Cable Myths Continued

[DrWho]

Btw, that was my 16,000th post....I think that may have been the first time I was willing to identify myself as an "audiophile" - haha, it took a while

[Don Richard]

Generally, we try to use visual language to describe audible conditions where we just don't have words. Distortion is completely vague.

Not really. There is linear distortion, meaning a difference in level, or a difference in frequency response between two examples of an audio event. Then there is nonlinear distortion, which most people mean when they say, "distortion". Nonlinear distortion occurs when there are frequencies present during playback that were not present in the original source.

[Don Richard]

Generally, we try to use visual language to describe audible conditions where we just don't have words. Distortion is completely vague.

Not really. There is linear distortion, meaning a difference in level, or a difference in frequency response between two examples of an audio event. Then there is nonlinear distortion, which most people mean when they say, "distortion". Nonlinear distortion occurs when there are frequencies present during playback that were not present in the original source.

I think you are referring to the few types of distortion we measure with instruments. I am referring to the kind of distortion between what I think it should sound like and what it does sound like. For example, we don't have a "grain-o-meter," and yet it is easy to hear the difference in grain structure from amp to amp. And that is a form of distortion.

What causes "graininess" in an amplifier?

[Deang]

I've heard Nelson Pass talk about this stuff before, and it led him to design the way he does.

"Typically using MOSFETs as the gain devices, (often as current sources rather than voltage sources), most of his designs operate in Class A, with all the low distortion of small signal benefits generally associated with that. He is also known for his advocacy of "minimalist" designs, which reduce the circuit complexity and number of components through which the audio signal must travel, resulting in any residual distortion having a very simple distortion spectrum. His more powerful designs that go beyond what can practically be realized in Class A (such as the Threshold and Pass Labs products) operate in Class AB1 but retain a heavy bias so that they operate in Class A during quieter passages, and great emphasis has been given to avoiding or minimizing switching effects."

Some opamps sound grainy, but I don't know why.

[Islander]

I get the argument that bias is part of the experience, but that doesn't mean it is always a "good thing", or that it is beneficial to the audio community as a whole. The fundamental audiophile is supposed to be enjoying the auditory experience (at least that's the boring strict definition that I understand). To the extent that non-auditory factors affect perception, they are distracting from the goal of the pure audiophile.

It's hard to get rid of bias because it's a survival skill that's probably been hard-wired into us for hundreds of millennia. Like prejudice, bias is a mental shortcut we use to quickly evaluate a situation and determine the presence of a threat, based on previous experience with a similar situation.

You can imagine a primitive man spotting a bear in the distance: "Hmm, when I saw one of those large creatures before, it tried to eat me, so this one probably will as well. I'd better get out of here before it sees me."

Avoiding bias takes mental effort, and sometimes some mental tricks. It sounds like Dr. Who is aware of this, and knows what to do about it, in spite of being an audiophile.

[DrWho]

...it is easy to hear the difference in grain structure from amp to amp.

Haha, nice! I'm gonna borrow that one at work.

[mungkiman]

The following words aren't mine, but I do find them to be interesting:

"The failure of a given individual or group to share my views arises from one of three possible sources -

1. The individual or group in question may have been exposed to a different sample of information than I was (in which case, provided that the other party is reasonable and open minded, the sharing or pooling of information should lead us to reach an agreement);
2. The individual or group in question may be lazy, irrational, or otherwise unable or unwilling to proceed in a normative fashion from objective evidence to reasonable conclusions; or
3. The individual or group in question may be biased (either in interpreting the evidence, or in proceeding from evidence to conclusions) by ideology, self-interest, or some other distorting personal influence."

[DrWho]

Grainy opamps.....while some inherently have higher distortion, I've found that there are often external factors that dominate this kind of sound. I'm not sure how many are public knowledge, but it's pretty well known that the layout of the circuit around the opamp can impact distortion.

