Some questions about amplifier classes, distortion and efficiency - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 14th January 2013, 02:34 PM   #11
diyAudio Member
 
Join Date: Oct 2012
Quote:
Feedback fixes everything......this is non-sense. Much of the crossover distortion components are well above the bandwidth of the global loop. This means the feedback does not get rid of the distortion but merely shifts it to higher order frequencies. Distortion components that are well above the 20KHz audio band limit can affect the sound for sure. You want the feedback to have to 'correct' for as little as possible. IOW, the goal should be that the global loop sees as linear of a transfer through the stages of the amplifier as possible.
Could you explain a little further? Where would these higher frequencies come from? About the feedback having to 'correct' for as little as possible, why would this matter? If an op-amp input stage is taken as an example, if then these unlinearities in the power stage are within the limits of the op-amp, shouldn't then the op-amp do it's job and output whatever it needs to output? As long as the op-amp isn't asked to output a voltage it can't output, why would there be distortion? Other than the tiny distortion you would get from an op-amp regardless of the output stage. I can see how a pure class B design will create crossover distortion though as in this design the op-amp is asked to go from 0V to ~0.7V in 0s which is clearly impossible. But taking the class AB design, this threshold voltage doesn't have to be overcome and the same moment the output of the op-amp changes, the output of the power stage will also change.
  Reply With Quote
Old 14th January 2013, 02:42 PM   #12
diyAudio Member
 
Join Date: Jul 2007
Quote:
Originally Posted by DF96 View Post
Feedback reduces distortion. It does not eliminate it. Whether distortion is audible is a matter of argument, hearing ability and personal taste.
Here is what I think feedback is.
Negative feedback amplifier - Wikipedia, the free encyclopedia
I dont know how this works. It reduces gain my the feedback factor and reduces distortion also by the same factor. How useless, why not just turn volume down.

Here is my version of feedback, which i guess doesnt reduce gain and removes all distortion and not just reduce it.
A is input signal
amplifier multiplies by 10
D is distortion

A ---Amp---> 10A+D
A -1/10 D ---Amp---> 10A - D +D = 10A
^ ^
| |
input - (input - 1/10 output)

- (input - 1/10 output) = feedback

Whay say ?

Regarding class A, B and AB
Here is response of transistor
http://www.vlsiinterviewquestions.or...rent-response/
Class A runs in small middle linear region, so obviously more liner and waste of power
Class B starts from bottom so no output until it input reaches 0.7v or something threshold, Class AB get rid of this 0.7 threshold problem by running at aprox 0.7 transistor even in idle.
Also notice Class B, Class AB uses more region of the response curve and as you can see the curve is Not liner.
extra info Class D operates out of all this liner region and runs in saturation region.

Last edited by Amit_112dB; 14th January 2013 at 02:51 PM.
  Reply With Quote
Old 14th January 2013, 02:52 PM   #13
DF96 is offline DF96  England
diyAudio Member
 
Join Date: May 2007
Your algebra is faulty.
  Reply With Quote
Old 14th January 2013, 03:12 PM   #14
CBS240 is offline CBS240  United States
diyAudio Member
 
CBS240's Avatar
 
Join Date: Aug 2005
Location: K-town
Quote:
Originally Posted by Plecto View Post
Could you explain a little further? Where would these higher frequencies come from? About the feedback having to 'correct' for as little as possible, why would this matter? If an op-amp input stage is taken as an example, if then these unlinearities in the power stage are within the limits of the op-amp, shouldn't then the op-amp do it's job and output whatever it needs to output? As long as the op-amp isn't asked to output a voltage it can't output, why would there be distortion? Other than the tiny distortion you would get from an op-amp regardless of the output stage. I can see how a pure class B design will create crossover distortion though as in this design the op-amp is asked to go from 0V to ~0.7V in 0s which is clearly impossible. But taking the class AB design, this threshold voltage doesn't have to be overcome and the same moment the output of the op-amp changes, the output of the power stage will also change.
Propigation delay is one factor. Every stage adds just a little bit of propigation delay to the system. Capacitances within the circuit and components do not charge instantly. The negative feedback loop is always trying to play catch up. It's not about asking the op-amp to provide an output voltage it isn't capable, in your example above, it is asking the op-amp or IPS+VAS to output a slew rate beyond it's limit, thus leaving the distortion from the output stage in the output signal. Although class AB is bias above cut-off, as I explained above it is the non-linear change in output Z that is the issue. Class A does not have this problem because Zout is not shared between two devices that are switching on and off. Here is an example of crossover distortion being seen dispite being class AB bias. The experiment is with mosfets dealing with error correction, a more sophisticated local nested feedback/feedforward circuit showing where they suffer a drop in Gm at the crossover region. The top waveform is the output. The bottom waveform is the error signal, or the inverse of the distortion products created, and clearly shows the required slew to drive the devices to linear output at different bias levels. This gives validity to Mr. Cherry's admiration of using nested feedback loops.
__________________
All the trouble I've ever been in started out as fun......

