At What Point Does Reducing Electrical Noise Reach No Audible Difference?

[rusty jefferson]

...In layman terms l’m referring to soft passages or sounds between subtle transients such as a flute in the background while a piano is playing. This is where the fidelity can suffer. That is because those subtle sounds are much closer to the noise floor. The human ear has more difficulty translating certain sounds as the level drops.

All this happens very quickly with recorded music so the ear - brain is working hard to pick up everything. If the sounds closer to the noise floor aren’t accurately made out then the perception of what’s heard will change. This is because most of the dynamic range in recorded music sits below your normal listening level. If these subtle sound aren’t as pin sharp as the louder sounds then Houston we have a problem.

Of course this is just one facet of high fidelity sound reproduction but one that is constantly overlooked....

Completely agree with this statement and what I was referring to when I mentioned the internal PS possibly stepping on the delicate digital information. You're not going to hear tone or timbre differences between the 2 power supplies but likely a difference in low level detail, and/or spatial cues. Choice of music is critical for being able to more easily identify this.

[Mr. Widget]

I find your post is without sufficient technical understanding between measurements and what people are actually able to discern.

Hi Ian, you have made similar statements in the past. There is nothing in my post on this thread that suggests any level of technical competence or lack of. I would suggest you find my 20+ years of posting on this site to be technically insufficient to satisfy you.

My initial point regarding the audibility of a reduction in power supply introduced noise/distortion was that while we can measure improvements at some point they are inaudible. I did not state that I was able to definitively state at what level further improvements are inaudible. My point was that we trust our ears and I think that can be a mistake. Rusty is right... it is possible that removing the PS may improve the audible performance, I am just stating to be certain, we really need to verify by blind/double blind test.

That’s why l made the comment about your Lydorf processor. If l replaced it with another processor without you knowing it would you realise it?

Psycho acoustics isn’t an exact science and Nelson Pass has stated there is still a long way to go before we can rely on measurements alone to validate what we hear.

I don't think you understand what I am proposing to evaluate.

The purpose of my exploration is to see if I and others can "hear" a digital pre/pro. In this case it will be a Lyngdorf MP-40 2.1. I will be comparing two Roon endpoints through a very high quality analog system. The comparison will be using the same digital source code through a Mark Levinson No 519 with its custom 32 bit link DAC and the Lyngdorf. The idea is that I want to know if a multi-stage digital pre/pro can audibly match the performance of a $25K No 519 which is at or near the state of the art.

Simply, I want to find out if it is possible to hear a difference between a dedicated high end streamer/DAC and a pre/pro that is substantially less expensive and has a much more complicated signal path. According to the subjectivists, the No 519 should be more "musical", "transparent", pick your adjective... according to the objectivists there should be no audible difference. I don't have an opinion and hence the reason for the exercise.

Psycho acoustics isn’t an exact science and Nelson Pass has stated there is still a long way to go before we can rely on measurements alone to validate what we hear.

If you look at some THD measurements versus amplitude of preamps in Stereophile test reports the distortion rises very rapidly at low levels below 100 - 200 mv.

But it’s not distortion. It’s the residual noise because the analyser doesn’t discriminate between noise and distortion. At low amplitudes which are often the nominal listening levels in a room the noise is significant in relative terms.

In layman terms l’m referring to soft passages or sounds between subtle transients such as a flute in the background while a piano is playing. This is where the fidelity can suffer. That is because those subtle sounds are much closer to the noise floor. The human ear has more difficulty translating certain sounds as the level drops.

All this happens very quickly with recorded music so the ear - brain is working hard to pick up everything. If the sounds closer to the noise floor aren’t accurately made out then the perception of what’s heard will change. This is because most of the dynamic range in recorded music sits below your normal listening level. If these subtle sound aren’t as pin sharp as the louder sounds then Houston we have a problem.

Since I am not a manufacturer or hobbyist interested in building electronics, I have no interest in how you measure differences in electronics, but I agree that psycho acoustics is an area that requires much more study.

My goal is pretty simple. I want to see if we can reliably hear ANY differences between digital circuits/DACs. For evaluation I am planning on using 5 tracks that Sean Olive uses for evaluation. If you have a specific recommendation of some cuts that have these subtle transients I'd be interested in adding them to the playlist.

I agree that some musical tracks are better at demonstrating differences than others and I also believe the differences can affect soundstage and not necessarily frequency response, or affect apparent "clarity" and not make obvious distortion or frequency response changes. I do not think noise, distortion, and frequency response improvements are the only changes that we are capable of hearing.


Widget

[Mr. Widget]

If you want more info on the devices that I plan on comparing see the attached documents.

The rest of the system is described here: https://www.audioheritage.org/vbulle...l=1#post424651

The sources that were described in the Project Widget thread have been updated, but the analog system is still the same.


