Master Clock for your EtherREGEN

[feelingears]

17 minutes ago, Clockmeister said:

My personal dac collection number over 100 now including some highly desirable audiophile delights plus some real howlers. Yet I use dac's between 28 and 16 years old simply because the make music correctly and place you firmly inside the recording environment. A lot of the new high-end bling infested are just merely polish hifi with zero grain and very little involvement. However, audio is very personal and what one man really enjoys the other is not so fussed.

 

Clockmeister,

 

Thanks for the extra detail and color commentary! I appreciate learning the perspective and context from which you're coming from because absolutely, one person's peach is another's, um, ploop. (Not that it matters, but I just read impressions of some rooms at the Pacific Audio Show here near Seattle. Wow, the reviewer's "most beautiful, lushest, hypnotic" was my worst or second worst-sounding room...and then there were a couple rooms we did agree on. Crazy hobby, this.)

 

DACs...thanks for that tidbit, and I have felt a tug in your direction (the past). I have been finding that there's a careful balance between "past" technological approaches versus/combined with modern noise "aware" designs and implementations.

 

Please don't let your digital testing fetish become too confident in its hypothesizing–words can't convey some of the things I hear (and love), and perhaps that's partly why this hobby is so wacky. Maybe you will help coin new expressions or words to bridge these sonic chasms, ha ha.

 

[Jakenz]

9 hours ago, Clockmeister said:

 

 

Hello FG

 

 

 

Totally understand your thoughts perfectly

 

 

 

I will clarify the 5-7% for you as it will put this into perspective in relation to my personal evaluation system is somewhat different from a great many system's its has been designed to show ANY differenices across a wide spectrum of variables.

 

 

 

To obtain that 5-7% from this single aspect of ethernet noise in raw terms was in a $ value around 12K, now we have clients that would bat an eyelid at this outlay. For others it's a system cost.

 

 

 

So the sound quality point of that system is significant to raise the bar that much more is a good figure of merit, HOWEVER it is an accumulative effect in that you will require, mains conditioning/grounding/room treatment/quality LPS's and cables.

 

 

 

I did take the spare N20 into one of the smaller system which are well set up, although only have moderate mains conditioning and good cables in a well set up room, the resultant difference wasn't nearly as marked. With the RJ45 grounding cable attached it produced a better result. Very system dependant imho.

 

 

 

One does have to keep a perspective on all of these steps forward, how and why each one happens and what results are transposed into the finial SQ.

 

 

 

Here is  a bit of fun for you, for my sins I am a self-confessed digital whore & test equipment Fetishest.

 

 

 

My personal dac collection number over 100 now including some highly desirable audiophile delights plus some real howlers. Yet I use dac's between 28 and 16 years old simply because the make music correctly and place you firmly inside the recording environment. A lot of the new high-end bling infested are just merely polish hifi with zero grain and very little involvement. However, audio is very personal and what one man really enjoys the other is not so fussed.

 

 

 

Here I was investigating finding some spurious RF emissions on an oldie b ut goodie dac, the worst starting point when using the dedicated app for 'oddball' emissions was -68dBc a solid 15dBc below the Cisper 14 standards. However, a few investigative hours with the right equipment can yield some very positive results.

 

 

 

However you need to apply this philosophy right across all of the audio system imho

 

 

 

 

Hi @Clockmeister I've followed your posts here and on the TAS forum with interest.

 

By contrast my only tool to measure the performance of my gear is my ears, trial and error experience with a much smaller range of gear than you have to hand. Still, I have enough to notice a few repeatable patterns / correlations.

 

So one repeatable phenomenon I've observed recently I'd be interested in knowing if you've ever tried to measure, is the audibly beneficial effect of signal grounding and vibration damping on external clocks and DACs. Addition of a Quartz Acoustic ground box to a spare BNC or s/pdif RCA of the Leo Bodnar, the LHY OCK-2 and my R26 DAC lifts their performance markedly, a more dynamic, spacious yet less digital sound. Vocals softer, more full bodied and papable. Ditto re adding A5 sheets of fo.Q's potent 2mm thick damping sheets (SH-22K/21K) to the chassis of each of those three as well as my LHY SW-8 switch. In the case of the petite LB I have it sitting on a sheet. This signal grounding benefit is on top of my baseline of a Puritan PSM156 power conditioner and their Ground master with dedicated ground rod.

