HOLO Audio MAY DAC

[Extreme_Boky]

1 hour ago, Miska said:

 

No it is not. Why would it be?

 

Because the USB card would be powered by clean, linear regulators. Not DC\DC switching converters. Some streamers do not use them, but there are only a few of those. If we are talking PC, we are automatically talking DC\DC  switching converters. 

 

1 hour ago, Miska said:

 

This is exactly why it is best powered from the USB VBUS (IOW, from USB port). From the +5 VDC lines on USB.

 

In addition it is good because the USB interface will stay alive regardless of power status of the DAC.

 

 

Your solution would make what I itemised below for you yet again, impossible to achieve:

 

No need to use USB cable to feed +5V. No interference with data lines. The USB cable would consist only of a data twisted pair, shielded, and as such would provide much better, optimal impedance matching with USB transceiver requirements et either end. It would be much easer to design, preserve its specifications, and manufacture such USB cable correctly. 

 

Full control over how the USB card is powered within a DAC. I.e. nor relying on an (mostly) inferior USB +5V VBUS regulation and USB cables. Some streamers provide very good solution, but ultimately still (have to) rely on a USB cable variances. 

 

In addition, two pints of impedance degradation caused by contact resistance on that +5V DC line would simply disappear. 

 

Mains (within a DAC) would be sourced from a clean point, not a (usually used) dirty point ruined by switching power supply 110/120  /  230/240 sides.

 

Of course, the USB card can stay powered up at all times with my solution as well. It would draw around 150mA with no data processing.

 

1 hour ago, Miska said:

 

Well, you don't need either one if you have the galvanic isolation the way I explained.

 

I am now convinced that for some reason you don't see  what I am proposing as an ultimate solution to power the USB module whilst ensuring the full galvanic isolation is preserved. The benefits are, once again, listed just above.

 

It is also okay to (continue to) disagree with me or not see what I'm saying.

[Miska]

1 minute ago, Extreme_Boky said:

Because the USB card would be powered by clean, linear regulators. Not DC\DC switching converters. Some streamers do not use them, but there are only a few of those. If we are talking PC, we are automatically talking DC\DC  switching converters. 

 

But there's no particular need to have data-to-data interface have any special kind of power as long as it doesn't affect correctness of the data transferred. USB card doesn't involve any critical timing operations.

 

1 minute ago, Extreme_Boky said:

No need to use USB cable to feed +5V. No interference with data lines. The USB cable would consist only of a data twisted pair, shielded, and as such would provide much better, optimal impedance matching with USB transmitter / receiver requirements et either end. It would be much easer to design, preserve its specifications, and manufacture such USB cable correctly. 

 

Interference is not a problem, for example USB powered HDD's work just fine and can transfer data correctly - at much higher data rates.

 

1 minute ago, Extreme_Boky said:

Full control over how the USB card is powered within a DAC. I.e. nor relying on an (mostly) inferior USB +5V VBUS regulation and bad USB cables. Some streamers provide very good solution, but ultimately still (have to) rely on a USB cable variances. 

 

This is usually a problem, because USB interface power control should be based on presence of VBUS, not some external method. This is main source of problems when people power down such DAC when software is using it.

 

1 minute ago, Extreme_Boky said:

I am now convinced that for some reason you don't see  what I am proposing as an ultimate solution to power the USB module whilst ensuring the full galvanic isolation is preserved. The benefits are, once again, listed just above.

 

It would increase cost and complexity of the DAC without benefits. While implementing it the way I presented provides better galvanic isolation since there may be complete independence of power sources, for example if you are playing from a battery powered laptop.

 

And you can preserve full galvanic isolation between two totally independent power domains. No need for common mains source.

 

[Extreme_Boky]

24 minutes ago, Miska said:

 

But there's no particular need to have data-to-data interface have any special kind of power as long as it doesn't affect correctness of the data transferred. USB card doesn't involve any critical timing operations.

