KeithL
Administrator 
Posts: 10,261
|
Post by KeithL on Jan 2, 2018 14:47:19 GMT -5
It's actually relatively simple to design an amplifier whose output CURRENT rather than its output voltage is proportional to its input voltage. The problem is that, in practice, they don't work very well - at least not with current speakers. One of the problems, as I detailed in another post, is that current drive cannot compensate for variations in mechanical properties. Another less obvious one is that a current drive amplifier, by its nature, has total control over the current passing through the speaker, but no control over the voltage. However, because of this, it is dissociated from effects that rely on controlling the voltage... and one such effect is damping. So, while a perfect voltage amplifier would have an infinite damping factor, a true current source has an infinite source impedance, so a current source amplifier has a damping factor of ZERO.... With no signal, a normal voltage amplifier acts like a short circuit across the speaker terminals. Back EMF causes current to flow through this short circuit, and that current then creates force that opposes the original motion.... we call this entire process "damping". Because a current source amplifier acts as an OPEN circuit, back EMF will cause no current to flow, so no opposing force will be generated.... so NO electro-mechanical damping will occur. (A voltage source amplifier provides damping by sinking current to cancel out back EMF; a current source amplifier will not do so... because its output current is determined SOLELY by its input signal voltage.) It might be interesting to try this with a headphone amplifier - since most headphones use a single driver... and many have sufficient mechanical damping to not need electrical damping. (It would be relatively simple to reconfigure a headphone amplifier for "current output".) I am getting used to your responses of no, false, and wrong to almost anything I write. If an amplifier driving a single dynamic coil type speaker were to be current regulated, then the varying impedance of the speaker would be a non issue. This is all I am saying. Sorry, not going after you, just striving for accuracy. The varying impedance of a loudspeaker is a non-issue with any amplifier, mainly because it cannot be otherwise. If you desire constant power into any load, the regulation you need is in the amplifier power supply not in the outputs, which is one reason why the newest Emotiva P/S designs are interesting. |
|
|
|
Post by mgbpuff on Jan 2, 2018 15:01:30 GMT -5
I've got to comment on your last statement..... because, even though it seems to make sense, it is actually far from true in practice. The VERY BASIC theory that suggests it would make sense goes as follows... - The force applied to the speaker cone depends on the current through the voice coil and the magnetic field strength. - Therefore, if we used a current drive amplifier (where the output current depends on the input voltage) the current would be unaffected by things like resonance. - So, since we've removed resonance and other frequency effects from the equation, our speaker should be more linear. Now, here's why that idea doesn't really usually work... - All speakers have resonances, and usually a single major primary resonance. - At the resonant frequency both the impedance and the mechanical efficiency go up. - With our current drive amplifier the current will remain constant (proportional to the input voltage) - even at the resonant peak. - As a result, the drive force will remain constant at the resonant frequency (which seems like a good thing... except...) - Because the drive force is the same at the resonant frequency, BUT THE MECHANICAL EFFICIENCY IS HIGHER, there will be a peak in the mechanical output (audio). - In contrast, with a normal voltage drive system, at resonance, while the mechanical efficiency goes up, the drive power goes down, and the two tend to partially cancel out. A normal voltage drive amplifier delivers less current, and so less power, at the resonant frequency; a current drive amplifier delivers the same current, and so MORE voltage, and so MORE power, at resonant frequency. However, the mechanical efficiency is higher at the resonant frequency; the voltage drive amplifier tends to partially cancel out this effect, while the current drive amplifier potentiates it. As a result, at least with most real-world speakers, a current-drive amplifier will actually produce a worse peak in the acoustic output at resonance. I am getting used to your responses of no, false, and wrong to almost anything I write. If an amplifier driving a single dynamic coil type speaker were to be current regulated, then the varying impedance of the speaker would be a non issue. This is all I am saying. What you are describing is a non linear relationship between coil movement and coil current due to a mechanical efficiency change in the transducer at resonance. Sounds reasonable to me, because the desirable controlled variable is displacement and not current. A shaped compensating circuit allowing for the resonant band could possibly be developed, but now we are into filters and cross overs and differences between speakers. The real solution will be to close loop control actual coil movement with a displacement sensor. We seemed to be doomed in this audio reproduction business to tolerating open loops and tailoring such open loop systems to get the best results, but varying conditions mean we well never get consistence results until we control the right variable. |
|
klinemj
Emo VIPs 
Official Emofest Scribe
Posts: 15,088
|
Post by klinemj on Jan 2, 2018 15:52:22 GMT -5
"Are we there yet!!!!?!?!?!?!"