Widget

[toddalin]

Why would someone even go to the effort of a double blind test if one piece of equipment in the test had an obvious peak or dip in its frequency response at 1khz? Voltage matching at 1khz is the only method for comparing equipment in listening tests I've seen used. I've not compared equipment of radically different specifications. We're usually trying to compare things that measure similarly (as Mr Widget is planning) to see if we can hear a difference between them, or as in the Stereophile example, if an 'upgraded' power supply can have an audible difference in a component.

We are including speakers here. It doesn't need to be a "peak" per se. If the 1kHz is just 0.5 dB reduced from the rest of the spectrum and you turn it up to compensate, while turning up the remainer of the spectrum, you've essentially made one unit 0.5 dB louder than the other and this difference is detectable.

[toddalin]

I agree that some musical tracks are better at demonstrating differences than others and I also believe the differences can affect soundstage and not necessarily frequency response, or affect apparent "clarity" and not make obvious distortion or frequency response changes. I do not think noise, distortion, and frequency response improvements are the only changes that we are capable of hearing.


Widget

I've always thought that the older Hammonds should be good for such a demonstration, especially at "Red Book" resolution.

This is because the older Hammonds (especially mine that had all of the electronics removed) more "suffer" from "key click" due to keying on a transient. Hammond reduced this click by using special materials on the buss bars to "soften" the contact. If the transient doen't fall at the beginning of a digital "cycle" it seems like at least some of it should be lost. Essentially, what the Hammond does is put a "square wave leading edge" on the sine waveform that may or may not be caught.

https://www.sweetwater.com/insync/key-click/

https://www.youtube.com/watch?v=ftqAMvhfLiE

[rusty jefferson]

We are including speakers here...

No, no we're not including speakers here. We've been talking about electronic equipment. Speaker comparisons are quite different especially in a home audio application where it would require 2 sets of identical amplifiers for instantaneous switching of your A/B/X box. There is no practical application for a double blind test of different speakers in your living room. Leave it to Harman.

I do appreciate your original thread starting discussion about potential noise in a piece of equipment and whether or not we'd be able to hear it, but we have drifted...

[Mr. Widget]

Leave it to Harman.

That was my thought too... without a speaker shuffler most loudspeaker AB or ABX comparisons are not particularly accurate.

I do appreciate your original thread starting discussion about potential noise in a piece of equipment and whether or not we'd be able to hear it, but we have drifted...

Apologies for derailing yet another thread.

Once I've received the equipment to test and set up the experiment I will start a new thread. I will start the thread before I conclude the testing as there will likely be valuable suggestions on how to better perform the experiment.


Widget

[toddalin]

No, no we're not including speakers here. We've been talking about electronic equipment. Speaker comparisons are quite different especially in a home audio application where it would require 2 sets of identical amplifiers for instantaneous switching of your A/B/X box. There is no practical application for a double blind test of different speakers in your living room. Leave it to Harman.

I do appreciate your original thread starting discussion about potential noise in a piece of equipment and whether or not we'd be able to hear it, but we have drifted...

Well no. I was also discussing minute differences from crossover capacitors and speakers by extension. I think I am getting more "noise" from the Audyn caps than the oil can/Clarity combo. Why limit this?

Why would one use two amps? That would distort the results because the two amps would not be identical. You need to change between the speakers on the same amp and maintain the volume setting at the mic.

Why does there need to be a practical application? What if I just want to compare my L200/300s to my L112s at the same volume level and be able to switch between them instantly? What if I wanted to know which images better?

[Mr. Widget]

What if I wanted to know which images better?

I think if you want to know if one speaker images better than the other and you want to know how they REALY image you will need remove speaker pair A and replace it with speaker pair B like Harman does with their shuffler.

Obviously this is tough to do so most of us sit the speakers near each other and swap back and forth. This may give you an idea of how they compare, but not necessarily an accurate representation of what they are capable of.


Widget

[toddalin]

Imaging was just a "for instance" and I'm sure that if you were to really think about it, one could come up with other uses.

I made this one for the AudioKarma crowd and it allows us to switch between any two line level components and/or speaker systems, all with line level matching. This is purely passive and uses Alps pots. In addition to what you see, I've subsequently added four 261 ohm, 30-watt resistors that can be switched to an amplifier's outputs and stays with that amp. This then alleviates the possibility of leaving the speaker terminals unloaded, even for the time it takes for the relay to fire, and was added to satisfy the tube guys who worry about "open outputs" on the transformers. The guys can't wait to try it out.

This one does not include the "X" feature, but is controlled either manually or by remote and the telltales may or may not be activated. An operator could serve as the "X" function if desired.

https://cimg5.ibsrv.net/gimg/www.cor...673907133f.jpg
https://cimg7.ibsrv.net/gimg/www.cor...4374d1a680.jpg

[Ian Mackenzie]

Hi Ian, you have made similar statements in the past. There is nothing in my post on this thread that suggests any level of technical competence or lack of. I would suggest you find my 20+ years of posting on this site to be technically insufficient to satisfy you.