 Any USB cable if done half right will provide correctness, but people have their preferences based on perceived sound quality that various USB cables provide.

24 minutes ago, Miska said:

 

 

Interference is not a problem, for example USB powered HDD's work just fine and can transfer data correctly - at much higher data rates.

 

Just see what I wrote above. Of course, you won't be able to measure the differences but various USB cables sound vastly different. And those that do nor carry +5V (and run 200-250mA in parallel to data lines) sound superior.

 

24 minutes ago, Miska said:

 

 

This is usually a problem, because USB interface power control should be based on presence of VBUS, not some external method. This is main source of problems when people power down such DAC when software is using it.

How can it be a problem if the USB card is powered up at all times, internally, by a clean source?

 

You managed to omit that sentence from my previous post!

24 minutes ago, Miska said:

 

It would increase cost and complexity of the DAC without benefits. While implementing it the way I presented provides better galvanic isolation since there may be complete independence of power sources, for example if you are playing from a battery powered laptop.

 

The laptop would be  a terrible source of +5V DC VBUS even when running of a battery, due to its switching mode power supply used internally, to generate +5V from a higher voltage rail. 

 

24 minutes ago, Miska said:

And you can preserve full galvanic isolation between two totally independent power domains. No need for common mains source.

 

Nonsense. For this, you'd need two separate power feeds from the MV to LV transformer station, straight to your household, with independently grounded neutrals at each point (at the transformer station). You can not have two independent power domains within a household.  

[barrows]

6 hours ago, Extreme_Boky said:

The more I look at it, the more I'm convinced that the USB card simply has to be powered up internally, by means of a separate transformer & voltage regulation - completely isolated from the rest of DAC's ground fill (no common ground with the rest of the DAC), to preserver galvanic isolation. I did it with the DAC I made 10+ years ago, so I'd expect that DAC manufacturers will eventually do the same.

I would say the above is not entirely accurate:  As long as the quality of the USB power is clean, and the USB cable is well designed, there is no need for the USB receiver circuitry to be powered from an internal DAC based supply.  but of course, this means using a USB source with very clean USB power, not a commercial computer USB source, etc.

I have tested such, using an internal USB power supply (fully isolated, dedicated tranny as specified) vs. USB power from a Signature Rendu, and the internal power supply in the DAC approach gave no benefit.  

[Extreme_Boky]

I understand your angle.

 

But, can you understand mine? My approach would do exactly what you do not want to see - it would make your product obsolete to those who have 1/2 understanding of what I'm writing here. 

[barrows]

1 minute ago, Extreme_Boky said:

I understand your angle.

 

But, can you understand mine? My approach would do exactly what you do not want to see - it would make your product obsolete to those who have half 1/2 understanding of what I'm writing here. 

Hahaha!  I was just using the Sonore product as an example of something with a clean USB power line, I suspect there are others.  But if you think that is the only advantage sonically of the Sonore Signature, you are highly mistaken, clearly you have not heard one.  I suspect that it might be wise not to speculate on something which you have no experience of.  Sorry for any detour, back to the May now...

[Extreme_Boky]

You also do not have experience with what I am saying and have been using for more than a decade.

 

Irrelevant of which low voltage regulators your product is using for USB +5V DC, its overall success (the product implementation in a digital audio reproduction sound chain, with an aim to improve a perceived sound quality by us humans) would still be plagued by:

 

a fact that the current will be sent down the USB cable. You do not have any control over the type/quality/length of that USB cable. You simply must have that extra wire, as part of the USB cable, to send that current to power up the USB card.

a fact that the current runs in parallel to USB data lines, unless the USB cable separates this wire from the shielded twisted pair, by design.

a fact that USB connectors need to rely on two points of contact resistance, to transfer that current from your product, to the actual USB card. Again, you do not have control over this neither (gold plated, rhodium plated, the actual surface area of the contact at either end)

 

With a USB card powered up internally, by the DAC, whilst taking the necessary (simple) measures of ensuring galvanic isolation, all of the above is not in play nay more. You are in full control. 

