcting. As the amplifier will see a load changing with amplitude pulling more or less current as needed.
Voltage drive is in that respect self correcting. As the amplifier will see a load changing with amplitude pulling more or less current as needed.
I assume you would not like to do the same with woofers, George
The same but not with that thin fabric.
Acoustic resistance (R) and enclosed volume (C) are both adjustable.
Acoustic resistance can be made to be frequency dependent, thus it additionally affects the enclosed volume depending on frequency.
With woofers in particular, acoustic damping can be applied also at the front side of the speaker.
I agree.
Very true.
No, this shows you do not understand the reason of distorted speaker current. Even an ideal voltage source with zero distortion of the voltage output and zero output impedance will not prevent speaker current from distortion.
I = V/R
It seems that some people do not understand even the simplest equation.
If V is ideal and absolutely clean, and R is nonlinear, then I is distorted. Ideal voltage output does not help. You would need to monitor output current and make instantaneous corrections, with strong and effective feedback derived from output current
Pavel
Even if you could make instant corrections, then it would not make things so much different as a great part of the current distortion is actually desirable. This is the part of the back em(f)(C) that is part of the dynamic load. other parts is not so desirable, this is mainly the elements that comes from nonlinear elements like a changing impedance with the cone position due to VC inductance/iron contribution or in the form of dynamic hysteresis damping in Aluminum VC-formers.
I believe this is this main difference in amplifiers, how well they work with this dynamic current loading...
We still need to understand and explain the difference i listening performance of VFA vs CFA amplifiers, this may be one of the big elements in that.
Even if you could make instant corrections, then it would not make things so much different as a great part of the current distortion is actually desirable. This is the part of the back em(f)(C) that is part of the dynamic load. other parts is not so desirable, this is mainly the elements that comes from nonlinear elements like a changing impedance with the cone position due to VC inductance/iron contribution or in the form of dynamic hysteresis damping in Aluminum VC-formers.
I believe this is this main difference in amplifiers, how well they work with this dynamic current loading...
We still need to understand and explain the difference i listening performance of VFA vs CFA amplifiers, this may be one of the big elements in that.
Last edited:
Michael,
distortion in speaker current reflects sinusoidal cone move at the amplitude of measurement and most of the mechanical issues. As you know, this is all reflected through a gyrator to the electric side of the speaker. Let's forget back EMF explanation for a while, it is too popular and simplified.
Current is the driving variable in a speaker, as you have stated correctly. Once the current is distorted, acoustical speaker distortion cannot be lower, only higher. I am sure you are aware of experiments that show that speaker distortion is lower when driven from almost ideal current source (i.e. not a constant current, DC is boring, but from a source with lowest current distortion) than from almost ideal voltage source. The only advantage of voltage drive is the most flat amplitude response of the driver. With current drive it would have dips and peaks exactly according to impedance curve. I am sure that voltage drive of speaker is not ideal re speaker distortion, but best from nowadays methods, best from bad methods. And I am sure that once we shall find a proper way how to drive a speaker for the lowest distortion. Zero impedance wire with gain will not be the solution, that's for sure.
distortion in speaker current reflects sinusoidal cone move at the amplitude of measurement and most of the mechanical issues. As you know, this is all reflected through a gyrator to the electric side of the speaker. Let's forget back EMF explanation for a while, it is too popular and simplified.
Current is the driving variable in a speaker, as you have stated correctly. Once the current is distorted, acoustical speaker distortion cannot be lower, only higher. I am sure you are aware of experiments that show that speaker distortion is lower when driven from almost ideal current source (i.e. not a constant current, DC is boring, but from a source with lowest current distortion) than from almost ideal voltage source. The only advantage of voltage drive is the most flat amplitude response of the driver. With current drive it would have dips and peaks exactly according to impedance curve. I am sure that voltage drive of speaker is not ideal re speaker distortion, but best from nowadays methods, best from bad methods. And I am sure that once we shall find a proper way how to drive a speaker for the lowest distortion. Zero impedance wire with gain will not be the solution, that's for sure.
No, this shows you do not understand the reason of distorted speaker current. Even an ideal voltage source with zero distortion of the voltage output and zero output impedance will not prevent speaker current from distortion.
I = V/R
It seems that some people do not understand even the simplest equation.
If V is ideal and absolutely clean, and R is nonlinear, then I is distorted. Ideal voltage output does not help. You would need to monitor output current and make instantaneous corrections, with strong and effective feedback derived from output current
Do not jump to cocnlusions, Pavel, especially rather insulting ones. If one's amp is a good voltage source and if its contrinbution to distortion is low enough at the delivery side of the speaker cable, I do not expect zero speaker distortion, but I do expect at least sightly reduced overall distorion because I expect the amp to be well behaved. Even if my own speakers vary much less in impedance and do not fool around much with phase shifts due to the fact that their impedance modulus is MUCH better behaved than is typical, the very fact that it does have 2nd/3rd order crossovers means that I will not get the perfect sound. Same goes for the drivers used - at the time, they were among the best Son Audax offered, and none too cheap, but they will still produce at least some anomalies at some point. But together, I expect less distortion than is typical of of them assuming the amp has no distortion worth talking of.