 

The effect of the signal grounding and vibration damping is additive and sufficiently pronounced (my system sounds much better with it) it wouldn't surprise me if was measurable /observable in a cleaner square wave and/or lower phase noise at lower offsets. I'd be interested in your observations here.

 

To acknowledge the thread topic I'm not yet an EtherRegen owner... Waiting for the ER 2 release, but would suggest ER owners might wish to try fo.Q damping on their ERs given how potent an effect it has on my LHY switch.

 

Indeed @Dandou is well placed to try this should he so wish...

[dylanesque]

In the early days with my Aqvox SE switch, I had such an improved effect of chassis and signal grounding of unused ports, using the Puritan GM. That I added Puritan route masters and connected all on my digital chain for good order. But did not see much cumulative improvement than just over the Aqvox switch. Did not do reduction and isolated testing. 
 

(For good order, then added 2nd Ground and route masters for the analogue chain, for signal and chassis grounding) 

 

My power grounding is untampered. 
This was long time ago.

So I’m also keen to understand what the measurements say re: Grounding. 

 

Since then, more  has happened. More ground shattering performance improvements. 

 

Meraki MS220 8P unmodded + Aqvox SE + ER + 120SE (separate power ground) reclocks ER (separate power ground) feeding OR and then tx-USB. And Lampizator Big7.

 

Each stage pulls its weight. With 120SE being the crowning glory. 

 

Alternate route. 
Meraki MS220 8P unmodded + Aqvox SE + ER + 120SE (separate power ground) reclocks ER (separate power ground) feeding 2x Stacked MC3 with ADark full mods, and using SPIDF out to Abbas DAC 3.2SE. Using Canare LV-77S clock cables. 
 

I prefer this 2nd configuration. Because the stacking the MC3 in RCLK mode with 120SE paired with 16Bit TDA1541 is simply magical.

 

By default, any thing that enters my chain is chassis grounded by default. 
Signal grounding of unused ports. And/Or RF protection caps. 

 

Biggest BFB on Digital signal
- 120SE clocking ER
- Stacking MC3 2x with ADark mods
- Fidelity Audio HF Filters
- Daisy Aqvox + ER
- txUSB ultra re-clocked by 120SE
- OR feeding txUSB
- Grounding Chassis and signal on all digital units. 

 

tbd: If ER2 should replace ER or get it in series..
Because, each stage has given me a step improvement.

 

Some day if I have to trim down, I would be happy with 120SE + ER2 + FA filters. +Grounding. 

 

Haven’t yet ventured into Clock cable optimization yet. Using Canare LV-77S clock cables. Taken notes from all you  wonderful folks on this gargantuan thread over the years.  But for now, increased/ing attention to family matters. 

[Jakenz]

1 hour ago, dylanesque said:

In the early days with my Aqvox SE switch, I had such an improved effect of chassis and signal grounding of unused ports, using the Puritan GM. That I added Puritan route masters and connected all on my digital chain for good order. But did not see much cumulative improvement than just over the Aqvox switch. Did not do reduction and isolated testing. 
 

(For good order, then added 2nd Ground and route masters for the analogue chain, for signal and chassis grounding) 

 

I like that we both have Puritan + GM setup, albeit with systems at rather different levels. At the risk of going a little off topic, how did you initially signal ground with a Puritan PSM + GM prior to getting the Route Master?

 

Back slightly more on topic, a cheap / easy and very effective clock cable tweak I and a number of others over at Headfi have recently tried with satisfying results is this: using copper foil tape with conductive adhesive to wrap the BNC male/female junctions at both ends of the clock cable. Funnily enough I was indirectly prompted to try this by @JohnSwenson 's emphasis on the quality of shielding being key for sine, which I predominantly use, sounds best in my system.