 

The extra benefits you mentioned? Every additional switching, clocking or other processing by means of having to use a power supply line(s)/rail(s) to power up switching IC's connected to the same ground fill, and the necessity to use a power supply to power something up, will affect the sound quality. I'll leave to others to decide if this is for better of for worse.

[Miska]

50 minutes ago, Extreme_Boky said:

 How can it be a problem if the USB card is powered up at all times, internally, by a clean source?

 

At the moment it is problem for example with Spring 3 when the DAC is in standby mode, USB interface disappears. This wasn't the case on Spring 1 and Spring 2.

 

50 minutes ago, Extreme_Boky said:

The laptop would be  a terrible source of +5V DC VBUS even when running of a battery, due to its switching mode power supply used internally, to generate +5V from a higher voltage rail. 

 

It works just fine with correct data transfer. And doesn't affect the DAC output if the galvanic isolation is implemented the way I presented. Just like it works correctly with attached HDD's and such.

 

50 minutes ago, Extreme_Boky said:

Nonsense. For this, you'd need two separate power feeds from the MV to LV transformer station, straight to your household, with independently grounded neutrals at each point (at the transformer station). You can not have two independent power domains within a household.  

 

No you don't when the source is for example battery powered.

 

[Miska]

12 minutes ago, Extreme_Boky said:

a fact that the current will be sent down the USB cable. You do not have any control over the type/quality/length of that USB cable. You simply must have that extra wire, as part of the USB cable, to send that current to power up the USB card.

a fact that the current runs in parallel to USB data lines, unless the USB cable separates this wire from the shielded twisted pair, by design.

a fact that USB connectors need to rely on two points of contact resistance, to transfer that current from your product, to the actual USB card. Again, you do not have control over this neither (gold plated, rhodium plated, the actual surface area of the contact at either end)

 

Not an issue as long as there's for example optical isolation between the USB interface and the DAC.

 

When you are worried about such, use ethernet instead of USB. Problem solved. You can use even optical ethernet. Or WiFi for air gap, but then you need to be a bit careful about your RF design.

 

12 minutes ago, Extreme_Boky said:

With a USB card powered up internally, by the DAC, whilst taking the necessary (simple) measures of ensuring galvanic isolation, all of the above is not in play nay more. You are in full control. 

 

If you have proper galvanic isolation, it doesn't matter how the USB side is powered.

 

12 minutes ago, Extreme_Boky said:

The extra benefits you mentioned? Every additional switching, clocking or other processing by means of having to use a power supply line(s)/rail(s) to power up switching IC's connected to the same ground fill, and the necessity to use a power supply to power something up, will affect the sound quality. I'll leave to others to decide if this is for better of for worse.

 

So the minimal approach is just to power the USB interface from USB power as things have been designed to be done. Less IC's, less PSUs, etc. Much simpler.

 

[Extreme_Boky]

1 minute ago, Miska said:

 

No you don't when the source is for example battery powered.

 

So, which one is it????

 

You mentioned initially how you have segregated the analog units and digital units by means of two separate power points. (I explained to you that this does not fly... and does not improve anything by a quantitative margin. Your understanding of "two totally independent power domains" is flawed and incorrect

 

Or, you are using laptops that are battery powered??

 

If if you are using the laptops, I explained to you that the laptop uses switching DC-to-DC converters to generate +5V DC for a VBUS, which is inferior to what I am proposing. You should stop doing this if you appreciate good sound.  In addition, this approach of generating that +5V DC is almost always common for at least 2 USB sockets - vastly inferior to what I am proposing.

 

 

[Extreme_Boky]

4 minutes ago, Miska said:

 

Not an issue as long as there's for example optical isolation between the USB interface and the DAC.

 

Hahah.... but eventfully, you will need to power up that USB card by a piece of copper run. I.e. you will need to eventfully use that USB cable and feed the VBUS from somewhere. Your suggestion is just adding more complexity. 

