The Opamp thread

[Aynjell]

Quote:

...


Favourites

Oh that's just geek porn right there.
 

 

[dlaloum]

Yes - but in my opamp measurements with RMAA, I believe (very strongly) that I am seeing increased crosstalk due to the adapters, along with increased IMD...
 
So the design of the adapter is absolutely critical - path lengths need to be eve and so does the shielding of paths from interference etc.... (especially given the paths are now longer and more exposed)

 

[leeperry]

 

but in my opamp measurements with RMAA, I believe (very strongly) that I am seeing increased crosstalk due to the adapters, along with increased IMD...

using internal soundcards? surely the browndogs will pick up interferences there. As far as I know, you don't care for IMD in a playback device.

 

[Avro_Arrow]

That's good to know, but taking a SOIC chip, soldering it to an adapter and sticking it in a socket to boot
just destroys any sonic quality the chip might have had in the first place...

 
Quote:

 

In many ppl's experience, two single opamps on an adapter sound much better than a dual.
 
even TI acknowledged the problem, but didn't really fix it: http://focus.ti.com/docs/prod/folders/print/opa1612.html
 
"The dual version features completely independent circuitry for lowest crosstalk and freedom from interactions between channels"
 
We don't solder them on adapters for the hell of it, and besides I very much pretty to solder DIP8's than SOIC8's....much easier.

 

[leeperry]

 

taking a SOIC chip, soldering it to an adapter and sticking it in a socket to boot just destroys any sonic quality the chip might have had in the first place...
 

 
Talking from personal experience I'm sure...

 

[Mad Max]

 
Quote:

That's good to know, but taking a SOIC chip, soldering it to an adapter and sticking it in a socket to boot
just destroys any sonic quality the chip might have had in the first place...

So soldering it in should make an incredible day-and-night difference with it?
 

 

[Sebhelyesfarku]

Quote:

I guess I'll go with the slammin' LT1364.

I've put an LT1364 into my headphone amp in the place of an NE5532 but haven't got "night and day" differences, to put it mildly. Probably because it's only in the input stage, the amp is discrete class A. Amps fully based on opamps can realize bigger improvements, I guess. Or I'm simply deaf... 

 

[Avro_Arrow]

Well, if you mean it can make the difference between it sounding good and not working at all,
then I guess it can make a big difference. Just try putting an OPA690 in a socket...
Don't get me wrong, I too use adapters and sockets and so on...I just know I am presenting
the op amp with a less than optimal situation. The final circuit is always proven with the
op amps soldered directly in whether it is DIP or SOIC.
 
Quote:

 

So soldering it in should make an incredible day-and-night difference with it?
 

 

[Avro_Arrow]

When you are rolling op amps, there is never any guarantee that any particular op amp is always
going to perform well, you are presenting it with a circuit that was not designed for it.
LT1364/LT1363 are fast op amps. If you put it in a circuit where the other components are not
as fast then it will not make as much difference.
 
You have to consider a circuit as a whole to understand what will work and what will make a difference.
 
Quote:

I've put an LT1364 into my headphone amp in the place of an NE5532 but haven't got "night and day" differences, to put it mildly. Probably because it's only in the input stage, the amp is discrete class A. Amps fully based on opamps can realize bigger improvements, I guess. Or I'm simply deaf... 

 

[dlaloum]

Quote:

 

using internal soundcards? surely the browndogs will pick up interferences there. As far as I know, you don't care for IMD in a playback device.

No actually - running in either of a Matrix M-Stage (Diamond Buffer discrete class A buffer stage, class A opamp operation with 3.9k res @ +/- 15V) - or in a Firestone Audio Cute Curve (also biased into class A for the opamp using the same resistor but at +/- 12V )
 
All cables to from sources have chokes fitted to stop RF getting through and polluting things - tests were initially baselined in loopback to check what internal noises were Soundcard generated (M-Audio 2496).
 
The IMD is being generated by the Headphone amps circuitry and particularly by the opamps - with substantial differences between them - and particularly noticeable increases with adapters whether soldered or socketed ones. 
 
IMD is considered by many to be the nastiest (or one of the nastiest) and most audible form of distortion and the one that it is most critical to reduce - typical Valve amps that sound great often have terrible THD but good IMD specs. Also much pleasant colouration is due to harmonics - ie THD - and preponderance of even order harmonics (which is part of THD)
 
From Stereotimes article http://www.stereotimes.com/DAC120709.shtml reviewing the BelCanto DAC (they are talking here about the analogue circuits :

"Inter-modulation distortion is a standard test measurement of audio performance. REF1000 amplifiers, for example, are rated at 0.0007% IMD (CCIF) at one watt into 4 ohms, using 14:15KHz tones. IMD occurs due to circuit nonlinearity and sounds especially harsh and unpleasant because it is not harmonically related to the audio signal. "........."The same thing happens with those spurious high frequencies I'm calling noise, they inter-modulate one another and generate products that can fall within hearing range. These products are very low level, they are random, they are unrelated to the music, they are virtually impossible to measure with certainty and accuracy (partly due to the noise floor of the test equipment), but their effects are audible. How do I know? Because their reduction in the DAC3VB produces an audible change. "

 

[dlaloum]

Quote:

That's good to know, but taking a SOIC chip, soldering it to an adapter and sticking it in a socket to boot
just destroys any sonic quality the chip might have had in the first place...

 

 

Well that would be "gilding the lilly" a little bit...
 
The socket/adapter is undesirable and can potentially introduce various problems (RFI, Phase delay due to differing trace lengths, etc...) but a really good chip in an adapter/socket will usually still sound better than a bad one  soldered into place.... its just that it could sound even better if it were soldered into place rather than socketed and mounted on an adapter.
 
Next question is - How much better.... well that can be measured - lots of things can be quantified - THD, IMD, Freq resp, rise time, etc... can all be quoted......
 
Final question - What is that difference worth to you?    heheheh 

  Some people would sell their souls for another 0.001% reduction in IMD....

 

[sw98]

Random question, but is there any difference in soundstage or SQ between the LME49720 and LME49720NA?

 

[qusp]

Quote:

Final question - What is that difference worth to you?    heheheh 

  Some people would sell their souls for another 0.001% reduction in IMD....

I sure would if I thought it was still worth anything

 
totally with you here and adapters do not have to be such a bad thing if they are well made, for sure I would still take an OPA827 in soic8 over OPA134 in dip8, in a second.
 
sw98: IMO no, none at all, i'm sure someone will pipe in and say its a night and day difference and that I must have woolen ears and that immeasurable differences are plain as day with their golden ears.
 
yes I have both, yes I use the can if any, but that is more about my neurosis than anything else. that being said, although it measures well, its boring as hell IMO and I never do. its popular as a buffer or in dac IV stages because of its measured performance and the fact its unity gain stable, but I just find it sucks all the life out of the music...... insipid is a word that comes to mind
 
LME49710 is excellent in regulators though, you might try checking out some of the more modern national chips like the LME49722 and tell us how it goes, has even better spec

 

[sw98]

Thanks for your help! I'm purchasing a new Essence ST that comes with 2 LME479720NA's, so if there's no difference to the LME49720 then I'll just pop in the NA's.

 

[igor0203]

Try LT1028ACN8