Once I have a set of two files (one normal, one treated), it does not matter if I play the pair across WiFi network to Android phone as receiver, or transfer the files to the Android phone I still get the same differences.
I have emailed files pairs to myself and same result.
Several others reliably detect and report same differences.
There is no change in PB system providing confounders.
I am recently having much fun 'remastering' digital files, and find that I can add particular 'signatures' at will.
Next round of experiments will be to explore reducing/nulling inherent formulation sounds of very old shellac recordings....I think it's possible.
Interestingly, the formulations of a selection of very old records sound distinctly different, ranging from quite good/friendly through to distinctly nasty....that PB mids/vocals bark/bite/harshness is not entirely due to the original recording chain or PB path.
Like I said, something intensely curious is going on and might be an answer to some listeners subjective dislike of digital audio.
Dan.
Wow, well this has certainly exploded.
Anyone reminded people that, given the gobs of gain (feedback), general "friendliness" (good practices get most out of most opamps) and general linearity of so many modern opamps, that audible differences, if there, are going to be SMALL? And, thus you really need to torture your tests to have a prayer of identification. Anything outside of large data sets with careful control (and, well, barely above 50%) should be looked at with immense incredulity.
If you don't know this, I invite you to spend less time with DiyA and much more hanging out with Bayes.
Anyone reminded people that, given the gobs of gain (feedback), general "friendliness" (good practices get most out of most opamps) and general linearity of so many modern opamps, that audible differences, if there, are going to be SMALL? And, thus you really need to torture your tests to have a prayer of identification. Anything outside of large data sets with careful control (and, well, barely above 50%) should be looked at with immense incredulity.
If you don't know this, I invite you to spend less time with DiyA and much more hanging out with Bayes.
In my estimation, the differences between IC op amps is not really very small, but can be lost with the inherent limitations of all the other IC op amps that either precede or follow the IC under test. Intellectually, at first, it would appear that low distortion (AM) implies the absolute sound quality of the IC, but it doesn't appear to work that way. That is why Bear or I don't have any IC's following the path to the speaker. It could be that a series of similar IC's will probably make slightly more distortion, but not enough to make a big difference, especially when following IC's in the audio path are ignored.
For example, when Dave Wilson made his IC op amp comparisons, he did it with my discrete electronics around it, not other IC's. This allowed IC comparison possible. I was not there for the IC evaluations, but I was told the 741 actually came in the middle of the spread. Apparently TIM is only a partial contributor to sound quality. Other factors must be just as important. It is these other factors that we still have to find.
For example, when Dave Wilson made his IC op amp comparisons, he did it with my discrete electronics around it, not other IC's. This allowed IC comparison possible. I was not there for the IC evaluations, but I was told the 741 actually came in the middle of the spread. Apparently TIM is only a partial contributor to sound quality. Other factors must be just as important. It is these other factors that we still have to find.
+1 for Dan Ariely, I have more than one of his, excellent work.
Also just finished 'Think like a freak' from the 'Freakonomics' duo Levitt and Dubner. Highly recommended.
So, anything evaluated since, well, the introduction of the SA5532, much less a product like an AD797 or the AD275? (I might be wrong about my dates, but thought the two latter were around the same time)
Yeah, a 286sx processor is also incredibly slow and power-hungry by today's standards. But we don't lump modern microprocessors by it. 🙄
Yeah, a 286sx processor is also incredibly slow and power-hungry by today's standards. But we don't lump modern microprocessors by it. 🙄
Daniel you weren't with it unless you had an 8086 math co-processor to go with your cpu and then an upgrade to a 20mb hard drive from a 10mb drive. My first fax machine cost over $3,000. Those were the days.
A very easy process that everyone can try is this:
Download a free reference 24-bit WAV, such as this
https://archive.org/download/onclas...r_tombeau-de-m-blancheroche_small-version.wav
Then download this ABX test software (link at bottom of this page)
ABX software
You don't need to prepare a second wave file. It will take the first wave file and introduce cross over distortion, bit depth distortion or amplitude difference and let you ABX right there
The cross over distortion that gets inserted is very small, but it is very massive on an audio analyzer, but to my ears I cannot hear it.
Try it. I think it will be very helpful to learn what the ear can achieve. And then it will be easy to know if an op-amp can make a difference or not.