Mark
|
|
|
|
Post by pedrocols on Jan 2, 2018 16:04:37 GMT -5
I've got to comment on your last statement..... because, even though it seems to make sense, it is actually far from true in practice. The VERY BASIC theory that suggests it would make sense goes as follows... - The force applied to the speaker cone depends on the current through the voice coil and the magnetic field strength. - Therefore, if we used a current drive amplifier (where the output current depends on the input voltage) the current would be unaffected by things like resonance. - So, since we've removed resonance and other frequency effects from the equation, our speaker should be more linear. Now, here's why that idea doesn't really usually work... - All speakers have resonances, and usually a single major primary resonance. - At the resonant frequency both the impedance and the mechanical efficiency go up. - With our current drive amplifier the current will remain constant (proportional to the input voltage) - even at the resonant peak. - As a result, the drive force will remain constant at the resonant frequency (which seems like a good thing... except...) - Because the drive force is the same at the resonant frequency, BUT THE MECHANICAL EFFICIENCY IS HIGHER, there will be a peak in the mechanical output (audio). - In contrast, with a normal voltage drive system, at resonance, while the mechanical efficiency goes up, the drive power goes down, and the two tend to partially cancel out. A normal voltage drive amplifier delivers less current, and so less power, at the resonant frequency; a current drive amplifier delivers the same current, and so MORE voltage, and so MORE power, at resonant frequency. However, the mechanical efficiency is higher at the resonant frequency; the voltage drive amplifier tends to partially cancel out this effect, while the current drive amplifier potentiates it. As a result, at least with most real-world speakers, a current-drive amplifier will actually produce a worse peak in the acoustic output at resonance. What you are describing is a non linear relationship between coil movement and coil current due to a mechanical efficiency change in the transducer at resonance. Sounds reasonable to me, because the desirable controlled variable is displacement and not current. A shaped compensating circuit allowing for the resonant band could possibly be developed, but now we are into filters and cross overs and differences between speakers. The real solution will be to close loop control actual coil movement with a displacement sensor. We seemed to be doomed in this audio reproduction business to tolerating open loops and tailoring such open loop systems to get the best results, but varying conditions mean we well never get consistence results until we control the right variable. So What is the "right variable?" |
|
|
|
Post by jcz06 on Jan 2, 2018 16:10:40 GMT -5
In the end, it comes down to whatever sounds better to your ear....
|
|
DYohn
Emo VIPs
Posts: 18,487
|
Post by DYohn on Jan 2, 2018 16:11:08 GMT -5
In the end, it comes down to whatever sounds better to your ear.... Which is I think where we started... |
|
|
|
Post by mgbpuff on Jan 2, 2018 16:14:41 GMT -5
What you are describing is a non linear relationship between coil movement and coil current due to a mechanical efficiency change in the transducer at resonance. Sounds reasonable to me, because the desirable controlled variable is displacement and not current. A shaped compensating circuit allowing for the resonant band could possibly be developed, but now we are into filters and cross overs and differences between speakers. The real solution will be to close loop control actual coil movement with a displacement sensor. We seemed to be doomed in this audio reproduction business to tolerating open loops and tailoring such open loop systems to get the best results, but varying conditions mean we well never get consistence results until we control the right variable. So What is the "right variable?" speaker displacement using a sensor such as a laser for feedback |
|
|
|
Post by gsand on Jan 2, 2018 16:33:19 GMT -5
In the end, it comes down to whatever sounds better to your ear.... I agree! I hope this thread ends soon |
|
|
|
Post by leonski on Jan 2, 2018 16:35:15 GMT -5
I wonder, and have a suggested 'test'. First? I wonder how the low resistance of an amplifiers output stage translates to damping? It would seem that the amp than acts as a passthru so the VOICE COIL of the speaker, 'self damps' when presented with the counter EMF it generates. A Test? Disconnect your speaker from the amp. Pull the grill cover. 'Flick' the speaker cone edge (woofer) with a finger. Take a Paper Clip and bridge the speaker terminals. Repeat the finger flick. Cone is now VERY WELL damped. The RDon of an output device will typically measure in the miliohm range. Not really high enough to provide much 'damping'. The speaker basically damps ITSELF. High output impedance amps (say, those with output transformers) will suck up the back EMF before it gets back TO the speaker. A couple other points. DY published a graph of speaker impedance. I see TWO lines, and a scale marked in DEGREES. Why do we continue to avoid the role of reactance in a speakers 'goodness of load'?? Given a reasonable sensitivity, and low reactance, even a speaker hovering at 4 to 5 ohms will be an EASY load for any reasosnable amp. Get an impedance minima of 3 ohms or perhaps a little less WITH a phase angle of over 50 degrees, and you'll start seeing differences between amps. And my final question has never been answered. What IF the guys emo decided to make an amp using a KRELL or ARC or MC schematic? Would it sound as good or just different? Would reliability suffer? Would such an amp be more or less prone to pick up environmental noise? How much $$ would such a creation sell for relative to the parent amp? And how would it measure against the parent amp? |
|
|
|
Post by Talley on Jan 2, 2018 16:35:21 GMT -5
So What is the "right variable?" speaker displacement using a sensor such as a laser for feedback Doesn’t rythmik use a Servo controlled speaker/amp closed circuit on their subs.... |
|
|
|
Post by leonski on Jan 2, 2018 16:39:37 GMT -5
speaker displacement using a sensor such as a laser for feedback Doesn’t rythmik use a Servo controlled speaker/amp closed circuit on their subs.... Just a funny way of saying 'Feedback'. This opens the can-o-worms that is NFB. |
|
|
|
Post by rtg97229 on Jan 2, 2018 17:02:56 GMT -5
I was dissed' for 'suggesting' that MC had a 'house sound'. I suspect that people misunderstood your intent in pointing out something that can be shown to be true. If people enjoy the sound it is not a bad thing. |
|
|
|
Post by mgbpuff on Jan 2, 2018 17:07:10 GMT -5
Doesn’t rythmik use a Servo controlled speaker/amp closed circuit on their subs.... Just a funny way of saying 'Feedback'. This opens the can-o-worms that is NFB. That can is mislabeled! |
|
|
|
Post by mgbpuff on Jan 2, 2018 17:11:31 GMT -5
Doesn’t rythmik use a Servo controlled speaker/amp closed circuit on their subs.... Just a funny way of saying 'Feedback'. This opens the can-o-worms that is NFB. Yes, a servo, which can be position controlled moves the diaphram instead of a magnetic voice coil. The band width is limited in a sub so this can work well. But it is a built in feature and needs (I think) a special amp. |
|
|
|
Post by leonski on Jan 2, 2018 17:34:42 GMT -5
What IS it called when the output either substantially influences or controls the input?
