Class i - Page 6 - 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 13th December 2012, 05:32 PM   #51
wahab is offline wahab  Algeria
diyAudio Member
 
Join Date: Nov 2009
Location: algeria/france
Quote:
Originally Posted by Edmond Stuart View Post
Multiple output devices or not, keeping the total idle current equal, you will get the same gm wobble at the crossover region.
The (external) emitter resistors contribution to output Z would be less but if increased
in inverse proportion of the current decrease, for obvious bias stabilization , then it would change nothing.
  Reply With Quote
Old 13th December 2012, 06:08 PM   #52
Struth is offline Struth  Canada
diyAudio Member
 
Join Date: Aug 2012
Location: books at londonpower.com
Hi Guys

wahab, I can't follow what you mean.

Cordell follows Oliver exactly when adding more output pairs: each pair has the same idle current and Re values, so idle current goes up with each pair added.

Self is ambiguous about idle current versus number of pairs of outputs.

Leech explicitly stated that a given idle current corresponds to lowest THD for the whole output stage, so the pairs could share the current without altering THD.

Each approach has its own impacts; the ambiguous one likely the most...

Have fun
Kevin O'Connor
__________________
London Power / Power Press Publishing
www.londonpower.com
  Reply With Quote
Old 13th December 2012, 06:24 PM   #53
Elvee is offline Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Quote:
Originally Posted by Struth View Post

Elvee, to me it looks like your graphs and explanation target internal gm variations as at least one source if not the major source of crossover distortion.
Yes, precisely because all the (static) linearity/transfer problems in the Xover region can be traced to/interpreted as 1/gm variations and non-linearity.
A proof of that is that all problems vanish if the load is removed

Quote:
The reason this comes to mind is twofold: First, a common fix for large signal nonlinearity (LSN) is to parallel more output devices. This has multiple effects itself: reducing beta droop, reducing the current through each device, reducing output impedance, etc, all of which widen and flatten the gain wobble of the output stage. There is also a potential to increase the "internal" gm of the stage depending on how you treat the idle current.
As others remarked, if the paralleling simply means spreading the same current between multiple devices, little will change.
If identical stages are stacked, and the initial Iq is multiplied by n, the relative variation of gm will have proportionately less impact, unless the load is reduced in the same proportions

Quote:
The second reason is when you consider the class-A situation. Internal gm is high as the current is high, so internal resistance is low - the "slope" I believe you are referring to. Yet, this condition could be interpreted in one way as being waaaaay overbiased, but the side effects of such a condition are not present until you drive the circuit out of class-A into -AB, as Self demonstrated. Again, this points to switch-off issues. Under-, over and optimal-B bias all have switch-off characteristics but the narrowness of gain wobble varies, along with the THD profile.
That is a matter of taste as I said above, but fundamentally I do not see the logic of designing a 50W amplifier to be used only up to 2.5W.
Except for pure class A, there will always be a class switching somewhere, and the later the switching, the higher the THD.
Class AC of J Broskie is an interesting way round this problem.

It is not the universal solution to everything though: first it is a pretty hot class, and second there are problems of accuracy because the hand-over between devices has to be accurate for optimum results, and that is not necessarily easy to achieve.
It is certainly a clever solution deserving attention
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote
Old 13th December 2012, 07:06 PM   #54
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Post53 plots are very interesting and confirm what has been described many times before.
I note, for those that haven't spotted it yet, that the optimally biased has substantially 2nd harmonic, whereas the under-biased and over-biased have gross amounts of odd order distortions. Even the ninth of the under-biased is still more than the second.
In the over-biased operation the 7th and 9th have fallen to lower than 2nd. That seems to indicate a "sound" quality that may be better than under-biased. This is what an earlier poster stated, that if not optimally biased then over is better than under.

These high and odd order are Spikes that get added to the harmonious signal. It is the relative "size" of the spikes that makes them audible. They are narrow and thus contribute less to total percentage distortion than would a wide harmonic addition of similar "size". This gives an under representation of the "size" of the spike contribution to the signal resulting in the misleading impression that low numbers for crossover distortion cannot be heard.
0.1% of crossover distortion will sound quite different from 0.1% of second harmonic distortion where there are unmeasurable levels of the higher orders.

A blanket philosophy has developed that 0.01% of THD is inaudible.
I don't have the expertise, nor the instrumentation, to challenge that philosophy.
What I can do is recognise that 0.01% of low orders of THD are a very different animal from 0.01% of high and particularly odd orders of THD.

