HOLO Audio MAY DAC

[The Computer Audiophile]

3 minutes ago, hopkins said:

or not in command of the English language.

I said perhaps my English wan't good enough. 

[DuckToller]

40 minutes ago, The Computer Audiophile said:

In my younger days, I would've asked what you are smoking. Now that I'm older and wiser, I'll ask what you're also drinking with that smoke. 

 

Did you read about the European heatwaves this summer ?

We are back at 95´°F and more in Paris and elsewhere in France, Bordeaux is facing a wall of smoke and fire, and that's for about several weeks now with some small lows on 77 °F for a day or two ...
IMHO this heat may spurn the loss of civility in conversations ... sometimes for some people at some places, but surely not for my friend hopkins  ;-)

[Miska]

4 hours ago, ssh said:

Hey @Miskahere can one find this list?

 

It's on the HQPlayer product page. I should add some more stuff there, such as Intona USB isolators.

 

[ssh]

Thanks!

[Miska]

2 hours ago, hopkins said:

All of a sudden we can assess the accuracy of an infinite number of DAC variations simply based on NOS vs OS?

 

OK, let's go this through some math.

 

To correctly reconstruct 44.1 kHz 24-bit data, with 20 kHz bandwidth, you need flat frequency and phase response up to 20 kHz. This leaves you with 44.1 / 2 - 20 = 2.05 kHz wide transition band. Or if you are ready to accept some imaging, 4.1 kHz wide transition band. Now you just need a reconstruction filter that reaches -144 dB attenuation (24-bit resolution) by 22.05 or 24.1 kHz. This without phase error at 20 kHz.

 

Now you can realize such filter is practically impossible in analog domain. Doing this with a digital filter is not a problem. Not even one with -300 dB attenuation with the same transition band.

 

Another problem is that running NOS at 44.1 kHz rate gives you severe roll-off in top octave reaching -6 dB at 22.05 kHz. This is result of the sample-and-hold function.

[copy_of_a]

27 minutes ago, Miska said:

 

It's on the HQPlayer product page. I should add some more stuff there, such as Intona USB isolators.

 

Loosely related to the discussion above (USB noise measurements) on the Intona Forum the Designer posted measurements of the "7055-C", also some comments as far as comparison the the "7055-B" goes:

https://forum.intona.net/viewtopic.php?f=10&t=11

On top they put out linear 5V through USB (in case you want to power an appropriate 5V USB powered DAC).

[Miska]

12 minutes ago, copy_of_a said:

Loosely related to the discussion above (USB noise measurements) on the Intona Forum the Designer posted measurements of the "7055-C", also some comments as far as comparison the the "7055-B" goes:

https://forum.intona.net/viewtopic.php?f=10&t=11

On top they put out linear 5V through USB (in case you want to power an appropriate 5V USB powered DAC).

 

For non-USB powered DACs, 7054 already performs all necessary functionality based on my measurements. Since the USB power is not used for any noise sensitive purposes, so galvanic isolation is main factor there.

 

I have two 7054's and two 7055-C's so far.

 

[copy_of_a]

6 minutes ago, Miska said:

I have two 7054's and two 7055-C's so far.

I have one 7054 and one 7055-B. But the latter more or less only because all the cabling is on the rear. As far as audio goes I couldn't tell a difference (by ear ☺️ ).

Cool thing is the 7054 still gets used in my desktop system.

[Qhwoeprktiyns]

54 minutes ago, Miska said:

Another problem is that running NOS at 44.1 kHz rate gives you severe roll-off in top octave reaching -6 dB at 22.05 kHz. This is result of the sample-and-hold function.

 

That is correct, but no one hears at 22,kHz, and at high frequencies the speaker (Tweeter) response has much more importance than the slight roll-off you get from a NOS DAC, IMO. You don't need to be an "expert" to know this, all you need to do is listen. 

 

I did not mean to get into a lengthy discussion, and will stop here. It is all pointless anyway. 

 

 

[The Computer Audiophile]

1 hour ago, Miska said:

severe roll-off in top octave reaching -6 dB at 22.05 kHz.

 

 

2 minutes ago, hopkins said:

at high frequencies the speaker (Tweeter) response has much more importance than the slight roll-off you get

Ah, the master at twisting words to fit an untrue narrative. 

[Allan F]

17 minutes ago, hopkins said:

I did not mean to get into a lengthy discussion...

 

With all due respect, the evidence strongly suggests otherwise.

[Qhwoeprktiyns]

27 minutes ago, The Computer Audiophile said:

 

 

Ah, the master at twisting words to fit an untrue narrative. 

