Cortez said:But thats just one thing, in my view there are lot of similar areas to explore...
Greets !
it is a good thing to explore those similar areas but if they make zero audioable differences to you, you are far better off concentrating your exploration on areas that DO make an audioable difference to you.
Unless of course you have unlimited resources at your disposal.
Those people that are discussing this subject with you are not ignorant
A lot of highly qualified guys are following your thread, and facing this serious.... problems that are already studied decades ago, as you told, and you open a thread to discuss that.
They may be kind, helping you to clean your own doubts tlf
regards,
Carlos
A lot of highly qualified guys are following your thread, and facing this serious.... problems that are already studied decades ago, as you told, and you open a thread to discuss that.
They may be kind, helping you to clean your own doubts tlf
regards,
Carlos
I didnt say that, lets discuss only about the feedback, it was just one thing.
But we must to know about the drawback of it, cause when we apply it wrong,
can make more trouble, then it should in ideal case.
For eg the well known TIM effect is produced also that way.
When the FB is too high, transient, overloadings can be created localy.
And we sometimes cant measure these things at the output point, cause
the global FB doesnt allow this, but lets suppose, that a problem like
this, doesnt stay there, and we have to pay in quality for them.
Maybe in an indirect way, but we have.
Yes, but there is a phisical and a virtual capacitance.
The phisical is a fixed thing, lets use device with less parasite LC's,
but the virtual one depends on the electronical environment !
And thats our chance to go further in this way. A circuit can amplify
this effects, and therefore can causes more or less phase-shifting.
Eg everyone knows, that in a CommonEmitter configuration, this parasitic
capacitance is increased with the amplification.
If its 1pf and the gain is 100, it will be act as a 100pF capac!
Specially couse this setup inverts the sign, and this capac depends
on the voltage between the input and output. When it inverts, this
increase a little bit more this effect. And the situation is more
worse, cause a stage like this amplify the current and the voltage in
the same time. If we now add the above about the high FB, we can suppose
that our inner stage amplifies a lot, and hence this parasitic capac is
increased a lot again... This is a problem here, but we correct it
globally. In my view this is a wrong solution. We must to produce
as less similar trouble as possible, and then correct it locally,
and just then is a little global feedback neccesary to reduce global
distortion, but not for correct inner wrong configured stages.
And if we now add, that phenomenons like thermal distortion exists,
this is a more important view, cause if we look to a transistor, that
it is device with energy-storing capability, this added, stealed energy
will modify, modulate the operation of the transistor, like its happening
when the operational point is changing. Before this plus energy would
gone away, other new signal are comming, and would be amplified wrongly.
And an other thing: if there is a capacitance between the input and the
output, the load of the next stage can talk back, to our input, capacitive.
Why not ? There can be phase delay between of his two inputs.
You just showed, that a voltage divider doesnt shifts the phase,
but while the input is amplified, it gets a lot of delay through
the chain, so a newer signal is compared to the feddbacked old's.
But lets really not going to discuss only about the FB delays...
A late intervention can cause lot trouble, it corrects the new signal with the error
of the previous one. The signal will try to chase itself in this circle...
Really ? I dont think so. I belive (or know ?
, that we can achieve a better(smaller) phase delay. And we must to do this in open loop, and when we have success,
then should we feedback the signal. The feedback should just decrease the vertical
distortion, but not correct the inner stage's phase and time problems.
The phase is not a problem only for our ears. Rather its cause his things in the electronics.
And when we success to reduce the phase delay, it would be good not only cause it is small,
but cause it is farly equal in the whole audio range. (a few degrees)
Why couldnt these effect be audible ?! We have a lot of amp schematic, and the most of them
works indeed. Then why do not we go further, and develop things like this ? Ok who doesnt
have pleasure doing this, it isnt a constraint, but no one should say, that it counts nothing,
and its not audioble. At the start we must to belive in our aims. (And at the and we can
enjoy the results of that...
Really ? Have you ever heard a fully DC-coupled, absolute phase hiend, power amplifier ?
I dont think so. There is always things to develop. So lets do it!
Agree, there is a lot of thing that has been discussed, but in general peoples are content, when it
roughly operate, and doesnt have oscillations or so. Iam not talking about PA amps with x kW-s...
Lets assume a little (real!) hiend world. Why couldnt we hear things that i talked about above ?
Some of them can be measured with a skope, like a writed in the list. There is no limit,
there is a lot of thing, that we can measure, to enhance the developing process.
Obviously it counts a lot, what kind of part are we using, and how we create our circuits in reality.
But i doesnt mean to the base things ! In my view its not the end of the journey, when our amp doesnt
oscillate, or doesnt humm. There can be effects what causes problems specially transients based on that.
