Kindhornman, those resonances you are talking about are typically breakups, and can indeed be triggered by harmonic distortion from lower frequencies, but this is not the kind of phenomenon we are talking about here.
In the case of a peak at Fs caused by current drive, this should be 100% minimum-phase, and should be EQable without issue.
In the case of a peak at Fs caused by current drive, this should be 100% minimum-phase, and should be EQable without issue.
The music coming from the recording will be reflected back at the driver from inside the box, from adjecent surfaces outside the box like the floor walls etc. Any kind of resonant tuning like a horn om bassreflex will store energy, which will effect the driver.
Please attach a oscilloscope to a loudspeaker in a room and sneeze anywhere the driver. Close the door or make any other kind of sound. It is a very sensitive microphone, picking up energy from all around. No EQ can stop that. Any high Q peak will be excited when playing music.
Why use a passive network when you can use more precise digital EQ and filtering?
If you EQ down a given frequency (Fs here) then reflexions from inside the box and from the room will also get attenuated accordingly.
BR tuning, horn loading, individual reflexions from within the box or the room, all these are minimum-phase phenomenons.
Last edited:
Lets pretend that all solid state class a/b global feedback voltage amplifiers love highly reactive loads, and that no amp ever sounds harsh and gritty driving a high Bl speaker and its typical large phase-shifts. It such a bliss ignoring 90% of the problems with the interface between a highly reactive load and a feedback-damped high gain voltage amp....
Please run a squarewave through a coil and capacitor in a parallel resonant circuit with your favourite amp. They usually just love to run near clipping for prolonged periods of time into highly reactive loads.... There is never any problems with stability....
Most normal class a/b global feedback amps out there in the stores or your hifi-rack is superficially stable only because of a lot of additional circuitry like millercaps and modified feedback loops. They are not inherently stable. They need to be tamed and slowed down to function properly and not turn into an am-transmitter.
When running a high Bl driver from a normal voltage drive amp the sound improves considerably with a coil or even a few ohms of resistance in series with the loudspeaker.
It might not be easily measured as distortion, but it is very easy to hear the change with listening tests. If the voltage-amp was so nice sounding and perfect with its extremely low levels of distortion, why would anyone ever want to mess with current drive?
Please read this AES paper:
35 "Comparison of Direct-Radiator Loudspeaker System Nominal Power Efficiency vs. True Efficiency with High-Bl Drivers," presented at the 115th Convention of the Audio Engineering Society, New York (Oct. 2003).
From: AES Papers -- Official website of D.B.Keele
Cheers,
Johannes
Last edited:
This is the impulse response from a 2 meter long and 1000 cm2 cross-section tube with a driver exiting 1 cm from the mouth (200 cm from the closed side)
This is a single impulse, very similar to the first transient of a drum. I can see a lot of energy being absorbed then gradually released by the high Q resonator, even though the driver is very loosely coupled to the resonator.
Please explain how to EQ all this resonant behavior away?!
A high Qms driver is a good resonator to, why would it not behave in the same way when driven from a high out-z amp?
Here is group delay and spl-response from the pipe alone. The driver response is ignored in this sim.
I am sorry, but phase or no phase, i have a hard time understanding how simple EQ can cancel or counteract the pipe resonance. It will resonante from almost any kind of complex signal, even without the Fs fundamental. Please play EQed music with a loudspeaker near the end of a quarter wave resonator like a large pvc-pipe closed at the other end. You can hear a resonant behavior , even though you are not playing the fundamental. Just like a guitar will twang when excited by a simple tap from my finger.
A driver never consists of a pure L//C circuit.
There are surrounding damping circuits, at least, a series RC (Re+Le) and an apparent resistor, Res, parallel to the reactive Les//Res circuit.
I do not see how a standard amp would have a problem with the resonance of a parallel LC resonant circuit as in that case, the amp is only offered a very high impedance load at resonance. Well conceived, it superbly ignores phase differences between current and voltage.
One generaly seems to ignore that negative feedback was invented to stabilize the amplification gain and to avoid it to be disturbed by the load.
My current favorite amp is the Renardson Mosfet amp wich has a huge amount of negative feedback. I extensively tested it with square waves, and sinewaves for harmonic distortion, on real L (16 µF !) ad C (1 mH !) components. It presents a much better behavior than expected, that means an absolute insensitivity to the reactive character of the loads.
In such tests, one must not forget to limit the bandwidth of the input signal, let's say to 100 kHz, otherwise the conclusions may be wrong.
Rather than asking other people to experiment the tests you propose, please show the results (not simulations) you have already obtained to base what you say, currently it looks as nothing else than a scientific speculation.
Last edited:
'Minimum phase' is just a name given to a definite relationship between frequency response and phase response. Know one, know the other exactly. Crops up in nature all over the place, but pipe and beam bending resonances are an exception IIRC, whereas spring-mass-damper systems such as speaker driver resonance conform. Any two minimum phase systems with opposite EQ will exactly cancel in both f and phase domains.
I'd suspect that over the narrow range of frequencies we're interested in surrounding the driver resonance, all systems would approximate min phase.