 

I had also read separately in a RF radio context that very high frequency RFI/EMI finds easy ingress / egress (particularly the latter) through even the smallest pin holes or discontinuities in shielding, easily reducing shielding effectiveness by tens of dB. In particular RF pigtails* with their additional junctions are not great, shielding has been shown to drop by tens or potentially scores of dB. So when I got the Minicircuits HP filter recommended by John I found myself looking at a series of BNC junctions. Which got me thinking about the mechanical and potentially conductive/shielding imperfection of the often loose BNC junction vs say an SMA screw type connection. So I wrapped the whole filter (the two BNC junctions) at the DAC end in copper foil and it sounded clearly better. I did the same at the clock end for the cable/clock junction - a similar additive improvement. Tried it with a range of clock cables (I have about 5, tried it on the best 3), they were all satisfyingly improved.

 

The character of the audible improvement is similar to reducing jitter by other means I've experienced (signal grounding, better clock cables etc) - a more full bodied, spacious and dynamic sound with less edginess and glare. I can't listen without it now, my system sounds a little shrill, thin and flat without it. 

 

This is another easily audible change I'd be interested in knowing whether it shows up in measurements of clocks & DACs.

 

*I'm aware many DACs and other clocked devices have - of necessity - internal pigtails from the clock board to the BNC female jack on the back of the DAC. My R26 has this with an MCX to BNC pigtail. Following the lead of @Stellabagpuss and @rodthebod I replaced the pigtail with an MCX/BNC adaptor (no cable in between) to which I directly attached my Harmonic Tech clock cable, including foil wrapping the entire BNC junction, now sitting just above the circuit board. Sure enough this brought further improvements of the character I describe above. However it was quite impractical as put too much lateral stress on the tiny MCX socket and couldn't fit through the back panel without modification. Or even close the lid fully. I'm not in a hurry to do this, so just a proof of concept/validation of a theory for now. Perhaps I'll replace the pigtail with something better as I understand @MartinT is doing.

 

Sorry a bit of a spiel there, but hopefully not too off topic and of some interest, especially to the folks on this thread with the ability to see if this shows up in measurements.

[Clockmeister]

Grounding is a interesting subject one which I really could deleve into great depth, so many audio manufacturers white papers I've seen on this at quiet frankly they are very wide of the mark.

This is RF at work, primarily conducted emissions, however there are other reference rail issue to overcome as well.

 

The whole point of the exercise is to reduce the PD (potential difference) between each of the 0Vdc (reference rails) between the various connected equipment to a low a resistance as possible. Taking into account the whole transmission line itself and the following impedances. Absolute ideal is a single connection point rather than multiple boxes which will give you variously different PD's. I'm not saying they don't work, what I suggesting is that it's not as optimal as it could be.

 

Over the years I have collected pretty much all of the latest and greatest grounding boxes, measured them in NON-Amir way (lol) in working systems with a calibrated set up to specifically designed to record and identify this very issue.

 

This investigation came about on a non-audio project we were involved with in the telecommunications sector nearly ten years ago.

 

What I can say is that we have measured over 150uA of RF current travelling through the various signal transfer pathways on audio systems.

 

@MartinT will be able to clue you in more, we demonstrated this to him on one of his previous dac's, one of the big areas of noise was the 100Mhz Crystek 575 wonder clock, the amount of harmonic noise that this circuit emitted was ferocious with 3rd order harmonics stretching right back to the 30th, quite something for dac having 3Ghz of bandwidth floating around in there.

 

Here are before and after the remedial work was carried out, there was other ac filter considerations worked on as well, however the ref level was @ -10dBc no pre amplifier on and this would have never passed ANY CE/UKCA/UL (which is more lax than the other two) ideally keeping below -55dBc is a good start normal standards start @ 9Khz and end at 30 or 100Mhz for consumer products. We replaced that Crystek with a very good NHK txco, spec's could compete with the 575, yet strangely it sound way more enjoyable and involving 😁

 

 Getting on to clocks & real-world performance, what really makes a clock audio word clock, very low Allen deviation? ultra-low phase noise, AN other.

