I guess a better approach is --> pls send me recommendations for the very best opamp IC (or discrete) that would have the lowest noise and the lowest thd at 10 v p to p up to 100KHz into 10K. The very best out there.... known... not guesses. I can do my own guessing.
Thx-RNMarsh
Dick,
I did a project where I compared the LM4562, the AD797, a NOS Signetics NE5534 and a Burr Brown (forget which.) A few months later got in more samples of the LM4562, AD797 and the Burr Brown. I also used a newer JRC 5534 (I think that was it.)
Well the first time the LM4562 measurements were RFI contaminated, The AD797 measured not very good, the Burr Brown decent and the NE5534 decent. Repeating the test The LM4562 with input chokes, but also different date code was way better, The AD797 newer sample was quite good, The Burr Brown had a lot more 1/F noise and the 5534 copy had serious RFI issues.
So my approach since then has been to install a good gold socket and test each individual chip in my stock. (Also eliminates the issue of working impedances.)
So even though I have done measurements, I don't have any useful answers.
P.S. I think the first AD devices were actually returns to the authorized distributor where I got them as some were DOA.
I'll repeat that layout, earthing & decoupling are AT LEAST as important as the OPA. You need to do at least what's recommended in AD797 datasheet for ALL the OPAs you mention if you're after 1ppzillion THD.
The RFI is an indication that you need to work on these. Chokes, especially SMD ones or anything with ferrite will also introduce THD.
Another good tutorial is Cyril Bateman's series on capacitors where he describes a 1ppzillion oscillator for his tests. Note the OPAs he recommends.
Also, I can tell you from experience is that a DIP8 socket WILL degrade performance at these levels cos layout bla bla.
BTW, LM4562 and its more expensive cousins suffer from latching which is bad enough to have serious problems on power up in single supply applications.
The problem is placing these diodes so they don't introduce additional distortion while protecting against latching, overload etc.
Last edited:
If working in a pure digital environment one can generally get away with it over the long term. However, as someone who used a typical cheap PC in a beachside environment and had it rendered it unusable in less than a year by salt induced green sludge seeming to magically develop everywhere there were "gold" connectors I'm still not so keen on the idea ...
Audio is an area where you have to be fussy, really fussy, if you want the best sound - every potential weakness should always be eliminated.
Frank
Last edited:
The most extensive (400 pages!) set of measurements I'm aware of is Samuel Groner's "Operational Amplifier Distortion" report, linked from < http://www.sg-acoustics.ch/analogue_audio/ic_opamps/index.html >.
The fact that he presents measured results as a series of graphs rather than a few isolated measurements (or a single "measure of goodness") underscores the oft-repeated comment that you really need to consider circuit details outside the opamp itself before claiming the superiority of one over another. His report includes both integrated and discrete amps but doesn't have data for many of the most recent high-performance designs. If somebody has test gear that can replicate his measurements for a few of his examples, they can work at extending the data base to include other devices.
Dale
Thank you for the info. BTW -- I am using the commercially available oscillators to try the opamp upgrades in.... Like a K-H or HP and the like.... the grounding, shielding, bypassing et al is done for me.
Thx-RNMarsh
Good stuff in there. I went and read it all. I had read it in its first incarnation years ago. The expanded testing is really helpful. I'll order some more opamps to try out. So far the LT1468 has given me excellent results and it tests by SG very well also.
I have SW's recommendation on order for awhile now. Trying to sqeeze some more dB's out of this stuff used in testing.
Thx-RNMarsh
So am I... especially reports on completed ones from those who wanted us to use smd's on thier pcb.
meanwhile, what else can we talk about that's
? Is there a symbol for twiddling your thumbs? Thx-RNMarsh
Low feedback sounds better because --->
Well then, let me throw this out there --->
One of the reasons for going low gnfb -in my mind- wa to get a constant amount of fb at all frequencies. back in the day, I didnt have such good test equipment for measuring distortion.
