30 years ago Keith Johnson and I designed and shipped subwoofers with a servo that sensed the position of the cone. It worked well in reducing the distortion and extended the response down as far as we wanted to go. Much easier than trying to predict the distortion.
Ok. I think I am missing something. My ignorant view is that an amplifier is meant to control the speaker/motor. If the speaker overshoots after an impulse (which is a mechanical issue) isn't the amplifier meant to try to reduce this overshoot? Isn't overshoot an undesirable thing?
So any distortion inherent to this control system is, well, a necessary evil. Isn't it better to tolerate this distortion than to leave the speaker uncontrolled?
No. As I said, the compensation networks has to be added *to each speakers*, IE after the filters. thus, the filters designed accordingly.
Take a boomer, try-it, and even with a high damping factor amp, you will, isn't strange ?, *see* a difference in the response curve of the boomer. Why do-you want to change the 'speaker alone' accordingly ? Just to see if the audible change is just the result of the litle response curve change ?
It is not (only) a question of response curve, it plays on the instant dynamic, damping, subjective balance between basses and treble etc...
You know, those awful sooo subjective differences that measurements don't underline ;-)
Why don't you try with your ears, instead of your brain that cannot take all the parameters into account ? You know ? Observe, then try to explain, that I'm sure, you will do better than me !
You still don't understand the difference between two people simply liking their tone controls in different places and making up stories about how one or the other setting is more "friendly" to the average amplifier.
I was simply addressing the technical stories. As Mark said you can simply change the volume and elicit stanzas of subjective poetry from many listeners. Virtually all the plots here show +-3dB peaks and dips that move around from design to design, of course careful listening can tell a difference, the preferences remain completely subjective.
As I said the other day I'm not an audiophile, I don't easily switch to critical listening mode and I only listen to the music. I usually dread having to endure the usual set of super disks and many of my favorite pieces have horrifying sonics.
Last edited:
Well, there is no free lunch, as says Scott.Erm...no.
So do you or don't you?
I tried, long time ago, by curiosity, like a lot of people ...
Not a lot of current power amplifiers on the market and speakers designed to work with them. May-be for good reasons ?
Arf, I used thousands of them on various mixing boards. With musicians and producers that, sometimes, like them in different places ;-)
The response curve of a speaker tell very little about the *nature* or *character* of the sound it produce. You can equalize, you will not change-it.
On a technical point of view, I consider that a speaker works better when dumped. That the serial impedance of the cable and the coils of the passive filters, in serial with the amplifier, ruin its damping factor. Thus the added damping that brings the compensation network is, anyway, beneficial.
Personally, I think that any effort to improve the motor characteristics of a driver by choice of a drive impedance profile should be kept low, it still is just a workaround of sorts. I use it to enhance cheaper drivers which often have significant Le(x) variation (high impedance helps) and to adjust the system resonance Q (sealed and ported), typically when the driver itself has too much internal feedback (low Qes). Also AMT tweeters are excellent when current driven, gives less distortion at HF (not always near low freq near cutoff, though) and takes power compression completely out of the picture which can be severe with smaller models only withstanding 10W or so steady-state.
Klippel has developped a more clever unit that adaptively models a driver including the nonlinear stuff and is able to pre-distort the drive signal to reduce distortion, it measures driver voltage and current. It also constantly monitors the "health status" of a driver, activates limiters etc.
Controlled Sound
I don't kow if any actual implementations yet exist, that is, if the unit is actually available (to OEMs, at least).
For HiFi/domestic use, it would be nice if driver manufacturers would offer drivers with integrated sensors for a true motion control. A nice linear velocity sensor isn't that hard to implement (sense coil in extra magnet field, orthogonal to the main coil, humbucking config, shielded). (Sub-)Woofers might be better off with excursion sensors and there would be ways to manufacture those conveniently (a RF capacitive approach similar to Sennheiser MKH condensor mic technology should give excellent results). Actually, both types could be implemented in a driver, at reasonable cost. A third sensor, this time acceleration type, could be added as well. By having sensors in all three domains and adding their signals one could have maximum output and dynamic range at any frequency, incl. DC.
Klippel has developped a more clever unit that adaptively models a driver including the nonlinear stuff and is able to pre-distort the drive signal to reduce distortion, it measures driver voltage and current. It also constantly monitors the "health status" of a driver, activates limiters etc.
Controlled Sound
I don't kow if any actual implementations yet exist, that is, if the unit is actually available (to OEMs, at least).
