Finding a practical feedback method would seem a logical step, it works everywhere else. I don't see anything inherently wrong with moving a diaphragm dynamically?
Moving diaphragms have an inherent, unavoidable distortion mechanism. It can be minimized by minimizing their axial movement. Microphones have it too, but they don't normally move enough to matter. (An exception is kick drum mic's who live inside the big SPL boomers. 25 acoustic Watts baby - selected for their "tone".)
And to all, a good night,
Chris
And to all, a good night,
Chris
Probably 'wrongly'. Why? Because sterile amps have a sound, and that sound is imparted to everything played through them. If they were truly accurate then recordings made on different equipment would not have a shared amplifier-type sound.
Jam and I have speculated that the sound might be associated with very low level high order HD components. Not sure about it though.
Amps that are not sterile are also imperfect, but in a different way which can be much more musical (by musical, it means it sounds like music is coming out of the speakers rather than a recording of music).
Last edited:
First, thanks for your inputs. You seem to have a good experience on the subject. Some correlation with your pseudo ? ;-)
Let discuss this point.
The large radiating area have at least a big advantage: it can match (and have to when you design one) the one of the bass driver. It means the directivity (total energy) will be the same at the crossing frequency than the one of a cone bass speaker. No break.
This is impossible with cone speakers.
The second advantage is a motor can cover a large bandwidth. This allow a two way speaker, and everybody know how prejudicial is a crossover.
On the other side of the bandwidth, it is, indeed, more problematic. I never succeeded to have good enough results on all my tries to add a tweeter to my horn. At this occasion, i realized that there is no real interest of such a tweeter. Yes, it adds "shine" to the sound. But this shine is quite unnatural (with most of the recordings) and the informations here seems to be mostly distortion components. So, my tweeter is "on/off". "On" to figure out the tonal balance in the extreme treeble when I wanted to monitor a mix, "off" to listen to music. This said, I know i'm old, so my bandwidth is very limited and probably, under the one of my horn assembly.
May-be the solution should be to use a coupling mount and use the same horn for high medium and tweeter ?
One last thing is the crossover. After lot of tries, I settled that, contrary of what lot of the audiophiles pretend, a 6dB/oct is not good. The double radiations with non correlated phase curves are prejudicial. I use acoustical 48dB/oct on both ways.
Next thing is to phase align the two ways. Makes an obvious difference on transients. Easier, nowadays, with digital delays.
The balance of the level between the two ways is very critical: A lot of energy on the two sides. The precision of measuring instruments of the acoustic response curves not good or revealing enough. I spend a lot of time, listening to lot of records to find the best spot. At 0.25dBsteps.
To answer the remark of Scott, I assure him that the big horn system I used in my studio was not fatiguing at all and allowed long days of mixing.
On the contrary, the details and attacks of the horns request less efforts of attention. The localisation and separation between instruments are "spot".
The negative aspect is that not lot of records "sound" perfect. Each little mistake in the mix that we are unable to figure out with cones are underlined.
Reason why I said such a speaker is not necessary the best for "easy" listening at home. Like a sport car is not the most comfortable on bad roads.
I found them very revealing of the character of speakers. they underline all the defects. You have, of course to abstract the tonal balance (Did-you noticed that all those enclosures have too much high medium and treble ? That boomers resonnances were obvious ?) But listen to the piano, percussion attacks even the wind noise of the saxophones.
One thing is for sure, if you are able to record a natural reproduction from your speakers this way, you can be sure you have an optimal result: Our ears are lot more forgiving to erase the room's resonances and tolerate response curve accidents.
Last edited:
Sorry for the flood, but i disagree. What I mean by "sterile" is a lack of micro informations. Especially on the reverberation tails. Like a brutal noise gate at the lowest levels. The difference between 16 and 24 bits with digital.
Not enough bits, see what I mean ? Add pink noise to hide the problem, the discomfort disappears. Sorry, but i'm sure it exists.
You will argue our listening dynamic is lot more limited ? Not true, omho. Like our eyes correct in permanence the white balance and the contrast curves in the dark areas, our hearing system seems to act like a limited dynamic filter that is permanently adjusted to the acoustic power.
With all the advantages that class D brings, be assured that, once this problem will be solved, it will be my nirvana for power amps.
Last edited:
No sorry that makes no sense at all. If anything the Ncore, with lower noise floor will excel in mythical decay tails.
Probably not too few bits. Bruno's self-oscillating class-D amps appear to have a lot in common with SD modulators used in DACs, etc. It may be more or less the same missing-reverb-tail issue that some have complained about for DACs. The bits are all there for a sine wave, but somehow reverb tails get lost. Complicated nonlinear dynamic behavior of apparently-less-than-fully-understood SD modulators it would seem. Nobody knows exactly why or those that do know aren't saying.
Last edited:
Lower noise floor ? Like a 16 bit CD erase all the noise under -96dB and give 1 bit of accuracy at this level ?
You are constantly referring to measurement numbers. They need to be interpreted. Give more room to your listening experience, that need too to be interpreted correctly, I agree ;-)
BTW: As i don't read audiophile reviews, I was not aware of this "mythical decay tails" before to be in this forum. I had this feeling since the early beginning with digital in my studio.
Reading this just reinforces my idea that I am not the only one to have this feeling.
You, folks, are often laughing at the Markw4 researches. Reading what he usually writes, it seems to me that both of us are focused on the same details with similar listening experiences.
He sometimes blindly seeks how to remedy the faults he hears? Do you have other solutions to offer, rather than denying the failings (that many also seems to had experienced), by attributing our feelings to self-suggestion or audiophile myths ?
"And yet it moves."