 

4 minutes ago, Miska said:

 

When you are worried about such, use ethernet instead of USB. Problem solved. You can use even optical ethernet. Or WiFi for air gap, but then you need to be a bit careful about your RF design.

 

We are not talking about Ethernet. May does not have an Ethernet card. 

 

4 minutes ago, Miska said:

 

 

If you have proper galvanic isolation, it doesn't matter how the USB side is powered.

 

It makes all the difference in this world how the USB card is powered. You saying this means that you do not have any actual experience with designing digital circuitry. The original USB card uses 7 separate voltage regulators for each section, of each IC. Why would they do that???  According to you, they should just feed one and only +5V DC rail to all IC's...???

 

The +5V bus is used to power a plethora of IC's located on that USB card, as well as the crystal oscillator. They (Vcc/Vee rails) all need to be (for each IC, actually for each section of each IC) properly decupled from each other by means of:

- separate voltage regulators for each section of each IC. May did an excellent job here with their USB card - the best in industry.

- proper local decupling of voltage rails, as close to the Vcc/Vee pins as possible (due to space constrains and an increase of the required number of IC's with  the new USB card... this implementation suffered.... a fair bit) 

 

4 minutes ago, Miska said:

 

 

So the minimal approach is just to power the USB interface from USB power as things have been designed to be done. Less IC's, less PSUs, etc. Much simpler.

 

 

I am not sure what you are trying to say here...??

[Miska]

19 minutes ago, Extreme_Boky said:

So, which one is it????

 

You mentioned initially how you have segregated the analog units and digital units by means of two separate power points. (I explained to you that this does not fly... and does not improve anything by a quantitative margin. Your understanding of "two totally independent power domains" is flawed and incorrect

 

Yes, I have, most critical computing stuff is running off an offline UPS. Rest are on separate power line filters.

 

But as I said, source can be also for example battery powered such as laptop. I can also power mains powered devices from a 12V 120VA lead battery and 230V inverter when I want to.

 

Devices that have two pin mains connections are in addition floating and don't have common ground connection.

 

19 minutes ago, Extreme_Boky said:

Or, you are using laptops that are battery powered??

 

Lot of times yes. Or a NAA that is battery powered. Simplest for a NAA that has 12V DC source.

 

I have about 30 DACs and about 10 music servers and about 10 NAAs. Scattered over two physically distinct locations (office and home office).

 

19 minutes ago, Extreme_Boky said:

If if you are using the laptops, I explained to you that the laptop uses switching DC-to-DC converters to generate +5V DC for a VBUS, which is inferior to what I am proposing.

 

It is well good enough for correct data transfer when the USB interface is galvanically isolated from the DAC itself.

 

19 minutes ago, Extreme_Boky said:

You should stop doing this if you appreciate good sound.  In addition, this approach of generating that +5V DC is almost always common for at least 2 USB sockets - vastly inferior to what I am proposing.

 

Many of NAA's run from 5V DC feeds that are directly fed to the USB interface. For example My Spring 3 is fed by a NAA that is powered by Ferrum Hypsos PSU.

 

Does it matter is another question.

 

[Miska]

16 minutes ago, Extreme_Boky said:

Hahah.... but eventfully, you will need to power up that USB card by a piece of copper run. I.e. you will need to eventfully use that USB cable and feed the VBUS from somewhere. Your suggestion is just adding more complexity. 

 

USB VBUS is property of any USB compliant host device. It will be there always.

 

16 minutes ago, Extreme_Boky said:

We are not talking about Ethernet. May does not have an Ethernet card. 

 

But it could!

 

16 minutes ago, Extreme_Boky said:

It makes all the difference in this world how the USB card is powered.

 

Why?

 

16 minutes ago, Extreme_Boky said:

You saying this means that you do not have any actual experience with designing digital circuitry.

 

Oh I do....

 

16 minutes ago, Extreme_Boky said:

The original USB card uses 7 separate voltage regulators for each section, of each IC. Why would they do that???  According to you, they should just feed one and only +5V DC rail to all IC's...???