I am aweful even at 1 dB of volume difference:
QA_ABX v0.92
Loaded file: onclassical_demo_riva_froberger_tombeau-de-m-blancheroche_small-version.wav @ 9/25/2016 12:19:01 AM
File MD5: b465c1c991b35345002b55aca3abc562
Distortion: Amplitude Value: -1.00
Sample Rate: 44100 Bit Depth: 32 Channels: 2 Format: IeeeFloat
Pass @ 9/25/2016 12:19:21 AM
Fail @ 9/25/2016 12:19:55 AM
Fail @ 9/25/2016 12:20:18 AM
Fail @ 9/25/2016 12:20:39 AM
Finished @ 9/25/2016 12:20:42 AM
|570698a374614ecc74dfe2dc81fd3b1f|
Cross over distortion at 0.96 value I can easy detect. This is about -60 dB THD and you will hear some buzzing on string pluck very easy
QA_ABX v0.92
Loaded file: onclassical_demo_riva_froberger_tombeau-de-m-blancheroche_small-version.wav @ 9/25/2016 12:23:30 AM
File MD5: b465c1c991b35345002b55aca3abc562
Distortion: Crossover Level: 0.96
Sample Rate: 44100 Bit Depth: 32 Channels: 2 Format: IeeeFloat
Pass @ 9/25/2016 12:23:48 AM
Pass @ 9/25/2016 12:23:56 AM
Pass @ 9/25/2016 12:24:08 AM
Fail @ 9/25/2016 12:24:13 AM
Pass @ 9/25/2016 12:24:27 AM
Pass @ 9/25/2016 12:24:33 AM
Pass @ 9/25/2016 12:24:40 AM
Pass @ 9/25/2016 12:24:44 AM
Finished @ 9/25/2016 12:24:47 AM
|9f759bb9f86dc20b00ed5ff718295517|
But at 0.98 setting for crossover distortion my result it is not so good. Author web page says this is about -72 dB THD
QA_ABX v0.92
Loaded file: onclassical_demo_riva_froberger_tombeau-de-m-blancheroche_small-version.wav @ 9/25/2016 12:25:33 AM
File MD5: b465c1c991b35345002b55aca3abc562
Distortion: Crossover Level: 0.98
Sample Rate: 44100 Bit Depth: 32 Channels: 2 Format: IeeeFloat
Fail @ 9/25/2016 12:26:00 AM
Pass @ 9/25/2016 12:26:10 AM
Fail @ 9/25/2016 12:26:32 AM
Pass @ 9/25/2016 12:26:43 AM
Fail @ 9/25/2016 12:26:53 AM
Finished @ 9/25/2016 12:26:54 AM
|4cbbdc6f4e609648dfd275e3e6c3aceb|
If one opamp has distortion at -100 and another has distortion at -90, the chances seem I won't be able to hear a difference on my equipment. Other listeners will have other results.
Steven, my family's 486dx-100 in 1993 had the math coprocessor. AND an EISA mobo!
Ain't modern tech great?! 😀
Ain't modern tech great?! 😀
Samuel Groner did an extensive exploration of measurements on a number of opamps.
I don't have the link, but it's easy to find online.
My problem at present is that I am confounded by the lack of my ears (and others who have been here to listen) ability to follow the sonic perceptions and have them track the measured results. That means that the opamps that appear in measurements to "sound best" do not necessarily correspond to the "ranking" based solely upon measurements.
At least I have not yet found the combination of or specific measurements that are predictive of the sonic result. Been staring at Groner's results for a while now. About this, I am both confounded and interested.
That's why I suggested using the "10 in series" test to see if the differences can be "magnified". Perhaps something would show up.
The AD797 is in Groner's test.
_-_-
PS. it is possible that a particular sort of "distortion" actually sounds "right" or "natural" and that technically cleaner opamps do not sound right because they are revealing the "true" signal - BUT, this begs the question of HOW such a mechanism would work.
For example how can possibly some sort of higher distortion part cause sibilence to be reduced and things like cymbals to have a more "natural tone and body"??
Whereas the lower distortion part(s) decrease the "body"?? Distortion products go UP in frequency, not down. So, what is going on? (assuming my description is accurate)
Frankly, right now this is what I hope to be able to discover - what the mechanism is - given the very clear subjective results I am seeing.
We know it can not be bandwidth, because we have a bandwidth/rise time limited signal.
We know it can not be slew rate, because all parts exceed the nominal slew rate requirements... etc. etc.
I don't have the link, but it's easy to find online.
My problem at present is that I am confounded by the lack of my ears (and others who have been here to listen) ability to follow the sonic perceptions and have them track the measured results. That means that the opamps that appear in measurements to "sound best" do not necessarily correspond to the "ranking" based solely upon measurements.
At least I have not yet found the combination of or specific measurements that are predictive of the sonic result. Been staring at Groner's results for a while now. About this, I am both confounded and interested.
That's why I suggested using the "10 in series" test to see if the differences can be "magnified". Perhaps something would show up.