 

Ever wondered why the Leo Bodnar clock is quite a way superior to many so call high end clocks?

I'll let you figure that one out.

 

With vibration within components, yes this is very common place due to inadequate design thought gone into board layout & stacking, piss flimsy case work like a lot of $6 extruded cheap assed aluminium lengths with a tacking slot in or bolt on end plates, the height of quality workmanship.

 

Or this little nugget makes me laugh, in fact we have one or two in the office as a rogues galley a txco or oxco clock crystal on an evolution board (Digikey or Mouser lol) slapped in a cheap case with the most basic of power supplies, coupled with a non-student friendly price tag 😆

 

Vibration both internal & external can be easly identified with a seismometer, every time you pass a current/voltage through an inductor/capacitor/resistor you will have microphony generated, but how many components are installed on that circuit board sometimes multiple hundreds, plus you will also incur a very small phase angle deviation as well.

 

It's like your system is conspiring to get you before you press or swipe play.

 

With the dual in series MC3+'s (fully upgraded internally with our own custom updates) on ultra-low noise LPS's and Ref 10 SE-120 (internal rebuilt with over 88 changes), I personally use at least twelve (yes 12) ground lead just for those only.

It is accumulative progression.

 

Some dac's really love grounding far more than others and some of these are the expensive one's Solution 760's thrive on grounding its not subtle in any way. So does DCS (lol)

 

You can trace this back to how these dac's were first concieved its awfully insightful.

 

Have a good day chaps, It’s a sultry 27C here and I have a couple of LPS to finish plus a D90 QFN regulator repair and MC3 full upgrade to undertake.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[ZeusOdin]

@Clockmeister, you should write a book containing your experiences and insights with clocks, clock cables, grounding, DAC, and the recent foray into ethernet. All audiophiles would buy it. Manufacturers--all manufacturers--would hate you however. Lol

[Clockmeister]

@ZeusOdin 

 

 

 I would be a very look book with a lot of dry subjects. How to boil and condense down over 100Tb of relevant information into a say a 12 chapter bible call 'The Noise abatement society strikes back'

 

here in the UK the term audiophile is a turn off, like plane & train spotters, all that missing are the long trench coat type mac's large ruck or haver sack with copious amounts of 'reading material' and a anal retentive score of mine goes to 12 😁

 

We tend to deal with music lovers rather than audiophiles, who on the whole feel the know all that is warranted by reading Paul 'Ego Master' Millers publications Hifi News/Stereophile or Absolute Sounds / Hifi + cover top cover for 10 years and attend every show. We have a very apt term for these chaps, 'Biker-less bikers' they have the full leathers, top skid lid, latest carbon fibre boots and ride a push bike hoping to cadge a lift on the Friday blast to the pub 🤪.

 

Music lovers seek to have the sounds their way rather than eeking out WBF's thread run for best can of platinum encrusted tire weld for their latest automotive penile extension acquisition or how many Taiko extremes can I have on my ground floor complex of sound systems 🙃

 

Although even these guys are bulking @ the Wadax's 🤑

 

 

Recently I was 'fortunate' to have a 90 minute conversation with a very intelligent chap who's back ground is integrate circuit substrates and was ultra-insistence that si IGBT were far superior to SiC Mosfets in regard to power supplies and amplifier outputs. Now in certain circumstances I would agree, but for that specific application not at the moment. Strangely no mention of GaN but hey what do any of us know

 

At the end of the conversation, I did enquire as to if he had built such a amplifier with these devices yet?

 

'Oh no' it models so well I don't need to. I believe he has even built a digital twin for this idea??