 

You cannot feed +5V DC to all the IC's because most of them use 3.3V DC.

 

16 minutes ago, Extreme_Boky said:

The +5V bus is used to power a plethora of IC's located on that USB card, as well as the crystal oscillator.

 

If it is correctly designed, there's crystal oscillator only for the USB bus. Some really cheap oscillator is just fine for the purpose. Like the stuff you can find on external HDD's.

 

You need three oscillators, one for the USB bus, and two for audio. The USB bus one is not critical at all.

 

16 minutes ago, Extreme_Boky said:

They (Vcc/Vee rails) all need to be (for each IC, actually for each section of each IC) properly decupled from each other by means of:

- separate voltage regulators for each section of each IC. May did an excellent job here with their USB card - the best in industry.

 

So far actually none of the XMOS interfaces are great. Luckily most have moved away from using XMOS.

 

16 minutes ago, Extreme_Boky said:

- proper local decupling of voltage rails, as close to the Vcc/Vee pins as possible (due to space constrains and an increase of the required number of IC's with  the new USB card... this implementation suffered.... a fair bit) 

 

OK, I'd like to see some objective measurement data that shows the "suffered a fair bit" part.

 

16 minutes ago, Extreme_Boky said:

I am not sure what you are trying to say here...??

 

That galvanic isolation matters. Data interface side matters only to the extent that data is transferred correctly. Rest is job of the DAC beyond the isolation barrier. And the audio clocks that are related to jitter in DAC output are at the DAC side of that barrier. If they are on the USB interface side, they are at the wrong side and you will certainly have lot of challenges making it perform.

 

[Superdad]

13 minutes ago, Miska said:

You need three oscillators, one for the USB bus, and two for audio. The USB bus one is not critical at all.

 

There are dozens of well-known audio manufacturers and thousands of audiophiles who will disagree with that! 

[Miska]

1 hour ago, Superdad said:

There are dozens of well-known audio manufacturers and thousands of audiophiles who will disagree with that! 

 

Still very little DAC analog output measurement data showing the difference. Feel free to post some!

 

[musicjunkie917]

2 hours ago, Miska said:

 

Still very little DAC analog output measurement data showing the difference. Feel free to post some!

 

 

Touche´

[barrows]

7 hours ago, Miska said:

Still very little DAC analog output measurement data showing the difference. Feel free to post some!

Yeah, I find this maddening.  Having personally upgraded the USB clock, and heard a significant (IOW, larger than anything imaginary) improvement, it is annoying that I cannot see any difference in measures.  Additionally, I find it maddening that I have not heard a single speculation of why/how the USB clock might influence DAC sound quality which makes any sense.  I do not believe in "magik" in audio, but it still seems we do not measure the exact right things all of the time.

I know Jussi that you reference many measurements which well out of the range of audibility as being "improvements" (of course technically they are, even if inaudible) so what is the point of those measures...  It seems there is more going on than is fully confirmed/understood and I wish we could find the definitive measures to define all sound quality characteristics-it is just audio after all, not quantum physics.

[Superdad]

1 hour ago, barrows said:

Additionally, I find it maddening that I have not heard a single speculation of why/how the USB clock might influence DAC sound quality which makes any sense.

 

So I guess you did not read the paper we put out, wherein John Swenson, an engineer who actually designed the power networks of ASIC chips for decades--including both USB and Ethernet PHYs--explains how the phase-noise of clocks in packet-data interfaces (USB or Ethernet) can cause clock-threshold-jitter, which in turn induces ground-plane noise in downstream devices and chips.

https://cdn.shopify.com/s/files/1/0660/6121/files/UpTone-J.Swenson_EtherREGEN_white_paper.pdf?v=1583429386

 

[Actually Barrows, I know you did read the paper. So it baffles me that you frequently claim not to have seen/heard any plausible explanation. And after all, it comes from the very person who engineered the Sonore Rendu series products you associate yourself with. ]

