EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Jebnor on December 30, 2014, 10:12:50 pm
-
I was just wondering why this isn't the case. It seems that in most lab equipment has sense lines, from what I understand, to maintain linearity of the amplifier/supply. Wouldn't a sense line help when the speaker load may change from 4Ohm to 8ohm or 16ohm depending on configuration. (not dynamically obviously, but from setup to setup) Wouldn't this help reduce noise/THD/ other measurable effects?
-
Well, the sense lines aren't typically to maintain linearity per se, just accuracy, and if the speaker load is heavier and increases the loss in the lines, that should at least be relatively proportional. It doesn't matter if the amplitude delivered isn't perfect, since the signal is being transmitted into ears (which tend not to have insanely high vertical accuracy anyway).
Also, audio amplifiers are tricky enough to stabilize as it is, and sense lines complicate that significantly.
-
Most solid state power amps have so low of output impedance it doesn't matter, practically speaking, if it is driving a 4/8/16ohm load. Below 4ohm could be probamatic.
Still, PA amps may have that.
-
If the load has changed, then the speakers have changed, and probably their placement/aiming, and in the case of PA amps, likely the venue and event needs, so what would be the point? Even if just the speakers have changed and nothing else, chances are you'd need to readjust volume/EQ anyway to compensate for different speaker performance. Maybe if you were building an amp for some sort of really precise measurement setup.
-
Wouldn't a sense line help when the speaker load may change from 4Ohm to 8ohm or 16ohm depending on configuration. (not dynamically obviously)
Why not? The dynamic impedance of a speaker varies far more dramatically than that in its normal operating range. A "4 ohm" speaker is just a nominal 4 ohm, the DC resistance is usually more like 3.2, the spike at resonance can be upwards of 40+, and the high frequency rise due to inductance can bring it back up into the 20+ ohm range.
Power amps must have a very low output impedance (often called "damping factor" in the audio world) in order to maintain [roughly] the same voltage at the speaker terminals regardless of frequency, despite the widely varying impedance. Big speaker cables are required to keep the damping factor/output impedance high/low as well. If your impedance is too high (DF too low), you'll get a decrease in output when the speaker impedance is low, and an increase in output when the speaker impedance is high. This can cause a sharp increase in output around the speaker's Fs, and an increase in output at high frequencies.
That said, amplifier output impedance is generally SO low, and cable runs are usually short enough and of adequate size, that this is rarely an issue.
-
Considering how messy a speaker is, and how relatively well-behaved a pile of cable is, it's not a problem.
A servo against the motion of the speaker itself would be quite interesting! Perhaps a right pain to compensate and stabilize, but that would fix all of your nonlinearity and most of your frequency response problems.
Supposedly, once upon a time, there was a laser range finder servoed speaker, or something like that, which was reported to sound amazing.
Tim
-
A servo against the motion of the speaker itself would be quite interesting! Perhaps a right pain to compensate and stabilize, but that would fix all of your nonlinearity and most of your frequency response problems.
Hmm. This idea occurred to me a while ago, but I came to the conclusion that "right pain" would be a massive understatement. Now I'm intrigued again...
-
A servo against the motion of the speaker itself would be quite interesting!
In USSR there was active speaker system Radiotehnika S-70 with electro-magnetic negative feedback.
Sensor within low frequency speaker (can be electro-magnetic or piezo sensor) with output going back to PA.
Combined with current-source amplifier (not usual voltage-source) it can deliver something interesting.
Here is an article on Russian (http://reanimator-h.narod.ru/emos.html), but at least you can look at some pictures.
-
is that piezo (in russian) page talking about the same acceloro feedback?
Something like that, yes.
Main idea: measure acceleration of the cone, cancel out everything else (magnetic noise, air flow issues, resonances of the sensor).
Sensor must weight as little as possible (to prevent changing the mass of the cone - or parameters of the speaker will change) but must be very rigidly connected to the cone (to provide best frequency/phase response).
Then filter signal in some way and inject it into feedback signal.
From what I can see in my link, that article was based on some others, from 1970-s (see the link list at the bottom, you can see dates there).
-
Sometimes ago, there was this notion of "current feedback" to driver speakers. Essentially, the feedback is based on the current going through the speakers / load. So the typical resistive divider is replaced with a speaker over a current sampling resistor and the feedback is based on the voltage across the sampling resistor, thus current through the speaker.