 

The case for the persecution rest's M'Lud . All joking aside can I suggest this publication by Mark Waldrep ‘Music and Audio: A User Guide To Better Sound’

 

https://www.musicandaudioguide.com/thebook/

 

There are some ground rules in there that will help a lot of you, doesn’t cover everything but it’s a very good start.

 

I know it took Mark nearly three years to finish this publication, I really have more interesting projects that require my attention, however I appreciate the thought

 

[MartinT]

I can confirm @Clockmeister account as that DAC was my old LKS, a DAC that uses the Crystek 575 and which sounded tremendously better after the 575 was swapped out for a Coherent clock. It then became a very decent DAC until the Gustard X26 Pro finally usurped it.

[Clockmeister]

 Afternoon chaps, thought I would share with you some images of a finished project recently for a gentleman in Ireland, this I feel is as close as it realistically is achievable with a low frequency extracted frame sync (FS) now usually in HSSD   (High speed serial data) the eye pattern is far more closed up can be as much as  30+dB due to frequency/noise/transmission line losses and ISI (inter symbol interference) In some cases software coding is used to open up the eye at the receive end.

 

Anyway here is updated Mutec MC3+ after completion showing both time & frequency measurements for waveform with some critical parameters that are used for checking signal integrity of wave forms, also the lower screen image of a zoom of the dwell period between the two transitions of this pulse wave/train (rising and falling edge). As you can see its rather flat and consistent and stable (the peak <> Peak measurement table also shows standard deviation for this parameter as well)

 

Also, the eye pattern itself is rather 'open' lol this is rather good, but remember it’s a lower frequency and a small transmission line with not much in the way of circuitry imparting many signal deviation issues.

 

I always use a genuine Lecroy 75Ohm adapter when measuring audio clock cables rto keep impedances corrent

 

The main R&D scope has a jitter of 60fS (on its own with using the Mutec ref10 SE-120 however all of my instruments are sync'd using a enhanced version), a full 12 bits of resolution and sample rates up to 20 bilion samples a second with an extended memory depth of 5Gpts.

 

Finally, the FFT plot for 1dbfs @ 1Khz using 256K with 16 averaging. as you can see zero spurs in the mud at all, it’s at the limit of the audio analysers range.

 

This was using a quality lab bench linear psu as the temporary power source, the fitted one was two orders of magnitude quieter and measured directly at the DuT’s point of use.

 

Fear not I'm working on that as we speak 🤣

 

What does this all tell us?

 

Well the more open the eye, the better the signal integrity and therefore the more closer to the ideal wave form for the purpose at hand.

 

Though you chaps would appreciate some insight

 

 

 

 

 

 

 

 

 

 

[Clockmeister]

 

 

 

You missed the point FG, I was extracting the urine from said gent.

 

I could demonstrate so many so called ultra-well measuring audio devices that sound as dull as ditch water or so polished it would seem like all songs would song like a 'Hello and welcome to BBC' in a 1950's RP dialect programme announcer Or conversely sound like a 'nice wall of sound'

 

However with clocks its not the case, the least amount temporal deviation you have simply the more coherent, three dimensionally textured and real the playback sounds.

 

When I come up with any designs, I have produce at least prototype boards and build and test every one, and yes I will make changes before finally settling on the finished article.

 

Just to put matter into to further perspective I am a musician as well (well the band tell I just hang around with them, but as I point out if I didn't they timing would utterly suck 🤣) We alos have two recording studio's within the group so oddly enough I really do understand how bands/orchestra's and performers sound in the hall/pub/stadium and that local club.

 

I use a mix of scientific/empirical and listening tests to truly get to grips with any product plus a group of eight beta testers with totally different systems.

 

After all a totally three 3D top to bottom linearity, textured, fuild, articulate, natural sound with realistic dynamics thats sound good at low volumes is your goal I would suggest

 

Not some hyper in yer yer face 28 foot Clapped out in front of you you can drive a double decker bus through the image. Thats just crass bollox not music.

 

Although as we have mentioned before, some folks like that sort of sounds, and some like real music you can pour into a glass with zero effort and just be captivated for hours on end.