 

Congratulations, one more useless comment. You really bring a lot to the conversation. 

 

-3db at 20khz with 44khz files is the figure commonly stated. 

[GoldenOne]

1 hour ago, copy_of_a said:

Loosely related to the discussion above (USB noise measurements) on the Intona Forum the Designer posted measurements of the "7055-C", also some comments as far as comparison the the "7055-B" goes:

https://forum.intona.net/viewtopic.php?f=10&t=11

On top they put out linear 5V through USB (in case you want to power an appropriate 5V USB powered DAC).

Realistically I think as @Miska said unless the DAC is USB powered, almost any of them are going to perform the same, they're all incredibly low noise.

I've got three 7055-C's and a 7055-D and have not been able to hear (or measure) any difference between the C and D with any DAC so far.
Also for DACs like the Holo May where the isolation is done internally anyway, it's not needed.

Though to be honest, if your DAC IS USB powered then it's questionable whether the money would be better spent on an isolator or upgrading the DAC to begin with.

[sworksone]

Maybe everyone needs to just take a step back and have a drink or two of choice, and get back to what was a good discussion?

 

 

[sworksone]

16 minutes ago, GoldenOne said:

I tried a 7055-C between my Innuos Zenith Mk3 and May KTE following the Stereophile review, and didn’t find it provided any subjective benefit contrary to the reviewers comments.    This was as I expected given the internal isolation in May but thought it was worth a shot.  

[Account Closed]

How about we return this thread to its original intent and start a new thread for all this extraneous stuff. 

[Miska]

1 hour ago, hopkins said:

That is correct, but no one hears at 22,kHz, and at high frequencies the speaker (Tweeter) response has much more importance than the slight roll-off you get from a NOS DAC, IMO. You don't need to be an "expert" to know this, all you need to do is listen. 

 

You have things called intermodulation distortion, etc. And any short-term events (transients) will be incorrectly rendered, because most of the time, sampling point at low sampling rate won't coincide with the signal change, but instead somewhere between the samples. For that reason, reconstruction always needs to take multiple samples into account and fit those on the sinc function, to correctly reconstruct any such effect. Otherwise your transient response and level will vary depending on your luck how the sampling point happens to coincide with the transient.

 

ADC disassembles the signal into data using a sinc function. And then DAC needs to reconstruct it back. NOS DAC running at 44.1 kHz completely fails on this reconstruction.

 

For me, slight roll-off is certainly less than 0.5 dB. I'd rather have it 0.1 dB or less.

 

1 hour ago, hopkins said:

I did not mean to get into a lengthy discussion, and will stop here. It is all pointless anyway.

 

Yeah, it is generic math and not specific to any particular DAC.

 

[Axiom05]

The exact amount of roll-off at 20KHz is less important than where the roll-off starts. -3 dB at 20 KHz may not be particularly audible but -1.5 dB at say 10 or 12 KHz is quite audible. You can look at the roll-off as having a low Q which, since it is occurring over two octaves (according to Stereophile for the May) should be quite audible compared to something that is flat to 20 KHz.

[Qhwoeprktiyns]

5 hours ago, Miska said:

 

You have things called intermodulation distortion, etc. And any short-term events (transients) will be incorrectly rendered, because most of the time, sampling point at low sampling rate won't coincide with the signal change, but instead somewhere between the samples. For that reason, reconstruction always needs to take multiple samples into account and fit those on the sinc function, to correctly reconstruct any such effect. Otherwise your transient response and level will vary depending on your luck how the sampling point happens to coincide with the transient.

 

ADC disassembles the signal into data using a sinc function. And then DAC needs to reconstruct it back. NOS DAC running at 44.1 kHz completely fails on this reconstruction.

 

For me, slight roll-off is certainly less than 0.5 dB. I'd rather have it 0.1 dB or less.

 

 

Yeah, it is generic math and not specific to any particular DAC.

 

 

Oversampling a 44khz recording is not providing more guarantee that you "coincide" with what was the original analog signal. 

 

The limits of NOS are well known, but once again, you are only looking at one aspect of DAC performance, through this single metric.

 

Oversampling is your bread-winner, I understand that you would focus on this, but you cannot conclude that a DAC is going to be more "accurate" by only looking at this one aspect.

 

So yes, it is generic math, not specific to any DAC, but it is not sufficient to evaluate "accuracy". It's as if you evaluated speakers by only looking at a frequency response curve. 

 

You probably believe that everything else is moot, but you have no math to rely on here, it's only guesswork.