It was said to have the advantage of fast / accurate sound reproduction. I found it to be fatiguing to listen to and exaggerating the high frequency content.
-
Sometimes ago, there was this notion of "current feedback" to driver speakers. Essentially, the feedback is based on the current going through the speakers / load. So the typical resistive divider is replaced with a speaker over a current sampling resistor and the feedback is based on the voltage across the sampling resistor, thus current through the speaker.
Yep. And my link tells about it as well. Amplifier is controlled by current through speaker + signal from acceleration sensor.
In Russian we have two terms: "INUN" and "ITUN", you can find them in that article. EEVBlog is incompatible with cyrillic symbols so I can not provide exact terms...
"INUN" means "voltage source, controlled by voltage" (typical power amplifier schematic).
"ITUN" means "voltage source, controlled by current" (feedback based on current through speaker).
If we look from output impedance perspective:
"INUN" has near to zero output impedance.
"ITUN" has near to infinite output impedance.
Sounds weird, because high output impedance means that with rising impedance of the speaker (vs. frequency) voltage will rise as well... But current feedback takes care of it.
-
It seemed to me that if one were trying to produce a 'perfect' speaker, what you'd want would be feedback from a sensor for air pressure at the face of the moving surface (cone, plane, whatever.)
Since the recorded signal is supposed to represent traveling pressure waves in air.
But with standard amps and speakers, the signal just gets applied as an amplified voltage at the output terminals of the amp. Then the speaker has inertial and back-emf effects related to its motion, so the cone motion isn't a duplicate of the recorded signal - neither offset or velocity. Plus even if it was, that still won't produce an identical series of pressure waves in air, since the speaker cone movement doesn't translate exactly to pressure.
Anyway... I'd guess the *main* reason stereo amps don't have sense lines, is that this would be far too complicated for average consumers to connect up. With two wires there's really no way to get it totally wrong (just in-phase or anti-phase), but with four per speaker... I wouldn't want to be on a customer help phone line.
-
A servo against the motion of the speaker itself would be quite interesting! Perhaps a right pain to compensate and stabilize, but that would fix all of your nonlinearity and most of your frequency response problems.
Philips did this in the early 70's : http://en.wikipedia.org/wiki/Motional_Feedback (http://en.wikipedia.org/wiki/Motional_Feedback)
The amplifier is built into the box, so the feedback is simple. The (piezo I think) sensor is mounted in the middle of the cone of the low freq. speaker.
I still have a set of those...
-
I was just wondering why this isn't the case. It seems that in most lab equipment has sense lines, from what I understand, to maintain linearity of the amplifier/supply. Wouldn't a sense line help when the speaker load may change from 4Ohm to 8ohm or 16ohm depending on configuration. (not dynamically obviously, but from setup to setup) Wouldn't this help reduce noise/THD/ other measurable effects?
An audio amplifiers job is to amplify an audio signal independently from its speaker load..expensive amplifiers use some interesting signal analysis that sit in the audio signal path performing pre-processing using a DSP using various audio algorithms from simple to complex to company patented stuff.
A sense system that assists with output control of a loudspeaker means, the system has to perform electrical and mechanical analysis (that ages, like TS parameters) even worse if its a one time setup where environments change and the acoustic properties changes with it.
My point is this, separate independent component based systems like an audio amplifiers will never try and "FIX" a native loudspeaker as both are trying to achieve "perfection" independently from each other. However as shown above by @PA0PBZ, with the Philips system, some "complete" audio systems do use these techniques internally.
-
Maybe a year ago, I've played with the idea. A 8 Ohm speaker with 3m 10 AWG cable, the cable is 2% the resistance of the speaker. This has the potential to make audible differences, at least that is what I believe. And I dont have a good reason why there isnt four wire for audio. This could be one of those things that never got improved, because everyone takes them granted, like cars or furniture. http://www.nandblog.com/wire-audio/ (http://www.nandblog.com/wire-audio/)
-
because high output impedance means that with rising impedance of the speaker (vs. frequency) voltage will rise as well..
The set-up basically turns a voltage source to a current source -> the output signal is current.
Tube amps, with its higher output impedance, are largely a current source -> much lower dampening factor. Thus, the rationale behind some audiophiles adding a 120ohm resistor in serial with its output of their headphone amps. The resistance is far bigger than the load so it acts like a current source. It also lowers the dampening factor significantly.
