Suggestions for a very low distortion class A headphone buffer - 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 23rd April 2012, 04:29 AM   #11
diyAudio Member
 
ionomolo's Avatar
 
Join Date: May 2007
Location: Los Angeles
Send a message via MSN to ionomolo
Quote:
Originally Posted by volume knob View Post
Hi

If you're not apposed to I.C's then the buff634 in high bandwidth mode could be a simple solution
Thank you for your suggestion. I did a considerable amount of research before deciding to build my circuit, and I found there was not an actual reference buffer that had a truly great following. In fact most of what's out there falls in three categories:
  • Simple followers (For some reason people love to build them using FETs, even when they usually have large distortion and some not-so-benign high-order harmonics)
  • Random stuff in the feedback loop of an op amp
  • Diamond buffers (either integrated like the one you mentioned or discrete)

The first ones just don't do well compared to more elaborate topologies. The stuff in the feedback loop of an opamp does well but I think its quite inelegant, besides the fact that the opamp is a class-b circuit and in a headphone amplifier it will literally not leave the crossover region. Crossover distortion is not as nice as the second harmonic from single-ended class-A. Stability may also be a pain if it operates as a follower. The BUF634 that you pointed out has the problems of class-B, and I think distortion-wise it does well but its not great.

So basically I've decided to try my shot at a reference buffer for headphone applications. Certainly if my present design behaves as well in reality as it does in simulations I may be close to have just done it.

Quote:
Originally Posted by bear View Post
Look at "the wire" thread here... it uses a bunch of ultra-super low distortion buffer chips in parallel as a power amp output stage, iirc. You could use just one.

Also, I would expect ringing into a capacitive load with an inductor at the output.
You don't need that if the circuit will not destroy itself into an ultrasonic virtual short circuit (very low Z) - which is merely a capacitive load...

Consider paralleled devices if you need speed and can not find speed in a single device?

_-_-bear
If i'm well informed, The Wire is based on the LME49830. This would be 100x overkill for my headphone amp!

Regarding the circuit destroying itself when connected to a capacitive load. I measured my headphone cable capacitance to be around 1 nF. This was too much for my previous design, but the present iteration seems to hold it well and therefore I will remove the RL network from the design.

When I discussed ringing, I meant ringing at the output of the amplifier, before the load. This ringing is due to the phase margin not being high enough, and it is bad stuff because in my experience phase margins are always lower in reality than they are in spice.

Now I have replaced the input transistor by a Sziklai pair taken from Douglas Self book and the performace seems much better. I get a THD 0.0002% of (almost) purely second harmonic at 20 KHz 1Vpk an and a nice, critically damped square wave response on 1 nF without filtering. Please see the attached schematic (Disclaimer, it is untested and may not work, or even blow your headphones). If I can beat the 0.0001% THD mark I'll design a PCB. It would be awesome to see people changing the parts I've chosen because they sound bad . (Please note that I have not invented this circuit. It's close to an output-stage-less blameless amp with two pole compensation.

EDIT: But it has my contribution! While Douglas Self says cascoding the input stage does not make a difference, I've found that in this circuit does indeed provide an improvement by a factor of 10 in the distortion. I guess it is because of the non-inverting unity gain operation.
Attached Images
File Type: png sch.png (36.4 KB, 330 views)

Last edited by ionomolo; 23rd April 2012 at 04:34 AM.
  Reply With Quote
Old 23rd April 2012, 05:58 AM   #12
jcx is online now jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
Default your assumptions about monlithic op amps are wrong/outdated

op amps driving output buffers usually work Class A in all stages - the output of most "audio recommended" op amps are internally biased ~100 uA or more (based on indirect evidence from open loop output Z plots), driving 100+ kOhms, single digit pF at the buffer input they never leave Class A for audio signals

if you want deeper Class A output bias people have long loaded op amp output with few mA ccs to one rail to force the output into SE Class A - depending on op amp, fab tech the rail polarity, current magnitude may vary for "best" distortion - disappointingly for most the only measured improvements with this technique I've seen published are < 10 dB distortion improvement