|
|
|
|
Post by mgbpuff on Jan 2, 2018 17:46:55 GMT -5
What IS it called when the output either substantially influences or controls the input? Input is not affected by feedback. Output is affected by feedback. And if correctly compensated, feedback improves the linearity and the accuracy of the input to output amplification function and eliminates external influences that may try to affect the output also. It is not a can of worms if engineered correctly. |
|
|
|
Post by leonski on Jan 2, 2018 18:54:08 GMT -5
I need to be more elaborate. In both a 'servo' and 'feedback' system, it would appear that a signal from the output is fed BACK to the input. You can generate this signal either electrically or use an accelerometer or even modulate a laser reflecting off the speaker. www.electronics-tutorials.ws/systems/negative-feedback.htmlHere's an article which is pretty easy and undertandable. But than you need to find the 'Anti-NFB' guys and listen to what THEY have to say. I've read where Bruno P., the head-guru of the nCore amps, uses LOTS of feedback which is claimed to be better than Too Little feedback. Can-O-Worms? You should read the huge disagreements a simple discussion of feedback, by stage or global, amount and such starts. Some sites even lump talk of NFB in with DBT and a few other 'hot button' issues in order to keep printed bloosshed to a minimum. |
|
|
|
Post by mgbpuff on Jan 2, 2018 19:15:03 GMT -5
I need to be more elaborate. In both a 'servo' and 'feedback' system, it would appear that a signal from the output is fed BACK to the input. You can generate this signal either electrically or use an accelerometer or even modulate a laser reflecting off the speaker. www.electronics-tutorials.ws/systems/negative-feedback.htmlHere's an article which is pretty easy and undertandable. But than you need to find the 'Anti-NFB' guys and listen to what THEY have to say. I've read where Bruno P., the head-guru of the nCore amps, uses LOTS of feedback which is claimed to be better than Too Little feedback. Can-O-Worms? You should read the huge disagreements a simple discussion of feedback, by stage or global, amount and such starts. Some sites even lump talk of NFB in with DBT and a few other 'hot button' issues in order to keep printed bloosshed to a minimum. I'm talking about feedback in the modern day and age, mainly with OP amps. An OP amp without feedback is unusable (except for digital circuitry) because of it's high gain approaching infinite. So if FB reduces gain, that is only good. The principles of feedback are mathematical in nature, fb is not snake oil. FB and it's effects are calculable and controllable. I don't read ANY of the anti-feedback stuff out there because it is non scientific in nature. Good ole boys don't understand it! |
|
|
|
Post by leonski on Jan 2, 2018 20:04:20 GMT -5
I built a CMOY opamp headphone amp and of course, it has feedback. The ratio of a couple resitors is the 'gain'. Simple calculation that if you can add 2 @2 digit numbers,
you can do it without paper or pencil.
I don't, however, know of a single power amp with an opamp output stage. And yes, MOST (90%+, at least) of all modern amps do have some form of feedback
Either global, from output to input OR, as Nelson Pass advocates, a stage-at-a-time.
I'll find something that'll pass the 'scientific smell test' from someone who is not quite fully on-board with feedback as applied to modern designs.
One aspect of NFB that pops up is one of signal latency. The signal generated at the output is a finite time after introduction at the input. By the time the correction signal
gets back to the input, it is NO LONGER precicesly in phase with the signal it is intended to correct. It would appear that in so doing, LOT of low level harmonics and such are generated.
|
|
|
|
Post by pknaz on Jan 2, 2018 21:14:11 GMT -5
Get an impedance minima of 3 ohms or perhaps a little less WITH a phase angle of over 50 degrees, and you'll start seeing differences between amps. In my humble opinion, this is an example of a speaker that was poorly designed. I'd ask why someone would want such a poorly designed speaker to begin with?  |
|
I am getting used to your responses of no, false, and wrong to almost anything I write. If an amplifier driving a single dynamic coil type speaker were to be current regulated, then the varying impedance of the speaker would be a non issue. This is all I am saying. 