Last edited:
There is also acoustic intermodulation (beats) and binaural beats are interesting as they are created between your ears 😉 Beat (acoustics) - Wikipedia
T.,
Bill is a reasonable guy, IMHO. I hope he will visit Auburn someday, it might interesting for both of us.
In regard to the reverb tail thing, he is thinking as trained engineers do. It is in terms of linear models that can be applied to very many things quite well. The engineering tools for nonlinear devices are not nearly so well developed so not much time spent on that in the early years of training. In addition, SD modulators are kind of a specialty area.
Folks need to listen in the same place at the same time to understand each other's perceptual point of view. Arguing in a forum just won't do it.
Bill is a reasonable guy, IMHO. I hope he will visit Auburn someday, it might interesting for both of us.
In regard to the reverb tail thing, he is thinking as trained engineers do. It is in terms of linear models that can be applied to very many things quite well. The engineering tools for nonlinear devices are not nearly so well developed so not much time spent on that in the early years of training. In addition, SD modulators are kind of a specialty area.
Folks need to listen in the same place at the same time to understand each other's perceptual point of view. Arguing in a forum just won't do it.
Last edited:
Absolutely true.
In my motion control work using PID loops where the driving element is magnetic, all the standard models are linear. As such, there is no accounting for the system transfer function change when the motor is loaded, and the magnetic field spring constant changes. As the motor load changes, so does the system's Bode plot, meaning gain and phase margins are dynamic.
The controls people are limited to setting parameters so that their system under no load has sufficient phase margin that it won't oscillate. Unfortunately, that means under load, the accuracy drops.
For motion devices I troubleshoot, the PID system is sufficiently fast that I can built in adaptive system parameters to change as forces change. Meaning, during acceleration, I can increase system gain, and when the device's magnetic reaction forces increase, I can do same.
Speakers do indeed have the same characteristics, but I never see anybody consider this non linearity.
Applying an adaptive gain to an audio amp such that feedback gain increases as a function of current draw of the load would be the equivalent. For me luckily, I'm dealing with bandwidth at the ten Hz level, and static accuracies on the order of 5 nanometers with on the fly accuracies of 25 nanometers, 2 to 3 orders of magnitude beyond others in the field.
Jn
Ps...kinda hoping to distract away from the browbeating and silliness.
Last edited:
I posted this link earlier https://www.klippel.de/fileadmin/_m...ON_OF_NONLINEAR_LOUDSPEAKER_DISTORTION_02.pdf
He considers the static positional based nonlinearities. I'm speaking of dynamic nonlinearities.
What he describes in the opening paragraph is what we call feed forward compensation. It is a predictor of what the load requires to do the correct motion. The controls guys were taught to keep the gain down and rely on the feed forward parameters velocity, acceleration, and friction to make the load move closer to desired. It is quite incapable of dealing with a non linear load. The low gain I call a "soft tune", and it is the most common tuning issue I deal with.
The other problem I deal with a lot is the mechanics. The vendors and our mechanical engineers are not trained to consider motion control dynamics of the system, I end up showing them what the problems are and how to correct them. Sometimes the solutions are based in 6th grade geometry, sometimes it is more subtle, like spring loaded involute gears that are forced into the root, or two independently controlled motors mounted to one springy piece of 8020 aluminum rail.
In dealing with mechatronics trained engineers, I find there are still significant holes in their education.
Sigh
Jn
What he describes in the opening paragraph is what we call feed forward compensation. It is a predictor of what the load requires to do the correct motion. The controls guys were taught to keep the gain down and rely on the feed forward parameters velocity, acceleration, and friction to make the load move closer to desired. It is quite incapable of dealing with a non linear load. The low gain I call a "soft tune", and it is the most common tuning issue I deal with.
The other problem I deal with a lot is the mechanics. The vendors and our mechanical engineers are not trained to consider motion control dynamics of the system, I end up showing them what the problems are and how to correct them. Sometimes the solutions are based in 6th grade geometry, sometimes it is more subtle, like spring loaded involute gears that are forced into the root, or two independently controlled motors mounted to one springy piece of 8020 aluminum rail.
In dealing with mechatronics trained engineers, I find there are still significant holes in their education.
Sigh
Jn
Last edited:
TT..you caught it before I finished editing. What I meant was velocity feed forward, acceleration feed forward, and friction feed forward. (VFF, AFF, and FFF).
I was too lazy to write them out as I'm on an IPad..
In the control system block diagram, those three use the desired location as a feed to the output, bypassing the gain block. Friction is a particularly bad one as when the velocity passes through zero to opposite sign, the system jumps. Unfortunately, the control people are taught to use it to make a good step response, yet the application never needs a step. Having a multi-ton object settling down in 20 milliseconds after a step is simply absurd.
Jn
I was too lazy to write them out as I'm on an IPad..
In the control system block diagram, those three use the desired location as a feed to the output, bypassing the gain block. Friction is a particularly bad one as when the velocity passes through zero to opposite sign, the system jumps. Unfortunately, the control people are taught to use it to make a good step response, yet the application never needs a step. Having a multi-ton object settling down in 20 milliseconds after a step is simply absurd.
Jn
Last edited:
What does he mean here?
"The control law represented by Eq. (14) is static in the state variables of vector x. Alternatively, the control law can also be written in a dynamic form (mirror filter [17]) by using differentiators resulting in a structure with reduced complexity where each term has a physical meaning showing the compensation for nonlinear stiffness, nonlinear damping, parametric excitation, electromagnetic reluctance force and nonlinear self-inductance. Both control laws are based on the same physical model and provide exact linearization of the modeled nonlinearities in theory."
- Status
- Not open for further replies.