 

I suggest we stop here. 

 

 

 

[musicjunkie917]

11 minutes ago, hopkins said:

I suggest we stop here. 

 

Lol! You don't get to tell @Miska that he does't know what he is talking about and then tell him to stop. He DOES know what he is talking about and it is clear that you do not. It's kind of like a soldier at his first day of training taking on Achilles in a fight to the death.....we all know who is going to lose...

[Qhwoeprktiyns]

28 minutes ago, musicjunkie917 said:

 

Lol! You don't get to tell @Miska that he does't know what he is talking about and then tell him to stop. He DOES know what he is talking about and it is clear that you do not. It's kind of like a soldier at his first day of training taking on Achilles in a fight to the death.....we all know who is going to lose...

 

There is more to DACs than sampling theory, only an idiot would deny that. 

[The Computer Audiophile]

 

 

59 minutes ago, hopkins said:

I suggest we stop here.

 

Please follow your own advice

[Miska]

15 hours ago, hopkins said:

Oversampling a 44khz recording is not providing more guarantee that you "coincide" with what was the original analog signal. 

 

Yes it will, that is mathematical fact. Higher the oversampling rate, closer you get. And it gets more and more feasible for analog low-pass filter to do the rest.

 

15 hours ago, hopkins said:

The limits of NOS are well known, but once again, you are only looking at one aspect of DAC performance, through this single metric.

 

What else?

 

I'm looking at many factors of the analog output performance. D/A conversion hardware is only part of it, DSP front-end is as important. And the combination defines ultimate performance you can get.

 

15 hours ago, hopkins said:

Oversampling is your bread-winner, I understand that you would focus on this, but you cannot conclude that a DAC is going to be more "accurate" by only looking at this one aspect.

 

You need to tick all "accurate" boxes, you cannot leave any of the boxes unticked.

 

I'm not looking at only one aspect. I'm looking at every aspect, and not a single one can left imperfect.

 

One imperfect aspect can spoil the result.

 

15 hours ago, hopkins said:

So yes, it is generic math, not specific to any DAC, but it is not sufficient to evaluate "accuracy". It's as if you evaluated speakers by only looking at a frequency response curve. 

 

If speaker frequency response is screwed, the speaker is screwed. It is one of the aspects that must be perfect. Although that is one of the aspects you can try to correct with DSP.

 

15 hours ago, hopkins said:

You probably believe that everything else is moot, but you have no math to rely on here, it's only guesswork.

 

Math and physics are the baseline you build upon.

 

15 hours ago, hopkins said:

I suggest we stop here. 

 

Well, you didn't...

 

[Qhwoeprktiyns]

45 minutes ago, Miska said:

 

Yes it will, that is mathematical fact. Higher the oversampling rate, closer you get. And it gets more and more feasible for analog low-pass filter to do the rest.

 

 

What else?

 

I'm looking at many factors of the analog output performance. D/A conversion hardware is only part of it, DSP front-end is as important. And the combination defines ultimate performance you can get.

 

 

You need to tick all "accurate" boxes, you cannot leave any of the boxes unticked.

 

I'm not looking at only one aspect. I'm looking at every aspect, and not a single one can left imperfect.

 

One imperfect aspect can spoil the result.

 

 

If speaker frequency response is screwed, the speaker is screwed. It is one of the aspects that must be perfect. Although that is one of the aspects you can try to correct with DSP.

 

 

Math and physics are the baseline you build upon.

 

 

Well, you didn't...

 

 

 

In the past few years, I've had the chance to exchange frequently (and in depth) with a DAC designer about the many issues that are brought up here. I am not an expert, but the one thing I have taken away from these fascinating exchanges is this: there is no perfect solution in audio - everything is a compromise, every new solution has its limitations and introduces new problems. Yes, its obviously always physics behind. 

 

Its seems to me that you live in a fantasy world. In this world, an OS DAC is "accurate" :)

 

Forums are not the right place to have in-depth discussions.  If you really feel the need to reply to this message I'll try to refrain myself from responding again, or perhaps if you are interested we can discuss this by email. Sorry for having derailed this thread.

[Allan F]

1 hour ago, hopkins said:

In the past few years, I've had the chance to exchange frequently (and in depth) with a DAC designer about the many issues that are brought up here. I am not an expert...

 

FYI, @Miskais not only the developer of HQPlayer. he is also a DAC designer (design available free on his website). He is an expert. So your resort to the logically fallacious "appeal to authority" does not carry any weight on this forum. So please try to exercise some discipline and do "refrain [myself] from responding again".