-
I was just wondering why this isn't the case. It seems that in most lab equipment has sense lines, from what I understand, to maintain linearity of the amplifier/supply. Wouldn't a sense line help when the speaker load may change from 4Ohm to 8ohm or 16ohm depending on configuration. (not dynamically obviously, but from setup to setup) Wouldn't this help reduce noise/THD/ other measurable effects?
Hi,
There are various practical problems that come up when trying to do this. The most serious i think is the distance from the speaker to the listener, which is almost the same problem as viewing a TV monitor from different angles for different viewers.
There is no problem however with controlling the speaker itself, as that would be viewed as a simple servo system, of course it has to be fast though. This would involve feedback just like any other system, and the correction would come in the form of a changed drive signal from the now controlled audio amp. This should be able to correct for various flaws in the speaker, as long as the speaker can handle the increased peak power that would have to be necessary in order to make these corrections.
I had thought about this back in the 1980's but did not know then that there was something similar already on the market. My idea was totally different however so i cant give it out right now.
Another idea (apart from above) was to simply have high quality microphones placed around the room, and put the amplifier and speakers (as positioned at the time) through a series of calibration tests and adjustments. This could be done automatically, but it would have to be repeated any time the speakers are moved.
A problem then comes up as to what happens when someone new walks into the room and changes the dynamics :-)
-
The biggest problem of equalizing the speaker frequency response is that speaker does not have just one well-defined frequency response, but infinitely many different responses since it radiates different frequency response with different dips and peaks to each direction. Since all rooms are usually more or less reverberant, consequence is that all other frequency responses are also altered.
Net effect of this is that if one tries to make the frequency response flat, overall response in a room will get very bad due to the room reflections. It is nowadays trivial to equalize a speaker to have perfectly flat and linear phase response at one single point in room via FIR equalizer, just measure the impulse response of the speaker in the room, calculate inverse correction and apply that to the electrical signal. That would correct all linear distortions, either speaker or speaker-room interaction related.
While this sounds good in theory, in practice it will sound horrible due to the strong dependency of the frequency response by the location. See Bruno Putzeys excellent PDF on various audio things, page 101: http://www.hypex.nl/docs/papers/AES123BP.pdf (http://www.hypex.nl/docs/papers/AES123BP.pdf)
There are other interesting topics on that PDF, too.
Regards,
Janne
-
Really High End Kilowatt and Up Audio amps for concerts have sense lines. Its used to compensate for cable drop. I've seen what happens when a sense line gets lifted. This happened at really high end VIP party for a cities centennial. Ever seen what happens when 1000 folks in Tuxes and Open backed dresses get hit with two Kw of Endless Polka?
Five minutes in, they were climbing the truss to try and kill the speakers. In Tuxes.
Sound guy was old and deaf, we were screaming on the radio, but he wouldn't kill the feed.
Generally when your sound stack needs a three phase feed, you'll find sense lines on the Bass.
Steve
-
The biggest problem of equalizing the speaker frequency response is that speaker does not have just one well-defined frequency response, but infinitely many different responses since it radiates different frequency response with different dips and peaks to each direction. Since all rooms are usually more or less reverberant, consequence is that all other frequency responses are also altered.
Net effect of this is that if one tries to make the frequency response flat, overall response in a room will get very bad due to the room reflections. It is nowadays trivial to equalize a speaker to have perfectly flat and linear phase response at one single point in room via FIR equalizer, just measure the impulse response of the speaker in the room, calculate inverse correction and apply that to the electrical signal. That would correct all linear distortions, either speaker or speaker-room interaction related.
While this sounds good in theory, in practice it will sound horrible due to the strong dependency of the frequency response by the location. See Bruno Putzeys excellent PDF on various audio things, page 101: http://www.hypex.nl/docs/papers/AES123BP.pdf (http://www.hypex.nl/docs/papers/AES123BP.pdf)
There are other interesting topics on that PDF, too.
Regards,
Janner
It can sound exellent in practice, too, assuming tje valibration algorithm really understands room acoustics. The Steinway brand has what they call RoomPerfect. It takes multiple measurements, tames the room modes, does not try to compensate local nulls etc. But then, at those prices it better be realy good.
Also systems that try to eq only bass section and only to tame resonances (never apply gain, as any dips are likely local nulls) in general do more good than harm.
-
Kenwood had Sigma Drive in the 1980s-1990s.
-
Because usually it will cause more trouble than advantage. Long wires in the fast feedback loop generally is not a very good idea. But compensating in DC makes no sense.