 

[barrows]

2 minutes ago, Superdad said:

 

So I guess you did not read the paper we put out, wherein John Swenson, an engineer who actually designed the power networks of ASIC chips for decades--including both USB and Ethernet PHYs--explains how the phase-noise of clocks in packet-data interfaces (USB or Ethernet) can cause clock-threshold-jitter, which in turn induces ground-plane noise in downstream devices and chips.

https://cdn.shopify.com/s/files/1/0660/6121/files/UpTone-J.Swenson_EtherREGEN_white_paper.pdf?v=1583429386

 

[Actually Barrows, I know you did read the paper. So it baffles me that you frequently claim not to have seen/heard any plausible explanation. And after all, it comes from the very person who engineered the Sonore Rendu series products you associate yourself with. ]

 

Yes, I read it a couple of times...  Either I am just too dense, or perhaps JS' way with words eludes me (this would not be the first time for an engineer), as I was unable to really, explicitly, follow what he was trying to get across.

[Superdad]

41 minutes ago, barrows said:

…I was unable to really, explicitly, follow what he was trying to get across.

Well I thought I gave you the whole paper in a nutshell:

45 minutes ago, Superdad said:

…the phase-noise of clocks in packet-data interfaces (USB or Ethernet) can cause clock-threshold-jitter, which in turn induces ground-plane noise in downstream devices and chips.

Most of the rest of the paper was just our attempt to explain phase-noise to lay-folk. 
 

Yes, this stuff is not particularly easy to measure (though John has done some ground-plane noise and phase-noise measurements at DAC master clock pins—using differential probes and his Jackson Labs PhaseStation respectively).

But the effects of this are certainly not hard to hear!

In fact just today I finally had a chance (after modding one of my JS-2s to the required 15V) to hook up to my EtherREGEN a Cybershaft 10MHz reference clock. Let’s just say I am late to the grinning party that so many of our other clients are already at. And remember, we already use in the EtherREGEN the very nice Crystek CCHD-575–same as got adopted for the Rendu series beginning with the ultraRendu. 
 

Have a great weekend. Do you still have some nice snow out there in CO?

[Extreme_Boky]

Close to 10  years ago, I moved from SD to SDA clocks and never looked back. 

 

May has chosen the lowest phase noise oscillators currently available (CVHD-957) sans oven temperature controlled ones. But... the data sheets conveniently are showing the phase noise reading only down to 10Hz... not even 1Hz, let alone 0.1Hz. These are supposedly the best in business oscillators; why stop at 10Hz only is beyond any comprehension. 

 

Not to mention the complete omission of phase noise vs. temperature graphs, for at least few frequency curves (including 10, 1 and 0.1Hz).

 

At the same time, NDK data sheet from 2016 clearly shows phase noise specs at 1Hz, SD vs SDA

 

CVHD-957.pdfNDK SDA's.pdf

[Miska]

2 hours ago, barrows said:

I know Jussi that you reference many measurements which well out of the range of audibility as being "improvements" (of course technically they are, even if inaudible) so what is the point of those measures...  It seems there is more going on than is fully confirmed/understood and I wish we could find the definitive measures to define all sound quality characteristics-it is just audio after all, not quantum physics.

 

Roughly my point is that if something can be measured it can be also audible.

 

For example ultrasonic and RF output typically effects later stages in audio chain. If you have a class-D amp (Hypex for example), it is known that it will have aliasing effects on such inputs. So you should pay special attention on that area when choosing your source pairing.

 

I also know that audio band things down to at least -120 dB from source are audible depending on cases.

 

Then there's a lot of audio foolery around band-limited measurements. For example AP plots restricted to 20 kHz band where in fact the anti-alias filter of the analyzer is acting as reconstruction filter for the DAC. So those are not measuring reconstruction accuracy of the DAC, but instead how the DAC performs when combined with such extreme brick-wall reconstruction filter that you won't have in the signal chain in real playback situation.