Miscellaneous cross-coupling phenomenons between the parts, cables, etc. They can modulate our original
signal and worse the feedback etc. I hope the list i writed is enough for starting. Be creative and explore
these levels, layers too ! Move the parts, cables, pcb wires related to eachother, rotate them, slap them,
move and roatete the entire amp and the pseudo-load too. With such methods, we can discover unwanted effects that exists and live together parasiticaly with us...
An other measure: short circuit the input, put a speaker to the output, and a scope to all kind of points.
Then slap the cone, and watch the result, the behaviour of your amp, and the feedback. This is a transient,
and the amp will try to correct the "failure", cause on the output the level should be 0, but with the slap
it is not 0, cause you added energy externaly.
I don't understand this. Lets view it bigger. Lets say the delay is not 30deg, but 30,000,000deg, or the music comes 5 minutes too late from the input signal.
There is no problem with this. Our ear will hear it as "normal" music, although it comes 5 minutes too late from the CD player (as long as left and right channel delayed by the same)
I'm not against feedback. Which part said that I don't like feedback?
Yes, that's it. Knowing more things about feedback is better than knot knowing. My current amp is using global feedback amp (but using 2 stages), and sounded better than my non-global-feedback amp. I just haven't succeeded in making feedback amp with more than 3stages. More than 3 stages+feedback sounds not nice. For 4 or 5 stages, I still prefer non-global feedback type.
Agreed.
Between left base and right base of differential, there will always phase shift, big or small. Lets say in feedback type there is 10deg shift after the loop is closed. I'm certain, that if the same amp is not closed-loop by feedback, the shift is far more than just 10deg.
A differential pair allows this small delay because it is actually 2 transistors with their emitors tied together. But delay cannot happened in 1 transistor. It is impossible to have base signal that shifted 10deg, for example, than the signal in IT'S emittor.
Yes, 2 resistor voltage divider doesn't introduce phase shift.
I don't understand. Which part is introducing delay IF THE LOOP HAS BEEN CLOSED BY FEEDBACK? If the loop is open, there will be delay, but if the loop is closed, none of this will happened (or happened with only small delay). That's what feedback do (with its plus and minus). Natural delay only happens in open-loop power amp.
Really ? I dont think so. I belive (or know ?
then should we feedback the signal. The feedback should just decrease the vertical distortion, but not correct the inner stage's phase and time problems.
I agree. Before closing the loop, the delay (and other non-linearities) should be as small as possible. Fix things first before applying feedback. Use small pf transistors. Use linear topologies.
That's why I get luck in making good sounding "few stages" feedback power amp, but not succeeded in making "many stages" feedback power amp. Better non-global-feedback type for the latter one.
Hi,
Yor statement was "feedback is definitely evil", I said it's an "evil wheel", or ring if that word fits better, maybe we are meaning the same thing(?), however I think and what I tried to say is that applying feedback is a trade-off, we get rid (partially) of one thing for something else.
In reality it could be that we are trading mostly lower harmonics for a spray pattern of higher harmonics.
Overall levels are lower with feedback, but human ears are also much sensitiver to higher order harmonics, thats the trade-off, hopefully no one missunderstood my words as it was not an advocating for or against feedback.
I dont know about establishing something better without feedback, but Charles Hansen with his commersial high-end Ayre amplifiers, is one example of non-feedback circuit designs (in that sense theres no more feedback than the inherent feedback mechanism in an amplifying device with a degeneration resistor).
Hopefully most of us understand this, the semantics have been discussed heavily in past year here on DIYaudio, which can be found with some
efforts! 
Cheers Michael
tlf9999 said:
feedback is definitely evil.
You just need to figure if no feedback is a bigger evil or not,
tlf9999 said:if you were to step back a little and recognize that nothing in this world is entirely evil or entirely perfect, you would have very quickly realized that criticizing feedback blindly, as some of us do, doesn't make sense.
Because no one has established so far that having no feedback is a better alternative.
Especially when one is dealing with mass production of open loop electronics,
Yor statement was "feedback is definitely evil", I said it's an "evil wheel", or ring if that word fits better, maybe we are meaning the same thing(?), however I think and what I tried to say is that applying feedback is a trade-off, we get rid (partially) of one thing for something else.
In reality it could be that we are trading mostly lower harmonics for a spray pattern of higher harmonics.
Overall levels are lower with feedback, but human ears are also much sensitiver to higher order harmonics, thats the trade-off, hopefully no one missunderstood my words as it was not an advocating for or against feedback.
I dont know about establishing something better without feedback, but Charles Hansen with his commersial high-end Ayre amplifiers, is one example of non-feedback circuit designs (in that sense theres no more feedback than the inherent feedback mechanism in an amplifying device with a degeneration resistor).