I thought that the way sound was described sort of followed the thd curve. --- best at low freqs and progressively worse as the freqs when higher.... that high freq rise at the top end. Still a characteristic of many amps -- pre or power. As we know there are many mechanisms at work.
But limiting this to just the thd and amount of feedback here -- What if -- the harmonic structure did not change with freq ?? What ever the relationship between harmonics at low or mid was the same at the high freqs? Same ratios etc.
Wouldnt that have a sameness to the sound across the audible band? No more, the mids sound this way and the highs sound that way. The character would be the same/similar.
In fact, that is what you get in test performance and in listening results. If the harmonics are high enough to hear or the chain/system has harmonics high enough levels to hear differences. there is no other way to get this kind of favorable result (is there?)... assuming a falling open loop within audio feq band and varying feedback margins/amount of etal.
Thx-RNMarsh
Well then, let me throw this out there --->
One of the reasons for going low gnfb -in my mind- wa to get a constant amount of fb at all frequencies. back in the day, I didnt have such good test equipment for measuring distortion.
I thought that the way sound was described sort of followed the thd curve. --- best at low freqs and progressively worse as the freqs when higher.... that high freq rise at the top end. Still a characteristic of many amps -- pre or power. As we know there are many mechanisms at work.
But limiting this to just the thd and amount of feedback here -- What if -- the harmonic structure did not change with freq ?? What ever the relationship between harmonics at low or mid was the same at the high freqs? Same ratios etc.
Wouldnt that have a sameness to the sound across the audible band? No more, the mids sound this way and the highs sound that way. The character would be the same/similar.
In fact, that is what you get in test performance and in listening results. If the harmonics are high enough to hear or the chain/system has harmonics high enough levels to hear differences. there is no other way to get this kind of favorable result (is there?)... assuming a falling open loop within audio feq band and varying feedback margins/amount of etal.
Thx-RNMarsh
Last edited:
I was struck recently at gazing at the schematic of a piece of test gear; Princeton Applied Research Model 113 preamp. The stamp of the schematic says 1966. Yet, such sophistication did not catch up to audio until the early 80s. A discreet bench test operational amplifier of great sophistication, FET diff amp, ccs on everything, input with all the input protection against overload. It amazed me at how long audiophiles were forced to listen to designs that were of much less sophistication. Audio must be the step child of electronics. It also amazes me at how different topologies sound different. I'm not an EE, just a serious audiophile but it seems to me that design has gone in FADS. What will be the new hare that everyone will chase? Ray
The goal is, that there is no "sound" to the system. If the "theories" are correct then that should always be the situation at the moment -- because we can achieve below audible distortion quite easily. People with sensitive hearing know this is a nonsense in real systems ... which IME is all about poor total system engineering, not the competence of an individual component's design ...
Frank
As things progress... at least at the high-end... everything gets better. However, systems are for one reason or another - not the least of which is cost - often composed of various components of mixed blessings. Some, tube, some IC, some ss discrete, some bipolar, some jFEt, some MOSFET, combos of all and EMI/RFI and grounding and shielding and on and on... that keeps everyone guessing. Source material, acoustics, speakers, mechanical contraptions like the LP and all its variables of noise, and distortions. An infinite number of combinations to configure... not all of which is low enough in s/n, distortions of all kinds etc. Some are better than others. One goal here is to have everyone's complete/entire system be transparent and sound like real live music. Not there yet. Not at either end of the chain. But the potential is greater than ever.
Thx-RNMarsh
Of course, I would beg to differ ...
. Digital done right, tweaked sufficiently, ticks the first box, and at least IME speakers can be made to work "good enough" to deliver the live music experience ...Frank
as am I, I have many of the parts, need to find time for matching. waiting on an etching bubble/bath as well, I will make my own PCBs. have been distracted with other projects too, ones I really should get finished before taking on anything new.
i've also become distracted with playing with some new parts, EPC eGaNFETs EPC2012 to be exact.