If we wanna talk crazy impacts from wiring, then let's discuss the location and size of traces on a circuit board

The opamp benefits from crazy huge slew rate to cancel distortion via feedback. That only works well when your parasitic inductance doesn't prevent the slew from happening....and then those large instantaneous currents gotta be kept away from your signal lest the rectification mix with the audio and now you've got mix products and am modulation.....all low level of course so your average non-signal energy (distortion) doesn't show up in an integrated energy measurement per say, but you'll see it in an fft.

There's lots more going on, but this is a fun one to talk about because it's dominated by the "wiring".

The quality of the noise floor is another source of grainy....and that's complicated by the "real" PSRR of the opamp, not what is measured for the datasheet.

[DrWho]

I just wanted to add that I've seen a lot of bad layouts in audio circuits.....even using the fancy expensive opamps. I think this would really make it hard to subjectively pick an opamp as a favorite.

....especially once you start factoring in the rest of the system.

[DrWho]

We use all sorts of weird language to describe what we're hearing.

Just to be clear, I have no issue with the flowery language. The fake thing would be more akin to a book review where it's clear the person never read the entire book, or maybe didn't understand some of the obvious literary themes. The annoyance factor increases when that person claims expertise in the field. Btw, not saying you do this Mark, but this subconscious dishonesty is rampant in this hobby.

Here's a perfect example:

At an audio festival in Detroit, this older gentleman set up a room to educate the "dumb youth" about classical music and how it's supposed to sound. He had an entire room full of his own speaker creations that were "dialed in over several years of listening". What he demonstrated for me was essentially a heresy with no midrange horn, and the drive units he had weren't capable of a full two way.....Basically there was a big hole in the midrange. You can imagine my astonishment when he put on a piece that I've actual performed, and the entire brass section was almost completely missing due to the gaping hole in the middle. I asked him how he would describe the trumpet melody and received a lecture that this piece was never performed with brass instruments. I just left the room in utter shock and unbelief that this guy spent all this money to pound his crazy dogma. Even more crazy were the other naive people in the room trying to use flowery language to describe the sonic nightmare because they wanted to associate with his mission about classical music.

I wish I could say this was isolated, but it's been the majority of my audiophile experience. When talking book/poetry reviews, or even the book/poetry itself, there is some consistency where each flowery word carries some common meaning...or the artistic tension is going against this commonality to further enhance a description. This is why I know what you (specifically you) mean when talking about grain and we've never been in the same room. With other audiophiles, they'll use words in complete ignorance of the common meaning. Sure, it's their right but they're still clueless. I'm just hesitant to offer up the veil of subjectivity in defense of this insanity.

Edited February 14, 2014 by DrWho

[Don Richard]

The opamp benefits from crazy huge slew rate to cancel distortion via feedback. That only works well when your parasitic inductance doesn't prevent the slew from happening....and then those large instantaneous currents gotta be kept away from your signal lest the rectification mix with the audio and now you've got mix products and am modulation.....all low level of course so your average non-signal energy (distortion) doesn't show up in an integrated energy measurement per say, but you'll see it in an fft. There's lots more going on, but this is a fun one to talk about because it's dominated by the "wiring". The quality of the noise floor is another source of grainy....and that's complicated by the "real" PSRR of the opamp, not what is measured for the datasheet.

So is it fair to say that "grain" can be used to describe an audible consequence as a result of low level IMD, power supply noise, crossover distortion, and/or TIMD? Any other possibilities?

[Deang]

all low level of course so your average non-signal energy (distortion) doesn't show up in an integrated energy measurement per say, but you'll see it in an fft.

It there a way that you can help me understand this better? I'm convinced that this low level stuff is audible and is largely responsible for the things some people hear and others don't. I wish I knew why some can't hear it, I gave up on that one a long time ago. I guess I need to study FFT.

Edited February 14, 2014 by DeanG

[DrWho]

The opamp benefits from crazy huge slew rate to cancel distortion via feedback. That only works well when your parasitic inductance doesn't prevent the slew from happening....and then those large instantaneous currents gotta be kept away from your signal lest the rectification mix with the audio and now you've got mix products and am modulation.....all low level of course so your average non-signal energy (distortion) doesn't show up in an integrated energy measurement per say, but you'll see it in an fft. There's lots more going on, but this is a fun one to talk about because it's dominated by the "wiring". The quality of the noise floor is another source of grainy....and that's complicated by the "real" PSRR of the opamp, not what is measured for the datasheet.