while "discrete is superior" is well established meme in audio - I would claim for most consumer line level audio applications it has fallen to the technological advance of monolithic process development and design expertise at major linear op amp manufacturers competing for precision instrumentation, medical imaging, telcom markets over the past 20 yrs - it is no longer 1970s

you may say the “the wire” is overkill but it or related op amp + monlolithic buffer circuits take lots less design effort for results you are unlikely to approach within orders of magnitude with available discretes, less than decades of design experience

look at TI’s OPA164x series as a modern example of current performance for “audio” – if you have high source Z then consider OPA827 as a cheaper update to the OPA627 if low order ppm distortion from the input jfet parasitic C modulation worries you
  Reply With Quote
Old 23rd April 2012, 09:27 AM   #13
diyAudio Member
 
tiefbassuebertr's Avatar
 
Join Date: Jul 2005
Location: D-55629 Schwarzerden
From the audible results (not from the measurement results) this is my favourite design:
the Power Folllower
You must choice the suited idle current in order to the individual load impedance from your headphone.
  Reply With Quote
Old 23rd April 2012, 09:40 AM   #14
diyAudio Member
 
ionomolo's Avatar
 
Join Date: May 2007
Location: Los Angeles
Send a message via MSN to ionomolo
Quote:
Originally Posted by jcx View Post
while "discrete is superior" is well established meme in audio - I would claim for most consumer line level audio applications it has fallen to the technological advance of monolithic process development and design expertise at major linear op amp manufacturers competing for precision instrumentation, medical imaging, telcom markets over the past 20 yrs - it is no longer 1970s
I agree with you in the fact that modern op amps behave better than what is achievable with discrete parts, but they still require the buffer stage to drive a pair of headphones. What I'm trying to do here is to improve in the CFP buffer stage to get it to perform comparably to the globally closed-loop system with the opamp. It may be just a matter of elegance to improve a stage that is already on the design so it makes another stage unnecessary, and the overall performance may be marginally minor (You will probably agree with me that a class-A buffer with 0.0002% THD will not sound much worse than an opamp with 0.00004% THD), but I'm definitely interested in finishing this design.

Quote:
Originally Posted by jcx View Post
you may say the “the wire” is overkill but it or related op amp + monlolithic buffer circuits take lots less design effort for results you are unlikely to approach within orders of magnitude with available discretes, less than decades of design experience
I must disagree with you on the orders of magnitude thing. If there is the blameless amplifier that can repeatedly drive 8 Ohms loads to 50W with < 0.0008% THD, It must be possible to drive headphones to a few milliwatts with 0.0002% THD. This is not within orders of magnitude of audio opamps. Please note that I respect very much the designers of such IC's. They do 5 or 6 times better than my design goal with microamperes of bias current, while I'm running all the stages so hot that they only use a small portion of their transconductance curve. But I think an output stage can be made clean enough so as to simply not get any audible benefit from the opamp.

Quote:
Originally Posted by tiefbassuebertr View Post
From the audible results (not from the measurement results) this is my favourite design:
the Power Folllower
You must choice the suited idle current in order to the individual load impedance from your headphone.
This is a high-distortion design. While I agree that some distortion may add a nice character to the sound, what I'm trying to do in this project is to add such character by using a transformer, so I can choose the amount of distortion that I want by changing the transformer dc bias current, so it's important that the output stage is of low distortion.

Last edited by ionomolo; 23rd April 2012 at 09:59 AM.
  Reply With Quote
Old 23rd April 2012, 10:00 AM   #15
diyAudio Member
 
KatieandDad's Avatar
 
Join Date: Nov 2011
Location: UK
I'm just hunting around the Internet for a design that I found a few months ago, it was based in the CMOY amp but used two BURR-BROWN op-amps in parallel.

