[sajunky]

This is where I lose all my logic. Is this some sort of marketing 2 in 1 thing?

75Ohm was a standard for word clock in the professional audio equipment and TV installations.

50Ohm was a standard in measuring equipment, amateur radio, military radar installations, etc.

When 10MHz clock was introduced, designers decided for 50Ohm cabling, I think it was in order to differentiate from the existing audio word clock systems. It makes sense, as these clock systems are functionally incompatible, i.e., word clock frequency depends on the current sampling rate, while 10MHz clock is fixed frequency and requires sophisticated clock synthesisers to work.

Audio GD has 75ohm output for a word clock, it is compatible with professional audio and 50Ohm inputs for 10MHz clock, it matches most of 10MHz sources. I don't see any deviation from a common practice.

A for compatibility of connectors, it depends. On the long cable runs impedance must be maintained to avoid standing wave (reflection on both ends: back and forward). However tolerance of cables impedance is not maintained very strict, making matching difficult, reflections on connectors are much smaller than cables. So matching connectors is less important. On the short runs matching impedance is even less important. It is why for the 10Mhz clock applications it can be assumed that these connectors are compatible.

Hope it clears any questions.

[EDIT] Similar discussion is pending in parallel. Can anyone repost this in high-end R-7 section? Someone has requested to cut my access there and - it was done.

 

[OCC7N]

75Ohm was a standard for word clock in the professional audio equipment and TV installations.

50Ohm was a standard in measuring equipment, amateur radio, military radar installations, etc.

When 10MHz clock was introduced, designers decided for 50Ohm cabling, I think it was in order to differentiate from the existing audio word clock systems. It makes sense, as these clock systems are functionally incompatible, i.e., word clock frequency depends on the current sampling rate, while 10MHz clock is fixed frequency and requires sophisticated clock synthesisers to work.

Audio GD has 75ohm output for a word clock, it is compatible with professional audio and 50Ohm inputs for 10MHz clock, it matches most of 10MHz sources. I don't see any deviation from a common practice.

A for compatibility of connectors, it depends. On the long cable runs impedance must be maintained to avoid standing wave (reflection on both ends: back and forward). However tolerance of cables impedance is not maintained very strict, making matching difficult, reflections on connectors are much smaller than cables. So matching connectors is less important. On the short runs matching impedance is even less important. It is why for the 10Mhz clock applications it can be assumed that these connectors are compatible.

Hope it clears any questions.

[EDIT] Similar discussion is pending in parallel. Can anyone repost this in high-end R-7 section? Someone has requested to cut my access there and - it was done.

Thanks for the explaination, but why not keep it to one type instead. It does not make sense to me. All highend brands in hifi is mostly 75Ohm. They dont mix it up

 

[sajunky]

Thanks for the explaination, but why not keep it to one type instead. It does not make sense to me. All highend brands in hifi is mostly 75Ohm. They dont mix it up

Major high-end audio brands responded very slow to the 10Mhz clock. I think most of them are still using word clock systems. If you connect these both, it won't work. So what is a benefit of using the same cabling system? This is my question. At least ordinary users will notice a difference and won't waste a time in connecting incompatible systems. I think it was a logical choice.

 

[OCC7N]

Major high-end audio brands responded very slow to the 10Mhz clock. I think most of them are still using word clock systems. If you connect these both, it won't work. So what is a benefit of using the same cabling system? This is my question. At least ordinary users will notice a difference and won't waste a time in connecting incompatible systems. I think it was a logical choice.

Can you recommend a great setup with R8HE mk2 + DI20HE. What should I tell magnahifi for the configuration?

 

[Jake2]

75Ohm was a standard for word clock in the professional audio equipment and TV installations.

50Ohm was a standard in measuring equipment, amateur radio, military radar installations, etc.

When 10MHz clock was introduced, designers decided for 50Ohm cabling, I think it was in order to differentiate from the existing audio word clock systems. It makes sense, as these clock systems are functionally incompatible, i.e., word clock frequency depends on the current sampling rate, while 10MHz clock is fixed frequency and requires sophisticated clock synthesisers to work.

Audio GD has 75ohm output for a word clock, it is compatible with professional audio and 50Ohm inputs for 10MHz clock, it matches most of 10MHz sources. I don't see any deviation from a common practice.

A for compatibility of connectors, it depends. On the long cable runs impedance must be maintained to avoid standing wave (reflection on both ends: back and forward). However tolerance of cables impedance is not maintained very strict, making matching difficult, reflections on connectors are much smaller than cables. So matching connectors is less important. On the short runs matching impedance is even less important. It is why for the 10Mhz clock applications it can be assumed that these connectors are compatible.

Hope it clears any questions.

[EDIT] Similar discussion is pending in parallel. Can anyone repost this in high-end R-7 section? Someone has requested to cut my access there and - it was done.

Nice explanation and history thanks @sajunky.

On the topic of the criticality (or not) of impedance matching, reflections and their effect on jitter and considerations in using external clocks, over on the R26 thread I recently came across a fascinating 2021 white paper by John Swenson of Uptone Audio. Expect some of you may be familiar with it already as I found a couple older refs in two other forums but not in this or other related threads.