Simple total THD tells us virtually nothing, at least that's what I am led to believe. Aksa is a proponent of a gradual declining of all orders from 2nd down to wherever they become inaudible and/or unmeasurable, as an indicator of good sound.
Very few, if any commentators, have come to my notice to challenge this philosophy.

On the basis that some of the spikiness of crossover is what makes for bad sound then Optimal ClassAB bias must the the only way to progress if ClassA is energy wasteful and to be avoided by all that want to save the Earth.
__________________
regards Andrew T.

Last edited by AndrewT; 13th December 2012 at 07:11 PM.
  Reply With Quote
Old 13th December 2012, 08:04 PM   #55
Struth is offline Struth  Canada
diyAudio Member
 
Join Date: Aug 2012
Location: books at londonpower.com
Hi Guys

Elvee, have you ever measured how much power your own listening system and loudness preferences require? I have, and for me it is less than 1W per channel. So, for my own use, amps with a wide class-A region that covers the power I may ever use means that it is highly unlikely that I will ever push them to the point of class transition. Even if I did, the sound would be in excess of what I can humanly process so it is irrelevant. Maximum loudness here is about 90db, but more usual levels are 60db or so.

At that 1W/ch level, I could not listen for very long. More usual listening levels require one-hundredth the power and less.

Class-A eliminates major distortion mechanisms when implemented correctly. If you make an accurate assessment of your own power needs, then it is a highly viable and easy solution to achieve very-low THD. Distortion profiles are not much in question these days by anyone who knows how they impact sound. That an arbitrary THD figure is "inaudible" has always been the belief of a segment of the audio community, but it is again not what most knowledgeable designers believe.

Have fun
Kevin O'Connor
__________________
London Power / Power Press Publishing
www.londonpower.com

Last edited by Struth; 13th December 2012 at 08:09 PM.
  Reply With Quote
Old 13th December 2012, 08:19 PM   #56
wahab is offline wahab  Algeria
diyAudio Member
 
Join Date: Nov 2009
Location: algeria/france
Quote:
Originally Posted by Elvee View Post

As others remarked, if the paralleling simply means spreading the same current between multiple devices, little will change.
The combined Gm of the multiple pairs will be the same as the single pair
but the total value of the external emitter resistors (if their value is kept
the same as the single pair case) will be divided by the n number of output pairs.

Assuming 100mA total Iq and 0.22R emitter resistors , a single pair
output Z is about 0.24R , two pairs will yield 0.185R , 0.166R/3Pairs
and so on ,converging to an infinite number of pairs as yielding the
output Z of an emitters resistors less single pair ...
  Reply With Quote
Old 13th December 2012, 08:53 PM   #57
diyAudio Member
 
Edmond Stuart's Avatar
 
Join Date: Nov 2003
Location: Amsterdam
Default RE

Quote:
Originally Posted by wahab View Post
The (external) emitter resistors contribution to output Z would be less but if increased
in inverse proportion of the current decrease, for obvious bias stabilization , then it would change nothing.
Hi Wahab,

You're right and I like to add that in order to minimize the 'gm wobble' you have to increase the RE's inversely proportional to the number of OP devices (assumed that the total Iq is held constant, of course). So, as said before, nothing will change.

Cheers,
E.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en
goed verliezen dan dooft het licht…(H.M. van Randwijk)
  Reply With Quote
Old 13th December 2012, 09:03 PM   #58
Elvee is offline Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Quote:
Originally Posted by AndrewT View Post
Post53 plots are very interesting and confirm what has been described many times before.
I note, for those that haven't spotted it yet, that the optimally biased has substantially 2nd harmonic, whereas the under-biased and over-biased have gross amounts of odd order distortions. Even the ninth of the under-biased is still more than the second.
In the over-biased operation the 7th and 9th have fallen to lower than 2nd. That seems to indicate a "sound" quality that may be better than under-biased. This is what an earlier poster stated, that if not optimally biased then over is better than under.

These high and odd order are Spikes that get added to the harmonious signal. It is the relative "size" of the spikes that makes them audible. They are narrow and thus contribute less to total percentage distortion than would a wide harmonic addition of similar "size". This gives an under representation of the "size" of the spike contribution to the signal resulting in the misleading impression that low numbers for crossover distortion cannot be heard.
0.1% of crossover distortion will sound quite different from 0.1% of second harmonic distortion where there are unmeasurable levels of the higher orders.

A blanket philosophy has developed that 0.01% of THD is inaudible.
I don't have the expertise, nor the instrumentation, to challenge that philosophy.
What I can do is recognise that 0.01% of low orders of THD are a very different animal from 0.01% of high and particularly odd orders of THD.