In other words, I am not expecting a disappearance of all distortion, but I am expecting reasonably small distortion products. I have never measured them, nor do I have the necessary instrumentation to measure it, but my ears tell me that what I do have is nothing to worry about at low (up to 2Vrms) or very high levels (about 23Vrms), because the overall tonality remains the same, it only becomes louder.
Last edited:
Erratum
It was J.E. Benson (the examiner of R. Small’s PHD Thesis) in his “Theory and Design of Loudspeaker Enclosures”
George
Let's not forget Rs. The eddy losses are not linear, and decoupling to the amp will not decrease the distortion.
Non linear ain't the half of it.
It's also signal dependent. As is Rs.jn
I have an old Jands PA amp, proper/conservative 500/4 + 500/4, ten MJ15023/24 ouput transitters per channel, MJ15023/24 drivers also.
Infinite current capability, same deal, stays same sound just gets louder and bigger, much bigger.
At low levels it was as sweet as also...stayed in Class A.
I don't use it nowadays because, A I am running active cabinets, and B I don't like the quiescent power bill.
Else wise, it's a really great amp and was seriously good fun when running four stacked 12'' 3 ways...as loud as if not louder than concert levels.
Vibrated/massaged the hell out of the the couch/listener, cracked the rendered brick walls and lifted the carpet yet no ear bleeding.
AC/DC (or band of your choice) performing live in the lounge room !.
Dan.
So, how do parallel RC and RLC networks influence ?.
My ears tell me reduction in perceived distortion.
IME, these networks need to be incorporated directly at the driver connections or they don't work properly.
If applied remotely they can in some respects worsen the perceived distortions.
Dan.
My ears tell me reduction in perceived distortion.
IME, these networks need to be incorporated directly at the driver connections or they don't work properly.
If applied remotely they can in some respects worsen the perceived distortions.
Dan.
Last edited:
Pavel,
As you stated there are many non-linear elements in a loudspeaker. This has been my area of discovery and development. There are things in certain speakers that make them inherently better devices than some others. For one a short coil in a long gap will help tremendously to reduce the non-linear inductance of the voicecoil due to position. From that one point many other simple changes can improve a speaker in many of these areas. The spiders used by 99% of all manufacturers are the same, they are a real problem in that they do not present a constant force on the cone assembly, it is again a very non-linear action. This is an area I have been working on, a variable rate spider that matches the force needed over the entire distance traveled. The same goes for the non-linear surrounds restorative forces. I am often surprised at the stupid things I see in speakers but most of this is purely economics. using commodity parts and just mixing and matching components and trying for the best outcome. Designing and developing things such as spiders and surrounds see little real science, it is mostly aesthetics and BS, very little real changes are made in loudspeakers, we have not come that far from the very beginning.
As you stated there are many non-linear elements in a loudspeaker. This has been my area of discovery and development. There are things in certain speakers that make them inherently better devices than some others. For one a short coil in a long gap will help tremendously to reduce the non-linear inductance of the voicecoil due to position. From that one point many other simple changes can improve a speaker in many of these areas. The spiders used by 99% of all manufacturers are the same, they are a real problem in that they do not present a constant force on the cone assembly, it is again a very non-linear action. This is an area I have been working on, a variable rate spider that matches the force needed over the entire distance traveled. The same goes for the non-linear surrounds restorative forces. I am often surprised at the stupid things I see in speakers but most of this is purely economics. using commodity parts and just mixing and matching components and trying for the best outcome. Designing and developing things such as spiders and surrounds see little real science, it is mostly aesthetics and BS, very little real changes are made in loudspeakers, we have not come that far from the very beginning.
Adding a 1 ohm series resistor (for current transduction/measurements) changes the test conditions.
Is there not a better way of deriving current measuremrnts...ie current transformer or hall effect current transducer like ACS712: Fully Integrated, Hall-Effect-Based Linear Current Sensor IC ?.
These are easily/cheaply available on Ebay pcb mounted.
Dan.
Is there not a better way of deriving current measuremrnts...ie current transformer or hall effect current transducer like ACS712: Fully Integrated, Hall-Effect-Based Linear Current Sensor IC ?.
These are easily/cheaply available on Ebay pcb mounted.
Dan.
Last edited:
Pavel, this is not entirely correct, true the drive is current and not voltage, and sure if you look a narrow band like in how do we create the lowest possible acoustic distortion at 1kHz, then a 1kHz sine current would be the best, but if the question is how do we create the best possible solution over a broad range of frequencies and SPL's then voltage drive is the only option. Some of the current distortion you see is not necessarily the same as an acoustic distortion, the acoustic may be lower as the driver needs a distorted (opposite) current to perform distortion free.
I'am not fan of creating parallel RCL current escape paths for speaker currents. to me it robs the speaker of life and dynamics, the path exemplified by the cloth covered woofer is a much much better way.
It may be that Mr Small showed how to optimize a cabinet for the widest frequency extension, but it's based on springs and resonance of masses and springs, this is exactly what we (I) try to combat in my speaker design. Nothing is worse than a system playing one note exhaust bass. An analogy would be to drive a car with only springs as suspension. somehow there must be some kind of damper placed there, with speakers it can be made by controlling the airflow/motion inside the cabinet.
- Status
- Not open for further replies.