 

 

 

 

 

 

[dylanesque]

On 7/8/2023 at 2:41 PM, Jakenz said:

I like that we both have Puritan + GM setup, albeit with systems at rather different levels. At the risk of going a little off topic, how did you initially signal ground with a Puritan PSM + GM prior to getting the Route Master?

@JakenzI have not had a PSM.  

Mike was kind enough to make a pigtails for me with the GM, in early days. 

DIY’ed leads and used a rather crude but neat way, using a Home theater panel with banana sockets, off Amazon. 

[GoodEnough]

Apologies for the off-topic question. 

I have a friend asking about BNC coax cabling from CD transport to DAC.  Do the same considerations apply to this choice of cable as apply to a square wave clock?  Namely, prioritising matched impedance and low attenuation?

[Clockmeister]

Indeed they do and if a BNC connector is involved at one and an RCA at the other, please, please do not use an adapter, have the cable terminated accordingly. If possible AES is the best route, however SPDIF can sound fine to.

[Clockmeister]

As I am having a lab re arrangement currently, I thought I would dust off the VNA (Vector Network Analyser) these are used for cable measurements, S11/S12 & S21/S22 return loss (dynamically), antenna setting up (SWR). circuit pathway measurements/ base and broad band emission monitoring and 5/6G & higher communication rf transmission decoding etc

 

VNA explanation

 

So I found an appropriate 'N' terminated 50Ohm T-flex 405 cable, let the instrument warm up for a hour or so, calibrated then looked at the results of investigation.

 

Now on this unit only looking up to 1.5Ghz (not quite DC to daylight lol) anyway these images are of the Smith chart/ phase response / Polar chart and S11 reflection coefficient (gamma) ideally as close to one is the goal. Phase response here looks odd but is actually very good. With the DUT (cable) showing just over 3.2 cycles in 1 GHz, that's equivalent to an electrical length of about 3.2 feet, the actually cable length is 90cm so pretty close I feel. light travels at about 1 ns per foot or in a typical plastic-insulated cable a physical length of about 2.2 ft / .67 m in that time period.

 

[nickv]

If I was to buy a clock now with a view to picking up an etheregen gen2 in the future how important would it be for the clock to have a measurement of phase noise in the order of -140 at 10Mhz?

[Superdad]

5 hours ago, nickv said:

If I was to buy a clock now with a view to picking up an etheregen gen2 in the future how important would it be for the clock to have a measurement of phase noise in the order of -140 at 10Mhz?

A 10MHz clock with -140dBc/Hz phase noise at 10Hz offset would be a fine choice for an EtherREGEN (Gen1 or Gen2). 

[Superdad]

@Clockmeister Hi David:

I shared with @JohnSwenson your comments about the Crystek CCHD-575 (which we use a 25.0MHz instance of in the EtherREGEN as reference for the Skyworks clock synthesizer--though we are moving to two even lower jitter Analog Devices ADF-series synths for EtherREGEN Gen2).

Below were his comments:

 

"On the 575 it not so much that the 575 itself is radiating noise, but the current spikes on the power and ground when edges occur. You can decrease the spectrum of these spikes by slowing down the edges (a series resistor works for that). David is complaining about the broadband noise on the ground plane and power planes contaminating the clock receiver on chips receiving the clock.

If you make the edge slow the broadband noise on the planes goes way down but you get more jitter on the receiver from the slow edge. Of course the way to get around that is to use differential clocks that don't look at the planes at all, but you somehow have to get the single-ended clock converted into differential which can embed the noise on the planes into jitter on the differential signal. So this comes down to converting the single-ended clock to differential as close to the clock as is physically possible to cut down on the area that clock current is running through the board. Careful layout can drastically reduce such effects on the board.

