[mfueger]

I found this on a Swiss site, written 2009.
 
http://www.sg-acoustics.ch/analogue_audio/ic_opamps/pdf/opamp_distortion.pdf
 
What are your thoughts?

 

[Avro_Arrow]

A very interesting read...

 

[akgfan]

Interesting paper, but when you love mosfets or tubes, no distortion can stop you.

 

[nikongod]

 
Quote:

Interesting paper, but when you love mosfets or tubes, no distortion can stop you.

High order non-musically related distortions can. 
 
The irony of what you just posted is that many amps posting scope-jockey-approved THD/IMD use mosfets. 
 
Ooh, and as a last point: anyone who thinks that an opamp measures as well in a real life circuit as the datasheet indicates had a sad joke played on them. I guess the paper linked to just proves that.

 

[leeperry]

This paper is exactly what the average clueless roller will go through, pop it in and listen/measure...but any serious EE will call it clueless to the utmost, as each opamp requires a proper surrounding design in order to provide the best results. This is not the case here, he simply popped them in and measured them. Real men don't read datasheets.

 

[Avro_Arrow]

The part I found most interesting is the fact that the authors creations
measured the best...isn't that always the case?

 

[samsquanch]

Quote:

The part I found most interesting is the fact that the authors creations
measured the best...isn't that always the case?

Wait, are you trying to tell me that when pepsi says that they taste better than coke, that they are in some way bias?

 

[rds]

Wow that's a lot of measurement and obviously a lot of work.  But not peer reviewed (as are many of the references), with very questionable methodology ...and overall TLDR  

 

[Yoga Flame]

Quote:

The part I found most interesting is the fact that the authors creations
measured the best...

 
That is true. Most of the SGA discreet opamps measured very well in the paper. The exception is SGA-SOA-1 that had not a lot going for it. The author even wrote: "Perhaps interesting where low loads are to be driven. Otherwise behind most IC amplifiers distortion wise." On the first release of this paper in 2008, that was the only one of his opamps measured. Meaning that he pretty much trashed his own design at first.
 
But come February 2009, a whole bunch of other opamps were also tested, including the later discreet SGA opamps designed by the author. These were the ones that did well. So one interpretation is that the author learned from his mistakes and improved his designs to perform better.
 

 

[Yoga Flame]

 
I mostly just read the comments. These are the tested opamps that interested me:
 
The popular hi-fi discreet opamps (OPA-Earth, OPA-Moon, and the Buron Audio one) had relatively high distortion.
 
AD797 did well, but "stability is not easy to achieve". And the circuit had to be tweaked to accommodate it.
 
For LT1363, "there are better opamps out there at that price."
 
LME49860 (aka LM4562) was a "good all-round low distortion opamp which needs carefull attention to
common-mode effects."
 
OPA627 did very well. "A good though very costly choice for low distortion applications requiring
JFET inputs."
 
OPA827 was also "a good choice for low distortion applications" though not quite as nice as OPA627.
 

 

[akgfan]

Quote:

High order non-musically related distortions can. 

 
Unless you are a guitar player (I am).

 

 

[nikongod]

 
Quote:

 Unless you are a guitar player (I am).
 

Im sorry you haven't figured it out. This is a hi-fi site. 

 

[akgfan]

 
Quote:

Originally Posted by nikongod /img/forum/go_quote.gif
 
Im sorry you haven't figured it out. This is a hi-fi site. 

That was a joke. But this is also a DIY thread and not all amps we built have good parameters. Not to mention that not everyone can measure THD for example.
 
I remember guys in my school long time ago when they claimed that their ic amps had 0.007% THD and other nonsenses because they read it in datasheets. Sweet innocence.
 

 

[jcx]

Pease has done some interesting measurements too http://www.national.com/an/AN/AN-1485.pdf - but the best parts are discontiuned so  test methods/philosophy may be the most useful parts of the paper
 
of course I think the better approach for ehadphone use is to separately buffer (in the loop, take added gain too) the output with someting like the tpa6120