The situation I was talking about last page is the possibility that in a high power transient, say so loud the woofer is at Xlim, the drive signal is suddenly off, leaving the bias impedance, the woofer snaps back to neutral position, there is a possibility of the back EMF and generated EMF pushing current into the feedback resistor network feeding one side of the input differential amp, and now the diff amp would be improperly balanced for some sudden new signal.
I normally work out situations for that with robotic motors. The feedback loop is governed by the PID formula, output drive = Sigma (Proportional error + Differential error + derivative error). If the PID time dependent coefficients are not balanced for the range of transient errors, then the system gets out of sync and can do several erroneous things such as oscillating position and position signal, and as bad as slamming the drive switching transistors with a full slam of back EMF, which can cause double the full current drive through the pass transistors. On unprotected systems without transistors rated to absorb that condition, they are destroyed nearly in an instant. I found DC drives very vulnerable to that.
I had to assume the energy to position the cone to Xlim may be 200 watts, and now at that position the voice coil system (= motor) would have 200 watts stored and ready to dissipate if it snaps back to neutral suddenly.
Crazy complicated isn't it ?
I normally work out situations for that with robotic motors. The feedback loop is governed by the PID formula, output drive = Sigma (Proportional error + Differential error + derivative error). If the PID time dependent coefficients are not balanced for the range of transient errors, then the system gets out of sync and can do several erroneous things such as oscillating position and position signal, and as bad as slamming the drive switching transistors with a full slam of back EMF, which can cause double the full current drive through the pass transistors. On unprotected systems without transistors rated to absorb that condition, they are destroyed nearly in an instant. I found DC drives very vulnerable to that.
I had to assume the energy to position the cone to Xlim may be 200 watts, and now at that position the voice coil system (= motor) would have 200 watts stored and ready to dissipate if it snaps back to neutral suddenly.
Crazy complicated isn't it ?
This is true. In a good quality audio system the bass is still very nice at low volumes and doesn't need compensation.
Feedback restores bass frequency at lower amp power (extends frequency range) but it cannot do anything of this sort when up to a certain power output, meaning it will play worst than a no feedback amp with added distortion up to a certain power level. That is why it is an essential help to an amplifier quality but not a remedy to a poor amp. (ie. over 20 db sort of) Most solid state have over 25 db relative feedback which is very poor sounding no matter what.
Well a woofer doesn't really 'snap back' - it has a natural resonance that is quite low in frequency, so as far as snapping is concerned, it's more of a leisurely return to the mid position.
In addition, the drive signal to the woofer cannot be suddenly shut off; the cross-over filter will route any transients to the mid- or tweeter.
I understand that a coil moving in a mag field generates a signal - call it back EMF - but I am at a loss how this could push anything into the feedback network. The feedback network consists of two resistors in series, one end at ground, the other end at Vout (disregarding for the discussion the input impedance of the diff pair, which is reasonable compared to the feedback R values).
So how can the current into the feedback network be anything else than Vout/(sum of the two feedback Rs)???
Jan
Last edited:
A bass tone control that approximates the differences between the ISO-226:2003 equal loudness contours would be a rather more well-controlled means of correcting the bass at low volumes than reducing loop gain and hoping the resulting inaccuracies will enhance the bass by the correct amount.
Agree. A well-know audio guy (can't remember the name right now) made the case that stereo reproduction is aimed at creating a credible illusion in your living from two speakers. Often a very credible illusion, but an illusion nevertheless. It's not the real thing.
Jan
Energy unit is Joule, not Watt.
Back EMF is some voltage given under the impedance of the driver.
The typical dumping factor of a solid state amplifier is 50. Hence the back emf voltage is divided by 50 at the amplifier output. So the negative feed back network is not upset at all. There is no point to imagine havoc propagated by the feed back network, things stop at the amplifier output, damped by the very low output impedance. A damping factor 50 means an output impedance of 0.16 ohm for an amplifier designed for an 8 Ohm load.
Simple isn't it ?
Last edited by a moderator:
Hi,
Thanks You so much Jan
That is exactly what I meant in #73, "rather its´s error amplification with a gain of almost OL (that is global feedack)" ... and what Scott and cbdb seemingly missed to understand.
jauu
Calvin
In audio we are not used to calling the signal between the two input pins of the amp itself 'the error'. It's what it is called in control theory, and all those principles apply equally to audio of course.
But most audio guys are not familiar with their 'parent technology' so to say.
Jan
It has nothing to do with the amp or the audio system, but everything with your ears. The amp (if it is a decent one) will have a flat response down to low frequencies no matter what the level, and certainly at low levels.
The point is that your ears get less sensitive to low frequencies when the level drops. Google 'Fletcher and Munson'.
That's the reason for loudness control. The idea is to put a hump in the lf response at low levels to compensate the loss of hearing sensitivity in your ears.
Jan
Hi,
Thanks You so much Jan
That is exactly what I meant in #73, "rather its´s error amplification with a gain of almost OL (that is global feedack)" ... and what Scott and cbdb seemingly missed to understand.
jauu
Calvin
Ah, I did misunderstand, thanks for the clarification
Wavewhipper, I see where you are coming from. A servo motor feedback from the encoder is a lot slower and the motor has a mass that is not present in an amp feedback. Also, remember that the driver is outside the feedback loop and only seen by the amp as an error voltage and is corrected in real-time.
If the driver tries to generate a voltage above the amp output, the amp instantaneously flows a negative current that will prevent the voltage from appearing at the output of the amp. Voltage is controlled by current flow.
For me, where things fall apart is that we are controlling voltage and not current.
If the driver tries to generate a voltage above the amp output, the amp instantaneously flows a negative current that will prevent the voltage from appearing at the output of the amp. Voltage is controlled by current flow.
For me, where things fall apart is that we are controlling voltage and not current.
How so, Mark? Isn't current just a consequence of a voltage across an impedance?
Jan
It would be if the load is a resistor. We start with a voltage at the output that is in phase with the input, while the current has a frequency dependent phase shift. At a minimum, the waveform of the current is not going to be the same as the waveform of the input or output voltage.
What I am saying, is that if you want to look for an error in the way we use feedback, this might be the place to look, if only because we are looking for a method to control the movement of a driver.
Well, it is difficult to explain how the current can be negative while the voltage is positive at the output of an amp, as the current being the consequence of voltage.
Well, it is difficult to explain how the current can be negative while the voltage is positive at the output of an amp, as the current being the consequence of voltage.
If you mean that it makes more sense to feedback-control the current rather than the voltage, you have a point. People who tried current driving loudspeakers have achieved substantial reductions in loudspeaker distortion. Of course you have to find some other way to control the quality factor of the loudspeaker's fundamental resonance then (such as equalization, motional feedback or a thick cloth behind the speaker).
Well, it is difficult to explain how the current can be negative while the voltage is positive at the output of an amp, as the current being the consequence of voltage.
OK, I see what you mean now.
But what if you envision a capacitive load charged to a negative voltage?
You'd have more voltage sources in the circuit than the driving source, but a speaker can also act as a source so this is a similar situation, no?
Jan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.