Last edited by CBS240; 14th January 2013 at 03:17 PM.
  Reply With Quote
Old 14th January 2013, 03:19 PM   #15
diyAudio Member
 
Join Date: Jul 2007
Quote:
Originally Posted by DF96 View Post
Your algebra is faulty.

How
  Reply With Quote
Old 14th January 2013, 03:44 PM   #16
DF96 is offline DF96  England
diyAudio Member
 
Join Date: May 2007
Quote:
Originally Posted by CBS240
Propigation delay is one factor. Every stage adds just a little bit of propigation delay to the system. Capacitances within the circuit and components do not charge instantly.
You are conflating two different things: propagation delay and filtering caused by capacitance. In almost any audio amplifier propagation delay can be completely ignored when considering distortion or loop stability. Filtering caused by capacitance is an important issue which complicates things.

However, filtering does not cause re-entrant distortion. This would still occur with an infinite bandwidth but distorting amplifier. In fact, it doesn't even need feedback - simple algebra is sufficient arising from a linear component in series with a non-linear component.

Quote:
Originally Posted by Amit_112dB
A ---Amp---> 10A+D
A -1/10 D ---Amp---> 10A - D +D = 10A
^ ^
| |
input - (input - 1/10 output)

- (input - 1/10 output) = feedback

Whay say ?
You can't subtract off distortion by subtracting off a (supposed) distortionless output. If you have eliminated D in the output (as you claim) then you can't use this perfect output to cancel D in the input, as you have no D left to use. Hence feedback can only reduce distortion, not eliminate it.
  Reply With Quote
Old 14th January 2013, 04:32 PM   #17
GoatGuy is offline GoatGuy  United States
diyAudio Member
 
Join Date: Dec 2012
Location: SF Bay Area
Quote:
Originally Posted by DF96 View Post
...cannot subtract off distortion by subtracting off a (supposed) distortionless output.
Actually, I think you missed the point. The author wasn't positing that his algebraic notion was a solution to the problem, but is just an algebraic form that negates it.

It is an interesting notion: that (in theory) one could cobble together an anti-distortion preemphasis network that would compensate without feedback per se for the amplifier's non-linear transfer function. Of course, in the (now) ideal world of having brilliantly simple amplification stages (triodes), the complexity of even a modestly accurate compensation network would dwarf the complexity of the rest of the amplifier.

Agreeing... with you though: in the real world, building conventional amplifiers out of the handful of accepted circuit topologies that we have come to generally agree are among the better ideas, then both local and global compensation feedback serve to remove distortion from the output, and yield amplifiers delivering close to the ideal Output = Amplification * Signal + zero distortion.

And there always must be some distortion on the output in order for the error-amplifiers along the feedback path to have something to react against.

It is for this reason that I'm particularly fond of non-capacitor-bypassed cathode bias resistors in single-ended amplification stages: gain is substantially reduced, even more remarkably linearlized, and the floating cathode (or source, if using FETs, or BJTs) becomes effectively a negative feedback element, that is ultra-local. Not global. Realistically, things such as 12BZ7s are pretty cheap, as are damn-fine newly made JFETs ... so having a few extra amplification stages (at lower per-stage gain) to effect the desired overall gain, is hardly worth gnashing one's teeth over. Especially when all the gain stages are being run in their sweet-spot of gain, amplification and current flow.