In the meantime this looks interesting KS Projects - High End Headphone Amplifier
  Reply With Quote
Old 23rd April 2012, 10:25 AM   #16
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
you are looking at the wrong 'the wire' design. the wire is a series of well characterised (with an audio precision system) amps, the first of which were buffered BAL-SE instrumentation amps for headphones and followed through SE-SE, BAL-BAL and a reworked BAL-SE see wiki for schematics

then there is the hybrid lme49830 lateral mosfet power amp and the one the bear is talking about that features paralleled buffers is called the LPUHP (low power ultra high performance) amp which is a ~16wpc speaker amp that can also be used for VERY current hungry headphones and features performance that matches the headphone amps but drives speakers, was designed for multiamped digital/active crossovers. the numbers on it are about half way down the page, but even with the AP system, he was unable to get any meaningful results, as the input preamp and ADC of the AP is outclassed by the amp. at 10W/1khz we are below your 0.0002% THD

Last edited by qusp; 23rd April 2012 at 10:29 AM.
  Reply With Quote
Old 23rd April 2012, 10:54 AM   #17
diyAudio Member
 
ionomolo's Avatar
 
Join Date: May 2007
Location: Los Angeles
Send a message via MSN to ionomolo
Quote:
Originally Posted by qusp View Post
at 10W/1khz we are below your 0.0002% THD
I'm sure it sounds dramatically better, as long as listening to 105 dB music sensitizes you to -10 dB acoustic signals.

Seriously, I'm just trying to build a linear buffer stage. It is an inoffensive, 5 bucks project (I have the heatsinks and transformers) and it's likely to sound quite good from the simulations, besides of being extremely instructive. I started this thread becasue I hoped to get some feedback from the circuit, but everybody seems convinced that I should be building something else.

Besides this, I once built a class-B mosfet output stage to run inside an LM4562 feedback loop, and I must confess I did not hear much of a difference from closing the loop, and it simulated 0.1% crossover distortion!

Last edited by ionomolo; 23rd April 2012 at 11:02 AM.
  Reply With Quote
Old 23rd April 2012, 11:19 AM   #18
Elvee is online now Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Quote:
Originally Posted by ionomolo View Post
Thank you for your suggestion. I did a considerable amount of research before deciding to build my circuit, and I found there was not an actual reference buffer that had a truly great following. In fact most of what's out there falls in three categories:
  • Simple followers (For some reason people love to build them using FETs, even when they usually have large distortion and some not-so-benign high-order harmonics)
  • Random stuff in the feedback loop of an op amp
  • Diamond buffers (either integrated like the one you mentioned or discrete)
.
Here is the oddball that falls in none of these categories:
The Tringlinator: a MOS-based Tringlotron amplifier

The TringloMOS output stage is unity-gain, operates in class A, has a very low distortion, and is ideally suited to transformers.

And it won't set you off more than 5 bucks
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote
Old 23rd April 2012, 11:31 AM   #19
diyAudio Member
 
ionomolo's Avatar
 
Join Date: May 2007
Location: Los Angeles
Send a message via MSN to ionomolo
Quote:
Originally Posted by Elvee View Post
Here is the oddball that falls in none of these categories:
The Tringlinator: a MOS-based Tringlotron amplifier

The TringloMOS output stage is unity-gain, operates in class A, has a very low distortion, and is ideally suited to transformers.

And it won't set you off more than 5 bucks
It's really nice to see something new. I'll certainly find some time to build one of these .
  Reply With Quote
Old 23rd April 2012, 11:20 PM   #20
expert in tautology
diyAudio Member
 
bear's Avatar
 
Join Date: Apr 2002
Location: New York State USA
You know, the real world implementation of your 0.0001% design is very unlikely to do that right out of the "box" - I suspect that you would have to trim just about everything and match devices to get it there - not to mention the power supply too... I suspect you know that already.

_-_-bear
__________________
_-_-bear
http://www.bearlabs.com -- Btw, I don't actually know anything, FYI -- [...2SJ74 Toshiba bogus asian parts - beware! ]
  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
Designing a super low distortion headphone amp. Rainwulf Chip Amps 41 1st April 2011 01:24 AM
Low-distortion Class-A line/headphone buffer linuxguru Analog Line Level 8 13th November 2010 08:00 AM
Low-distortion Super-pair Buffer simulation schematics linuxguru Solid State 29 12th November 2010 10:34 AM
Simple Low Distortion JFET Buffer wrenchone Solid State 21 29th January 2007 09:16 PM
Best low-noise low-distortion buffer? borges Solid State 7 9th December 2003 01:40 PM


New To Site? Need Help?

All times are GMT. The time now is 04:12 PM.


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