So is it fair to say that "grain" can be used to describe an audible consequence as a result of low level IMD, power supply noise, crossover distortion, and/or TIMD? Any other possibilities?

To be honest, when I pull out the measurement gear I usually stop using subjective language, and start describing auditory traits with quantifiable attributes. It's not until the measurement gear goes away that I start using subjective language again - in fact, it's a conscious exercise to remove that expectation bias.

I've yet to find an auditory artifact that we couldn't somehow track down to a quantifiable metric, but my experience is that it is far harder to measure something before hearing it. It can take weeks to root cause complicated interactions that are readily audible within a few seconds of listening. I'm by no means a seasoned expert, but there's already things I hear that I don't need to run to measurements to fix. Is that then a subjective or objective exercise?

Is it possible to objectively use our ears? And is it possible to subjectively pick component values?

[DrWho]

all low level of course so your average non-signal energy (distortion) doesn't show up in an integrated energy measurement per say, but you'll see it in an fft.

It there a way that you can help me understand this better? I'm convinced that this low level stuff is audible and is largely responsible for the things some people hear and others don't. I wish I knew why some can't hear it, I gave up on that one a long time ago. I guess I need to study FFT.

I tried to find an easy link to describe what's going on here, but I think this may be one of those things that is dominated by internal knowledge. I did find this though which is a super basic description of how things can show up in an FFT of an opamp:

http://ww1.microchip.com/downloads/en/AppNotes/00681a.pdf

On Page 8 they talk about clipping being the reason for the crappy FFT....that's example of a circuit that would sound grainy at low levels. Talk about hooooorible clipping behavior. That translates back down to the low-level behavior too - it's just harder to see.

[DrWho]

I came across this too:

http://www.exasound.com/e20DAC/Measurements.aspx

...which kinda made me laugh. Their noise floor plots with and without galvonic isolation on the USB cable is actually mitigating a ground loop between their APx525 and their computer's power supply. If they used a laptop running off a battery for their interface with the APx525, then those spurs would go away

The power supply noise coming through in the ~50kHz region is interesting too, and those spurs will move around with the audio input signal, and depending on your downstream device they can mix down into the audible band. The reason you don't see these spurs with zero's being sent to the DAC is because you're pulling yourself closer to your reference (ground in this case is considered to have no noise). With a full scale signal, your impedance to ground is higher and your impedance to the power supply is lower, which means you get more power supply coupling with a larger output signal. The amplitude of these spurs then AM modulate with your input signal, which is a form of mixing. This is why you get spurs that aren't related harmonically to your "single" input signal. In reality, the power supply is imparting the secondary signal to create unexpected IMD artifacts.

[Don Richard]

All good stuff, Mike. Thanks.

[Deang]

Definitely. Nice job Mike.

[DrWho]

The problem is multiple unrelated factors can manifest themselves as a "grainy" sound. For any individual metric, I think we can set up your row of amplifiers. I know I've done it at work for other attributes to determine when the masking occurs...

To try and quantify grainy across the board would require way too many measurements. It's easier to just listen to it

[DrWho]

I'm trying to think of any caveats, but in general I would say yes. Without knowing what's happening up front, I think there may be some variation in language, but overall I think there will be some level of similarity.

I've done this a couple times for my own projects at work so I'm really speaking from my own limited experience. Unfortunately I don't get to explore everything I'm curious about. It is a business afterall.

[BLSamuel]

Not having the time to read through all these posts as I stumbled upon getting ready to leave work.

Did anybody ask one of Marty's original questions about the wire from turntable to phono pre-amp? The signals are very tiny so I think good shielding to keep out interference would be important?

I'll have to go find the links in the other thread DeanG posted.... probably won't understand them.

I used to think bigger was better but the 6 gauge jumper cables are kind of heavy to hook up to my little t amp.