Of course very few DAC designers know about such things so this issue is rarely optimal in DAC designs. As you know Alex, I HAVE very carefully worked out such things in the EtherREGEN to try and minimize these effects. And in EtherREGEN Gen2 the single-ended clock is only going to be from the 575 or BNC to the LTC sine-to-square chip which has filters that get rid of all that broadband stuff anyway. The clock from there on is fully differential. So these effects of the 575 are almost completely mitigated by the way clocks are being handled in the ER2."

[SQFIRST]

@Superdad thank you for your follow up on the CCHD 575. Please excuse the rudimentary understanding behind my question:

Is the broadband noise issue relevant when using the ER with an external clock as the internal clock is being bypassed? 

[Superdad]

39 minutes ago, SQFIRST said:

@Superdad thank you for your follow up on the CCHD 575. Please excuse the rudimentary understanding behind my question:

Is the broadband noise issue relevant when using the ER with an external clock as the internal clock is being bypassed? 

 

Well the takeaway from what John wrote is that implementation is everything! 

What someone experienced with the Crystek 575--either through technical measurements or via replacement/listening--in some DAC or other device does not apply to the EtherREGEN.

First off, EtherREGEN (both Gen1 and the forthcoming Gen2) have the Crystek sitting about 3mm from the input of our clock synthesizer--and that's about the only single-ended clock run in the unit; from there it is all differential clock lines, with LVDS>SE buffers converting back to SE right at the inputs of the main switch chip and PHYs.

 

Second, here are some other relevant comments I forgot to quote from John:

 

"On the 575, I think he was talking about power supply noise, i.e. the current pulses it draws from the power supply and that noise getting into other parts of the circuit. That is very understandable given it is a square wave output with fairly steep edges that is exactly what will happen. We try to mitigate this by using a low-inductance 0402 bypass cap right at the pins and a separate LT3042 regulator driving just the 575.

This whole thing is one of the constant conundrums of digital audio, fast square wave clock edges give lower jitter, but increased broaddband noise in the circuit in general. Designers are constantly going one way or the other. As to going into a clock synthesizer, that doesn't particularly make any difference. I think this is just an issue with DAC desings that are large, big boards where the broadband noise from a fast edge has lots of room to generate significant noise across the board.

BTW the infamous differential clock can significantly alleviate a lot of this by not referencing anything off the ground plane. I don't know how many DACs are doing this or not. Probably VERY few!"

 

 

But to clarify, yes when one uses an external 10MHz clock with EtherREGEN the Crystek 575 is not used. That's the whole thing with the power-on sequence and the black slide switch telling a tiny microprocessor which code to load to the synthesizer so it knows to run from either the internal 25.0MHz Crystek or a running external 10MHz clock.

[Clockmeister]

 

 Hello Alex

Thanks for sharing the information.

 

We did our own inverstigations into the Crystals a few years ago

 

The issue with this clock is manyfold, its circuit board lay out and stacking/veer placement (this is directly related to reference plane noise like the ES90038) Using LT3042 or similar doesn't bring exactly what you feel it should either. We produce psu's in the pA & fA noise ranges for IoT devices and bearing in mind the most these types of clocks take is around 30mA. We still observed the same results only with minor reduction in harmonic amplitudes throughout the entire measured frequency domain. So, the root cause was elsewhere.

 

Rise time is really the main issue with this particular range of clocks, if we take the 10/90 rule on this. Having a faster rise time was a seen as a desrible trait for audio. However the amount of 'gibbs ears' ringing and overshoot it displayed on the resultant wave forms was initially dismissed as a 'it went with the territory' attitude with a lot of designers in many applications not just audio. Therefore, using a clock with a less rampantly aggressive take off path 🤣 many of the shortcoming of those designs could be mitigated imho.

 

You can with a lot of care and paying particular attention to the circuit layout mitigate some of the that crystals issues.

However instead of using a sticking plaster actually design the product from scratch with alternative methodology.

 

The use of Synthesised PLL via FPGA and dedicated I/C's (TI/Microsemi etc) have been around for quite a while, but now the technology has matured it is becoming as mainstream as spread spectrum clocking.