Hopefully most of us understand this, the semantics have been discussed heavily in past year here on DIYaudio, which can be found with some
efforts! Cheers Michael
lumanauw said:
Lumanauw,
you are now mixing up some things a bit, but I guees you know that 30 deg or 30,000,000 deg is not a delay, but rather phase.
Time is delay, like "5 minutes".
Cortez,
you did put lot of energy in your Post #85, I feel releaved to not having to write all that myself now!
Cheers Michael
Cortez said:
Right.
Just cause we apply feedback, the delay (phase shift) of the stages will not be reduced or 0.
Let me try once more.
There is VERY small delay in any circuit, because of the finite (< lightspeed) propagation of the signal. We can ignore that as it is in the order of a pS or so.
The feedback signal is NOT delayed. There may be phase shift inside the loop. What does that mean? It means that the reaction like output of fb signal to a Vout of the circuit is INSTANTENEOUS (apart from the 1 pS). What the phase shift means is that the wave shape is changed.
Example: Put a signal into an RC filter. That will give phase shift, meaning that, for instance, the amplitude peak of the input signal will be at another point in time in the output wave. But, and this is crucial, the output of the filter will start to change IMMEDIATELY if we connect an input signal. There is NO delay of the output signal from the input. Whenevcer thers is an input, there is an output. Same with a step voltage connected to a coil. The coil current will start to rise IMMEDIATELY, although, because of the phase shift, the current does not have the same shape as the volage. But, again, there is no signal delay.
What does that mean for fb? With rising freq, the fb signal has increasing phase shift to the output signal it is supposed to correct and to the input signal where it is subtracted. That means that the expected cancellation of the input signal from the (attenuated ) output signal, which only leaves the error signal to be amplified in inverse phase, that this cancellation is no perfect. Therefore, the correction signal is not accurate enough to cancel the error: the error is not completely reduced to zero. But there is NO delay of the fb signal as such (apart, again, from the 1pS or so mentioned before, and that is immaterial in audio).
Jan Didden
Agreed, I also think the same thing.
Feedback itself is a good concept, but the transistor is not "perfect" enough to do the job.
I think the "trading" of the lower order harmonics to higher order harmonics happens because in those higher order harmonics, the transistor is not having enough speed to do the job. If a transistor has more speed, it will eliminate the higher order harmonics, but will produce (actually "not-able-to-process") the higher-er order ones.
"Not 0", I agree, but "not-reduced", I don't agree. The differential pair (feedback center) will FORCE the whole amp (no matter 5 or 10 stages) to have NO-PHASE-SHIFT. But there will be some phase shift, because the differential cannot do the job 100%, but that amount must be far lower than IF the loop is open (not closed by differential feedback).
This sort of confusion is why I get grumpy when people refer to "polarity" as "phase."
Yes. Unfortunately, loudspeakers have a rolloff (and hence phase shift) at frequencies an order of magnitude higher than capacitively coupled power amps. And a higher order rolloff, too. So it's not surprising that DC coupling made no audible difference for unclipped signal.
Yes. Unfortunately, loudspeakers have a rolloff (and hence phase shift) at frequencies an order of magnitude higher than capacitively coupled power amps. And a higher order rolloff, too. So it's not surprising that DC coupling made no audible difference for unclipped signal.
lumanauw said:
The transistor has a non-linear transfer characteristic. Initially, amplifying a signal, this will cause harmonics and intermodulation (depending on the input signal). Because the feedback returns those harmonics to the input, they are again processed by the same non-linear characteristic that caused them in the first place. So you get harmonics-of-harmonics and intermodulation products of intermodulation products. That causes the increase in higher harmonic (and dis-harmonic) products. The transistor remains the same, it is still the same process that does it. There is no trading and nothing magic because of feedback happens to the transistor.
Don't forget that whatever you do to the loop, the amp itself will ALWAYS work open loop. If you have an amp with 1000 gain OL, and you apply feedback to get the gain down to 10, the amp still needs the input signal for its 1000 x gain. Assume you have 10V Vout, the effective amp input voltage is 10mV. This does not change whether you use fb or not. With fb and CL gain of 10 x, the effective Vin is comprised of 1 V input signal (to get 10 V out with 10 x CL gain), and 0.99 V fb signal. The effective 10mV input signal will look very distorted (because it is Vout with Vi subtracted). That is kind of "pre-distorted" so the Vout will look almost undistorted.
If you want to improve it, use more linear transistors (or topologies) and/or use faster transistors so the phase shift is less so the error cancellation by the fb is more accurate (see my earlier post).
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.