Lastly, the elegance of current drive is that voice coil force remains constant, even at resonance. So how far the cone moves at resonance depends just on Qms of the driver, and the impedance presented by air the cone has to compress front and back. So cabinet/pipe loading has a big part to play in determining resonant response, or Qts of the driver in situ, for current drive. But Qts won't be worse than driver Qms.
I'd suspect that over the narrow range of frequencies we're interested in surrounding the driver resonance, all systems would approximate min phase.
Lastly, the elegance of current drive is that voice coil force remains constant, even at resonance. So how far the cone moves at resonance depends just on Qms of the driver, and the impedance presented by air the cone has to compress front and back. So cabinet/pipe loading has a big part to play in determining resonant response, or Qts of the driver in situ, for current drive. But Qts won't be worse than driver Qms.
You don't need the whole LCR, just the R. And it can be effective at a high value like 50 ohm to 100 ohm so you're not wasting power.
The parallel R also helps to 'tame' drivers with too high Qms.
Nelson went over all this long ago ...
www.firstwatt.com/pdf/art_cs_xvrs.pdf
www.firstwatt.com/pdf/art_cs_amps.pdf
You can also change the Qms of a driver by using a resistive flow enclosure (like they used to in the ol' days). I'm using layers of cloth sandwiched between 2 pieces of pegboard ...
F = (Bl).i
i is constant but (Bl) ?
In current drive, the force is more constant than with voltage drive but it is still submitted to the variations of (Bl), mainly B, depending on the position of the voice coil in the gap.
Current drive is not the panacea to get linearity as many think of.
Not sure what you are simulating here, but if it involves several reflexions or diffractions adding together (comb filtering) then of course the result cannot be addressed using EQ.
Still, each reflexion is minimum-phase.
Current drive or difference in Qms will not cause such phenomenon, and if it exists in a system (with or without current drive, and regardless of the Qms of the driver used), then EQing the response of the direct signal can correct each reflexion exactly as if it was emitted using voltage drive or a different Qms driver.
It is a simulation of a resonant pipe very loosly coupled to a sound-source. It is similar to a high Qms driver in a box or OB radiating sound then reacting to the energy emitted bouncing around both inside the box (no damping is perfect - since we don't want a black hole inside the box) and outside the box/OB reflecting from nearby surfaces like the walls, floor etc.
If i play music on one speaker with an activ highpass crossover 24 dB/octave at 100 Hz, i can still feel the cone on the other speaker move, even though it is not even connected to the amp (similar to current drive with no electrical damping). Since the Fs of the driver in the box is at about 43 Hz, i would guess a steep highpass filter at 100 Hz crossover frequency should remove enough energy at 43 Hz not to excite the other speaker at the other side of my room. How should i EQ the signal to have an absolutely still cone at the other speaker?
I need to actively highpass the first speaker at 1000 Hz 24 dB/octave to not feel any movement in the passive speaker across the room. How can this be when all i need is "some EQ to not excite a resonant system"
A oscilloscope will still show 10 khz tone in the passive speaker if played from the active speaker, so obviously the driver will absorb energy even many octaves above the Fs.
How do i EQ away this phenomenon?
Here you are talking about how a speaker reacts to another. These are two uncorrelated signals. This is not at all the same situation as a source reacting to the effect it has In you
That said, in your experiment I don't see how a driver strictly generating only >100Hz frequencies could trigger a <100Hz resonance. Are you sure what you are feeling is a 43Hz resonance?
That said, in your experiment I don't see how a driver strictly generating only >100Hz frequencies could trigger a <100Hz resonance. Are you sure what you are feeling is a 43Hz resonance?
Last edited:
MJR7-Mk5 Mosfet Power Amplifier Design Notes
I like the simplicity and well thought through design of this amp. It has some negative feedback, but i would not say "huge amount". Many modern amps has several extra gain stages just to maximize the open loop gain to "huge amounts". Add an LTP input stage, with cfp devices and current-mirrors and you could get 50 dB more gain and negative feedback. I don't think that would do much good for the stability of the amp.
It is the same thing as a current driven speaker first emitting sound the reacting to the sound once it has reflected back from the floor/walls/box etc. A high Qms driver will absorb power and resonate from any sound reflected from anywhere Those sound-waves will not be minimum phase or whatever it is called, but will be very complex since there is many different reflections in a room arriving at very different times with lots of comb-filter effects.
This is a real user scenario for a current-driven loudspeaker in a normal room. This is why i don't think one can EQ away a huge peak from a high Qms driver to have a flat response. It is better to have a well damped driver in the first place.
A high Qms driver with EQed flat response will suffer from intermodulation effects, it will absorb energy around Fs and multiples of Fs, and resonate unharmonious with the music being played. I can't see this as good thing.
Drive a car at high speed along a bumpy road, with suspensions having a Qms of 15, and at the same time try to draw a perfect circle freehand, and you might understand why it is bad having a resonant system easily excited by the work it is expected to do while at the same time trying to have a great precision (low distortion). I guess we could have a complex series of hydraulic pistons controlled by a very fast computer constantly correcting every movement in the chair in the car, and that way have a smooth ride and a possibility to draw that circle. But why not just design the suspension for the intended use in the first place?