For HiFi/domestic use, it would be nice if driver manufacturers would offer drivers with integrated sensors for a true motion control. A nice linear velocity sensor isn't that hard to implement (sense coil in extra magnet field, orthogonal to the main coil, humbucking config, shielded). (Sub-)Woofers might be better off with excursion sensors and there would be ways to manufacture those conveniently (a RF capacitive approach similar to Sennheiser MKH condensor mic technology should give excellent results). Actually, both types could be implemented in a driver, at reasonable cost. A third sensor, this time acceleration type, could be added as well. By having sensors in all three domains and adding their signals one could have maximum output and dynamic range at any frequency, incl. DC.
Last edited:
On this current-drive subject I'll try once again to be terse, my experience is germane to the discussion:
In the 1990s I shopped the design around for a variably current-mode enhanced amplifier adapter integrated with a 2-way JBL studio monitor. During development but before any current-mode work, I realized it was necessary to stabilize the impedance of the speaker (duh). I tried several ways: network at the speaker terminals, individual networks on each driver, and found the second approach by itself subjectively improved the program-dependent SQ to my ears. I didn't spend a lot of time analyzing it since that was not the design goal, but attributed it to stabilizing the crossover frequency. If I remember correctly the inductance of the LF driver in particular varied +/- 20% as it moved through its excursion, so I figured the crossover frequency was being subsequently modulated as well.
I moved on to integrate the current-mode drive and liked the improvement I heard from that as well. However, I would say the overall improvement from start to finish was largely attributable to reworking the crossover and stabilizing the impedance. This was many moons ago so I will not try to describe the overall improvement, but I do remember thinking the dynamics were improved and the impedance stabilization seemed to "clean up" the sound somewhat at higher volumes. Specifically I remember drums having more impact and sounding more natural.
I am sure better information to implement current drive is available now, my experiments were 15 years before the Meriläinen book...
As along as I am revealing myself to be a dumbass again, I'll relate that one of the firms I demoed it to ended up copying the design and made a nice amount of money despite signing a non-disclosure/non-compete. This was one of several cases where I found these agreements are only as good as the $$ you have to spend on lawyers.
Cheers,
Howie
In the 1990s I shopped the design around for a variably current-mode enhanced amplifier adapter integrated with a 2-way JBL studio monitor. During development but before any current-mode work, I realized it was necessary to stabilize the impedance of the speaker (duh). I tried several ways: network at the speaker terminals, individual networks on each driver, and found the second approach by itself subjectively improved the program-dependent SQ to my ears. I didn't spend a lot of time analyzing it since that was not the design goal, but attributed it to stabilizing the crossover frequency. If I remember correctly the inductance of the LF driver in particular varied +/- 20% as it moved through its excursion, so I figured the crossover frequency was being subsequently modulated as well.
I moved on to integrate the current-mode drive and liked the improvement I heard from that as well. However, I would say the overall improvement from start to finish was largely attributable to reworking the crossover and stabilizing the impedance. This was many moons ago so I will not try to describe the overall improvement, but I do remember thinking the dynamics were improved and the impedance stabilization seemed to "clean up" the sound somewhat at higher volumes. Specifically I remember drums having more impact and sounding more natural.
I am sure better information to implement current drive is available now, my experiments were 15 years before the Meriläinen book...
As along as I am revealing myself to be a dumbass again, I'll relate that one of the firms I demoed it to ended up copying the design and made a nice amount of money despite signing a non-disclosure/non-compete. This was one of several cases where I found these agreements are only as good as the $$ you have to spend on lawyers.
Cheers,
Howie
Indeed.. and at least one person I know of is researching ultra stiff non flexing speaker pistons using the latest materials spec (a sponsored PhD thesis) and improved methods for moving same. Put all that together with the work of klippel et al, and the future looks good for serious audio improvements!
(When I typed this I expected it to follow KSTR's post...)
(When I typed this I expected it to follow KSTR's post...)
Sooo you bought a DSP?
I think when things go wrong is when someone assumes they can entirely correct the room. It looks good on paper, or a screen, but when you hear it you're like 😕😕 . I think they're more valuable for how they deal with crossovers and speaker driver variances, than the room. And of coarse if you have a crazy speaker like the Kii3 then you can make it run Cardioid so the room is less of a concern to begin with.
DSP in the software domain makes the most sense to me. That way you can use an ESS DAC. The problem is if you like vinyl... Then you probably should be looking at some high end DSP like Hypex.
That could be argued forever with no resolution, depending on what you mean by equalize we could end up with the bit perfect files that sound different.
Destroyer, MiniDSP outputs to I2S, so you can pipe it into plenty of high end DAC boards. Capability wise, a RPi3 or similar can run circles around it, but you've got to know what you're doing AND make your own drivers, which isn't for everyone.
No need. You can use REW, rePhase to make a filter, make impulse response and use it in convolution plug-in in foobar. This way you may easily create an inverse plot for the current drive. Been there, but the result is not worth the effort.
- Status
- Not open for further replies.