 

This is the reason why I measure DAC outputs up to 100 MHz (or more) bandwidth, to get more accurate overall picture what is actually coming out.

 

Then another aspect is that lot of the things are in existing measurement results, but people just overlook those. And some things are just not commonly measured.

 

2 hours ago, barrows said:

Having personally upgraded the USB clock, and heard a significant (IOW, larger than anything imaginary) improvement

 

Did you measure the system before/after? And what is the DAC in question?

 

[Miska]

1 hour ago, Superdad said:

So I guess you did not read the paper we put out, wherein John Swenson, an engineer who actually designed the power networks of ASIC chips for decades--including both USB and Ethernet PHYs--explains how the phase-noise of clocks in packet-data interfaces (USB or Ethernet) can cause clock-threshold-jitter, which in turn induces ground-plane noise in downstream devices and chips.

 

That still doesn't have measurement data from DAC analog outputs...

 

[barrows]

3 minutes ago, Extreme_Boky said:

15 years ago I moved from SD to SDA clocks and never looked back. 

 

May has chosen, apparently, the lowest phase noise oscillators currently available (CVHD-957). But... the data sheets conveniently are showing the phase noise reading only down to 10Hz... not even 1Hz, let alone 0.1Hz. These are supposedly the best in business oscillators; why stop at 10Hz only is beyond any comprehension. 

 

Not to mention the complete omission of phase noise vs. temperature graphs, for at least few frequency curves (including 10, 1 and 0.1Hz).

 

At the same time, NDK data sheet from 2016 clearly shows phase noise specs at 1Hz, SD vs SDA

 

CVHD-957.pdf 595.43 kB · 0 downloads NDK SDA's.pdf 146.87 kB · 1 download

Indeed, the SDA series are the best "standard" options.  It is important to realize that the spec sheets do not tell nearly the entire story at all though, as there are quite a bit of variance in phase noise between different samples of the same parts, and some manufacturers may be publishing phase noise data from selected samples.  I have it on very good authority, from an engineer who actually measures the phase noise of many samples, that the SDA series NDKs consistently outperform the best from Crystek.  But the CVHD part in the May is not the best Crystek has to offer, the CVHD series is a Voltage controlled oscillator and voltage controlled oscillators, all other things being equal, perform worse than fixed oscillators.  My guess is that the May uses a VCXO in order to make their PLL work.

Of course there are better clocks out there...  A really good oscillator made with a really good SC cut crystal will outperform all of the above.  Ayre, for example, uses a custom clock from Morion-basically a non-ovenized version of one of their (normally) OCXOs with an SC cut crystal.

Another approach worth experimenting with is using a lower rate XO than needed, taking advantage of the lower phase noise inherent in lower speed clocks, and then using a DDS to derive the needed frequency.  The recent advances in DDS devices make this a really interesting approach these days.

[Miska]

26 minutes ago, Extreme_Boky said:

May has chosen the lowest phase noise oscillators currently available (CVHD-957) sans oven temperature controlled ones. But... the data sheets conveniently are showing the phase noise reading only down to 10Hz... not even 1Hz, let alone 0.1Hz. These are supposedly the best in business oscillators; why stop at 10Hz only is beyond any comprehension. 

 

You can always look at the DAC's output. Because effect of the clock also depends on your D/A conversion architecture and what kind of data you feed it.

 

For example with Spring 3, you have 6 dB relative difference on noise floor between PCM and DSD, because of 6 dB output level difference re 0 dBFS (OTOH, DSD source can go momentarily to +3.5 dBFS which PCM cannot).

 

But you also have difference on jitter performance:

 

But overall this is very nice performance and the main lobe is very narrow at the base. And source is USB port of my HP Z4 G4 Xeon W-2245 workstation (64 GB of ECC RAM and Nvidia GeForce RTX2080 Founders Edition GPU).

 

If the DAC would be something like Chord Mojo, it would be much much more sensitive to the source.