The AD797 is in Groner's test.
_-_-
PS. it is possible that a particular sort of "distortion" actually sounds "right" or "natural" and that technically cleaner opamps do not sound right because they are revealing the "true" signal - BUT, this begs the question of HOW such a mechanism would work.
For example how can possibly some sort of higher distortion part cause sibilence to be reduced and things like cymbals to have a more "natural tone and body"??
Whereas the lower distortion part(s) decrease the "body"?? Distortion products go UP in frequency, not down. So, what is going on? (assuming my description is accurate)
Frankly, right now this is what I hope to be able to discover - what the mechanism is - given the very clear subjective results I am seeing.
We know it can not be bandwidth, because we have a bandwidth/rise time limited signal.
We know it can not be slew rate, because all parts exceed the nominal slew rate requirements... etc. etc.
You must first decide if your goal is to hear it as closely as the engineer or musician heard it when they made it, OR if your goal is to hear it in a way that sounds more pleasing to your ear.
They are not always the same
Yes, I have read some of their books as well. Interesting and entertaining, for sure. There are some sort-of-similar books that would also classify as very good, but not quite at the top of my list for someone wanting to learn the most helpful stuff first, and who has limited time to invest reading.
Some books I would recommend after Kahneman and Haidt, would include: The Invisible Gorilla, Mistakes Were Made - But Not by Me, and Stumbling on Happiness. For behavioral economics, Misbehaving, is a good one. And, Influence, for marketing, advertising, and protecting yourself from same.
Right, that's my point, let's not even worry about opamps in the last 10 years, let's talk since ~1990! (and thank you for the correction wrt op275; mea culpa!)
And... when nothing is badly behaving??
Also, again, we're talking ONE opamp per channel - there is only one in this case.
_-_-
Also, again, we're talking ONE opamp per channel - there is only one in this case.
_-_-
Oscillations, clipping, overloaded current-wise = opamp badly behaving. And clearly measureable features.
Otherwise, you're showing all the tell-tale characteristics of finding exactly what you're looking for. (All up in your head)
On my phono preamp I can hear the difference between the LT1358 and the LT1364 when directly driving AKG K701 'phones.
I don't even look at sub 100v/us slew rate op amps.
The stuff in TAA in the 90's about class A biasing seems to have been forgotten.
The fast op amps need discrete voltage regs, current source shunts and polystyrene filter caps RIGHT NEXT to the power supply pins to let them shine.
It's no wonder you guys aren't hearing much difference with op-amps in the 20v/us range powered by 3 terminal regulators.
The noise and distortion figures in the data sheets don't tell the whole story.
I don't even look at sub 100v/us slew rate op amps.
The stuff in TAA in the 90's about class A biasing seems to have been forgotten.
The fast op amps need discrete voltage regs, current source shunts and polystyrene filter caps RIGHT NEXT to the power supply pins to let them shine.
It's no wonder you guys aren't hearing much difference with op-amps in the 20v/us range powered by 3 terminal regulators.
The noise and distortion figures in the data sheets don't tell the whole story.
Many records mixed and mastered in the early days of digital sounded bad. Thin, hard, grainy, smeared, etc. So, depending on which record you are talking about, the mix engineer may have been wishing he could have used tape instead. And, yes, those problems tend to be very noticeable on cymbals, at least if one focuses attention on cymbal fidelity, at the expense of perhaps ignoring the rest of the music for a moment. Somebody mixing, of course, has to do that, and very carefully so. Or not stay employed for very long.
Also, more fidelity and detail don't always sound better. If you can hear the vocalist's lips smack, teeth click together, hear 3D-like fully-articulated guitar squeaks, etc., all in super detail, those things don't necessarily do much if anything to enhance music listening pleasure. Rather, they can detract from it. Sometimes a good maximum bitrate mp3 or AAC made directly from a 24/96 mix can sound more aesthetically pleasing than the original. It will sound different of course, and less detailed, but in cases like that, it can be a realistic option.
Maybe a few things besides that. Jim Williams was quoted as saying about opamps, "always use them inverting, unless you can't." He was talking about CM non-linearity, something manufacturers were not always highly proud of, so they tended not to draw much attention to it. As opamps improved, there was then some bragging by people about improved CM performance, by the same people who were rather quiet about it before. Same type of thing with rail-to-rail opamps. First, it was a great new feature! Then, the next ones were new and improved! After that the next ones were even better! So, the point of all this is that you can't always rely on manufacturers to tell you everything you might want to know about the shortcomings of their products. If they come up with something better, they will bring to your attention at that time why you need it.
- Status
- Not open for further replies.