 

Wadia were generating synthesised clocks some 28 years ago,, now it’s just much smaller packages and less real estate to take up on the board lol

 

As with everything in engineering it how much of a trade-off you wish to make Vs project retail cost of the product.

 

Just to show you can have insane rise times and keep the noise down, this is a 40pS rise time generator powered via a usb 2 lead from a touch charger.

As you can see this is a significantly cleaner and and more accurate waveform, the harmonic's do go upto 1.5Ghz, however the amplitute is much lower and far more even handed.

 

Also included in the last image are two spectrum analysis one in linear mode the other in log mode.

 

 

 

 

 

 

[Dandou]

On 7/8/2023 at 11:48 AM, Clockmeister said:

Grounding is a interesting subject one which I really could deleve into great depth, so many audio manufacturers white papers I've seen on this at quiet frankly they are very wide of the mark.

This is RF at work, primarily conducted emissions, however there are other reference rail issue to overcome as well.

 

The whole point of the exercise is to reduce the PD (potential difference) between each of the 0Vdc (reference rails) between the various connected equipment to a low a resistance as possible. Taking into account the whole transmission line itself and the following impedances. Absolute ideal is a single connection point rather than multiple boxes which will give you variously different PD's. I'm not saying they don't work, what I suggesting is that it's not as optimal as it could be.

 

Over the years I have collected pretty much all of the latest and greatest grounding boxes, measured them in NON-Amir way (lol) in working systems with a calibrated set up to specifically designed to record and identify this very issue.

Hi @Clockmeister , 

 

Your post raised interest among contributors to Head-Fi who are building grounding boxes. 

I have two questions for you on behalf of @cdacosta from Heaf-Fi. 

 

1. Were you referring to star grounding the chassis/earth grounding of all components or signal grounding?  

 

2. How are you recommending this be accomplished? If the earth ground from the wall is already being split off from say a PLC to all other parts of the system, wouldn't the ground potential already be very close in relation to one another?

 

I apologize for being slightly off-topic with the subject of this thread. 

[Clockmeister]

Head fi eh?

 

All I am going to say on the matter that earth & ground get confused a great deal, ground is not earth this is mains (a/c) earth protection, ground is if you like the return path for the signal the 0Vdc rail. Surely Cameron has poo poo's this by now lol.

 

This is superdad's forum section it would inappropriate to discuss this topic in this thread.

[doitttt]

Thank you for your inquiry.
Yes, it's perfectly fine to use 75ohm cables for 50ohm to 50ohm connections.
50ohm output ---> 75ohm cable ---> 50ohm input No problem at all, clock characteristics are preserved.
This has been confirmed using a measuring instrument and has no effect on the characteristics of the clock signal.

< For your reference >
This is a characteristic because the output format of our clock products is sinusoidal.
If you use other clock products, be aware that if it's a square wave, you'll need to match the impedance exactly
from cybershaft
Kenji Hasegawa
Cybershaft

[Superdad]

16 hours ago, doitttt said:

< For your reference >
This is a characteristic because the output format of our clock products is sinusoidal.
If you use other clock products, be aware that if it's a square wave, you'll need to match the impedance exactly
from cybershaft
Kenji Hasegawa
Cybershaft

 

It is nice to see Cybershaft agree with what we keep explaining to people for the past couple of years: 

That impedance match of clock>cable>device matters only when square-wave clock is used; Does not matter at all for sine-wave clock.

 

[Jakenz]

2 hours ago, Superdad said:

 

It is nice to see Cybershaft agree with what we keep explaining to people for the past couple of years: 

That impedance match of clock>cable>device matters only when square-wave clock is used; Does not matter at all for sine-wave clock.

 

Indeed!

 

FYI Cybershaft repeat this as a standing statement on their product page for their new pure silver 75 ohm cable, without which assurance fewer of their target market of Cybershaft 50 ohm clock-owning clients would be interested methinks. 😉

 

https://cybershaft.shop/products/75ohm-pure-silver-bnc-cable