Simple total THD tells us virtually nothing, at least that's what I am led to believe. Aksa is a proponent of a gradual declining of all orders from 2nd down to wherever they become inaudible and/or unmeasurable, as an indicator of good sound.
Very few, if any commentators, have come to my notice to challenge this philosophy.

On the basis that some of the spikiness of crossover is what makes for bad sound then Optimal ClassAB bias must the the only way to progress if ClassA is energy wasteful and to be avoided by all that want to save the Earth.
I broadly agree, but don't fall into the trap of "hearing with your eyes": you have access to technology allowing you to see and plot the amplitude vs time, and it is tempting to draw audio conclusions from the visual appearance of the waveforms.
In some cases both are in good agreement, but that is far from being a general rule: two shockingly different signals (visually) may sound exactly identical, whereas one shockingly distorted signal (audibly) may look almost identical to a perfectly clean one.
The fact is that an underbiased OP is one of the most offending sound you can hear: completely unnatural, unpleasant and detectable at tiny levels

Quote:
Originally Posted by Struth View Post
Hi Guys

Elvee, have you ever measured how much power your own listening system and loudness preferences require? I have, and for me it is less than 1W per channel. So, for my own use, amps with a wide class-A region that covers the power I may ever use means that it is highly unlikely that I will ever push them to the point of class transition. Even if I did, the sound would be in excess of what I can humanly process so it is irrelevant. Maximum loudness here is about 90db, but more usual levels are 60db or so.

At that 1W/ch level, I could not listen for very long. More usual listening levels require one-hundredth the power and less.

Class-A eliminates major distortion mechanisms when implemented correctly. If you make an accurate assessment of your own power needs, then it is a highly viable and easy solution to achieve very-low THD. Distortion profiles are not much in question these days by anyone who knows how they impact sound. That an arbitrary THD figure is "inaudible" has always been the belief of a segment of the audio community, but it is again not what most knowledgeable designers believe.

Have fun
Kevin O'Connor
Unfortunately, I have to listen at levels well under 1W, and probably <100mW: I run into problems with wife, neighbors, etc as soon as I try to push the volume.
I personally could content myself with a small class A amplifier, even a small class AB is vast overkill, but as I said above, that is not a question of my personal taste (or limitations).
If I design a 50W amplifier, I primarily want it to keep its specs for its whole ~0 to 50W range.
If you don't aim for perfection, there isn't half a chance you can reach half-perfection.


Quote:
Originally Posted by wahab View Post
The combined Gm of the multiple pairs will be the same as the single pair
but the total value of the external emitter resistors (if their value is kept
the same as the single pair case) will be divided by the n number of output pairs.

Assuming 100mA total Iq and 0.22R emitter resistors , a single pair
output Z is about 0.24R , two pairs will yield 0.185R , 0.166R/3Pairs
and so on ,converging to an infinite number of pairs as yielding the
output Z of an emitters resistors less single pair ...
Agreed, but with a low Iq, keeping emitter resistors that low makes them decorative: for them to be effective, they have to drop at least one or two tens of mV, preferably a little more.
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote
Old 13th December 2012, 09:12 PM   #59
diyAudio Member
 
Edmond Stuart's Avatar
 
Join Date: Nov 2003
Location: Amsterdam
Default one or two tens?

Elvee: just 26mV (in case of BJT's).

Cheers,
E.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en
goed verliezen dan dooft het licht…(H.M. van Randwijk)
  Reply With Quote
Old 13th December 2012, 10:16 PM   #60
diyAudio Member
 
Edmond Stuart's Avatar
 
Join Date: Nov 2003
Location: Amsterdam
Default correction

> increase the RE's inversely proportional to the number of OP devices
should read as follows:
> increase the RE's proportional to the number of OP devices
My apologies for the confusion.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en
goed verliezen dan dooft het licht…(H.M. van Randwijk)
  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
Class-AB meets Class-D: Yamaha's EEEngine Topology - where are this Diy Projects? tiefbassuebertr Solid State 29 14th April 2014 05:12 PM
Collection of Class B topologies <100mA Idle and Sound closest by Class A tiefbassuebertr Solid State 37 27th July 2012 09:04 AM
Can a Class AB PP amp be said to be operating in Class A at low signal levels? ray_moth Tubes / Valves 19 23rd January 2009 08:52 PM
How about a round-up of Class A kit power amps, or collectable vintage class A? Brisso57 Solid State 4 14th February 2007 11:30 AM


New To Site? Need Help?

All times are GMT. The time now is 07:32 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