GoatGuy
  Reply With Quote
Old 14th January 2013, 04:57 PM   #18
Bigun is offline Bigun  Canada
diyAudio Member
 
Bigun's Avatar
 
Join Date: Jan 2009
Blog Entries: 2
I am coming to start to believe that the class of amplifier is not as important as the implementation. A very well implemented amplifier of class <insert your favourite letter> will meet your needs. But you won't hear that from commercial vendors (they will peddle their's as the best) and you won't hear it much around here because we're all mad about amplifier design and their minute details and enjoy arguing over them
__________________
"The test of the machine is the satisfaction it gives you. There isn't any other test. If the machine produces tranquility it's right. If it disturbs you it's wrong until either the machine or your mind is changed." Robert M Pirsig.
  Reply With Quote
Old 14th January 2013, 05:19 PM   #19
GoatGuy is offline GoatGuy  United States
diyAudio Member
 
Join Date: Dec 2012
Location: SF Bay Area
[Bigun] - you're right. To the point of near-hilarity, I read posts and treatises about the One and True Path [to amplifier nirvana], with staunch advocates of huge triodes, parallel triodes, pentodes, pentodes-as-triodes, cathode followers, gyrators, single ended, AB push-pull designs, H designs, cascode front ends, long-tailed pairs, differential signal paths, symmetric amplification, local feedback, global feedback, phase summing, yada, yada, yada.

In the end - the very most (absurdly) expensive pieces of cabinetry and steampunk valve art - which are described in exquisite (but utterly vapid) terms - inevitably all the designs "err" toward just being hugely overbuilt to become more linear, to offer unfettered output current sources, and to keep frequency response completely beyond question, whether impacting the sound, or not. And, more to the point, the huge coloration-of-tone that is the inevitable consequence of using moving masses to move air, to make sound (AKA "speakers") ... is sidelined in finely written prose. But... its all moving parts.

So long as an amplifier faithfully reproduces the VOLTAGE of the input at the output, in all its little twitches and wiggles, and as long as it can supply whatever CURRENT the speakers in turn demand as their combinations of magnets, windings, couplings, physical responses enact the old 3rd order reactive load equation ... then one can hardly ask more from the amplifier, regardless of how it is designed. The sad truth though is this: most amplifiers only do a "pretty good" job at reproducing the voltage wave, and even fewer do all that good of a job supplying the current resulting from the voltage across the reactive loads. And most that don't do a very good job are push-pull. And most that don't reproduce the overall voltage waveform well are either push-pull, or are using triodes (and transistors) in their raw, uncompensated mode(s). Many a grand design to "cure" or address a perceived ill (such as triode/FET Miller capacitance - a kind of HF negative feedback), will create a whole cartload of other evils in the nominal signal path.

And that's why I laugh.

GoatGuy
  Reply With Quote
Old 14th January 2013, 05:21 PM   #20
DF96 is offline DF96  England
diyAudio Member
 
Join Date: May 2007
Quote:
Originally Posted by GoatGuy
Actually, I think you missed the point. The author wasn't positing that his algebraic notion was a solution to the problem, but is just an algebraic form that negates it.
Quote:
Originally Posted by Amit_112dB
I dont know how this works. It reduces gain my the feedback factor and reduces distortion also by the same factor. How useless, why not just turn volume down.

Here is my version of feedback, which i guess doesnt reduce gain and removes all distortion and not just reduce it.
Perhaps I misunderstood, but that looks like a claim that his (faulty) algebra indicates removal of all distortion.

Quote:
Originally Posted by GoatGuy
It is an interesting notion: that (in theory) one could cobble together an anti-distortion preemphasis network that would compensate without feedback per se for the amplifier's non-linear transfer function. Of course, in the (now) ideal world of having brilliantly simple amplification stages (triodes), the complexity of even a modestly accurate compensation network would dwarf the complexity of the rest of the amplifier.
Predistortion is used in situations where NFB is difficult, such as linear RF PAs. It is only used in such situations, as elsewhere (such as audio) NFB works better.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Amplifier Classes AmpBuilder225 Solid State 23 1st February 2010 10:32 PM
Amplifier Classes Survey tom daghdha Everything Else 8 23rd April 2008 07:11 PM
How 'Green' is Class D compared to other amplifier classes? Lars Clausen Class D 38 9th October 2007 09:01 PM
[Theory] Amplifier efficiency Giaime Tubes / Valves 6 20th August 2006 07:23 PM
Amplifier Classes soundNERD Solid State 56 11th June 2004 09:16 PM


New To Site? Need Help?

All times are GMT. The time now is 06:06 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2