My initial point regarding the audibility of a reduction in power supply introduced noise/distortion was that while we can measure improvements at some point they are inaudible. I did not state that I was able to definitively state at what level further improvements are inaudible. My point was that we trust our ears and I think that can be a mistake. Rusty is right... it is possible that removing the PS may improve the audible performance, I am just stating to be certain, we really need to verify by blind/double blind test.


I don't think you understand what I am proposing to evaluate.

The purpose of my exploration is to see if I and others can "hear" a digital pre/pro. In this case it will be a Lyngdorf MP-40 2.1. I will be comparing two Roon endpoints through a very high quality analog system. The comparison will be using the same digital source code through a Mark Levinson No 519 with its custom 32 bit link DAC and the Lyngdorf. The idea is that I want to know if a multi-stage digital pre/pro can audibly match the performance of a $25K No 519 which is at or near the state of the art.

Simply, I want to find out if it is possible to hear a difference between a dedicated high end streamer/DAC and a pre/pro that is substantially less expensive and has a much more complicated signal path. According to the subjectivists, the No 519 should be more "musical", "transparent", pick your adjective... according to the objectivists there should be no audible difference. I don't have an opinion and hence the reason for the exercise.


Since I am not a manufacturer or hobbyist interested in building electronics, I have no interest in how you measure differences in electronics, but I agree that psycho acoustics is an area that requires much more study.

My goal is pretty simple. I want to see if we can reliably hear ANY differences between digital circuits/DACs. For evaluation I am planning on using 5 tracks that Sean Olive uses for evaluation. If you have a specific recommendation of some cuts that have these subtle transients I'd be interested in adding them to the playlist.

I agree that some musical tracks are better at demonstrating differences than others and I also believe the differences can affect soundstage and not necessarily frequency response, or affect apparent "clarity" and not make obvious distortion or frequency response changes. I do not think noise, distortion, and frequency response improvements are the only changes that we are capable of hearing.


Widget

We agree to disagree.

Anyone even unfamiliar with this hobby would question your post. It’s high school clear English. You made the claim without supporting it with any logic, relevant evidence or foundation of proof. As if anyone reading it is that stupid to believe you.

It was a glib remark made to undermine the credibility of the Stereophile test as you saw and you wanted anyone reading it to believe it. That’s misinformation.

Adding insult that you didn’t understand my post. Unbelievable.

[Mr. Widget]

We agree to disagree.

Anyone even unfamiliar with this hobby would question your post. It’s high school clear English. You made the claim without supporting it with any logic, relevant evidence or foundation of proof. As if anyone reading it is that stupid to believe you.

It was a glib remark made to undermine the credibility of the Stereophile test as you saw and you wanted anyone reading it to believe it. That’s misinformation.

Adding insult that you didn’t understand my post. Unbelievable.

Unbelievable perhaps, but I really have no idea what you are getting at.

FWIW: The only claim I made was that we need to follow Harman's lead and use blind/double blind testing if we want valid human feedback.

My premise that sighted bias was the reason the Stereophile reviewers came to their conclusions could be wrong, but I think we all need to open our minds to the possibility that we have been fooling ourselves for years about what is and is not an audible improvement.

I hope to have more clarity on this in the coming weeks.


Widget

[toddalin]

Widget, this is a good track to audition. On so many systems it just sounds like noise. On a good system you can hear each clavinet in space. Use the original, unremastered version if possible (noisier and harder to resolve).

https://youtu.be/9c9KvU6cE6E

You may also want to consider tracks that were intentionally recorded bright (e.g., Boston, Peter Gabriel) and listen for "splashyness"

And speaking of "splashyness" this is where I find the difference between the Audyn and oil can caps with the Audyns being just a bit "splashier." Again, has to be the "right cut" to hear the difference. I ordered a pair of Mundorf EVO oils to try out.

This also brings up the problem with auditions and testing sessions. None of the pieces that you select may show the differences of interest and it takes regular listening to your regular tracks with lots of "back and forth" to find that one instance where there may be a difference, then queueing in on this one instance to try to improve it without doing "damage" to anything else.

[Mr. Widget]

This also brings up the problem with auditions and testing sessions. None of the pieces that you select may show the differences of interest and it takes regular listening to your regular tracks with lots of "back and forth" to find that one instance where there may be a difference, then queueing in on this one instance to try to improve it without doing "damage" to anything else.

Agreed that music content does make a big difference. Sean Olive who has been doing this as his day job for decades has picked a number of cuts that listeners can most reliably detect sonic differences. As it turns out pink noise is about the best for frequency response deviations and Tracy Chapman's Fast Car is often used when listening for various distortions.

Here is a chart of music types he has tried and their success rate for evaluating systems.


Widget

[toddalin]

To be clear, that's frequency response deviations between components, and not from the ideal pink noise curve. That takes a trained ear.

This would make sense because you are hearing all of the frequencies (though not at once), so recognize if any are altered. If one listens to a "cut" it may not even include those frequencies where the differences lie, or they go by before you can recognize the difference.