TL;DR - in the view of the paper author:

• square wave super sensitive to reflections in cable/signal path but not to external sources of RF (etc) noise affecting signal amplitude (takeaway - precise impedance of cable and matching to components is critical, shielding from noise not so much);
• sine wave - cables, at least short ones - not as sensitive to impedance mismatch reflections (no precision square wave-front to have its rise time messed up) but the integrity of its shape for timing purposes receiving circuit of the sine signal is likely more sensitive to RF noise/ in the amplitude axis so good shielding is critical.
• Impedance mismatches require careful design and implementation to avoid at every point in the signal chain from oscillator to end load/DAC and can occur internally within a clock - there is no way to know whether a particular external clock design or instance has entirely avoided this. I guess an example of this may be use of 75ohm style BNCs rather than 50ohm, though it may be a designer like LHY can compensate through upstream circuitry to still deliver the desired characteristic impedance in the circuit(?).
• Swenson's recommendation - if in doubt use sine wave clock signals as more forgiving of impedance mismatches within the cable and the clock and recipient DAC/DDC, but instead of spending on an expensively shielded or low reflection cable just use a specifically designed filter at the load end (eg DAC) to strip out any noise beyond 10mhz the cable has picked up to clean up the sine wave. The theory being a basic and potentially noisy cable can be used for sine as the filter will clean it up by getting rid of the noise outside the clock signal band.

The paper and its recommendation of going sine + filter prompted an Audiophilestyle forum group buy of the Mini-circuits filters. MC is a specialist RF cable/equipment manufacturer I linked in a previous post re their precision testing cables with v low internal reflection specs. AfterDark bought up all Mini-circuit's stock for the group-buy. I see the filters are still stocked by AD and are back in stock at Mini-circuits.

Has anyone here tried these filters?

I haven't tried the filters but am really intrigued. At USD40 they're not expensive so would give em a go but they're extortionately expensive to ship from either vendor to NZ. May get a mate returning from the States at Christmas to bring a couple back. Incidentally when I finally got around to AB'ing sine vs square from the OCK-1 to my R26 and/or with U18 using Gustard C2 cables I found sine sounded way better. No comparison. So applying John Swenson's logic, notwithstanding my attempt to use decent quality hopefully low reflection cables, there was likely still one or more impedance mismatches in the signal path from OCXO to R26/U18 PLL circuit.

 

[sajunky]

Has anyone here tried these filters?

I haven't tried the filters but am really intrigued.

Maybe it is a question in a wrong place. Your experience with R26 can be completely different. Kingwa preference to the square wave source reflect on tuning of the clock synthesiser. The output from the clock synthesiser drives all internal circuits, it should be evaluated how analogue filter (destroying square shape) influence output jitter. This technology consists of digital PLL (PLL coupled with clock dividers). If you are familar with PLL, it is an effective filter, it can be tuned to very high Q.

In other words, when you are using such filter, you are converting square wave source to a very rough sine wave. It shouldn't be used with a square wave source, IHMO.

I am actually more concern about other issue. Low frequency jitter is more important than high frequency, as a source is always weak in this area. I think we should focus on a low frequency jitter while these analogue filters are completely transparent to the acoustic frequencies, and placing inductive elements on a cable make it very sensitive to the environmental noise which carries a lot of power lines noise. This is a huge issue in this idea.

 

[gimmeheadroom]

75Ohm was a standard for word clock in the professional audio equipment and TV installations.

50Ohm was a standard in measuring equipment, amateur radio, military radar installations, etc.

When 10MHz clock was introduced, designers decided for 50Ohm cabling, I think it was in order to differentiate from the existing audio word clock systems. It makes sense, as these clock systems are functionally incompatible, i.e., word clock frequency depends on the current sampling rate, while 10MHz clock is fixed frequency and requires sophisticated clock synthesisers to work.

But as we see, it doesn't help at all but rather just adds confusion, since you can plug a 75 ohm cable into a 50 ohm socket and vice versa.

Audio GD has 75ohm output for a word clock, it is compatible with professional audio and 50Ohm inputs for 10MHz clock, it matches most of 10MHz sources. I don't see any deviation from a common practice.

The DI-20HE uses a 50 ohm clock output for WCLK. I was surprised to learn this but @FredA reminded me:

http://audio-gd.com/R2R/DI20HE/DI20HEEN_Use.htm

"CLK OUT ：
Outputs 256fs main clock or LRCK/WCLK (3.3V LVDS @ 50 Ohm) . User-selectable "

So we're back to gratuitous incompatibility with long-standing standards from the 1980s.

[EDIT] Similar discussion is pending in parallel. Can anyone repost this in high-end R-7 section? Someone has requested to cut my access there and - it was done.

Outrageous!

 

[sajunky]

The DI-20HE uses a 50 ohm clock output for WCLK. I was surprised to learn this but @FredA reminded me:

http://audio-gd.com/R2R/DI20HE/DI20HEEN_Use.htm

"CLK OUT ：
Outputs 256fs main clock or LRCK/WCLK (3.3V LVDS @ 50 Ohm) . User-selectable "

So we're back to gratuitous incompatibility with long-standing standards from the 1980s.

You are right, now I remember we had a discussion before. It was your main objection, I can't disagree.

Regarding 10MHz clock input, it is following a growing standard which is 50Ohm, should stay as is.

 

[Jake2]

Maybe it is a question in a wrong place. Your experience with R26 can be completely different. Kingwa preference to the square wave source reflect on tuning of the clock synthesiser. The output from the clock synthesiser drives all internal circuits, it should be evaluated how analogue filter (destroying square shape) influence output jitter. This technology consists of digital PLL (PLL coupled with clock dividers). If you are familar with PLL, it is an effective filter, it can be tuned to very high Q.

In other words, when you are using such filter, you are converting square wave source to a very rough sine wave. It shouldn't be used with a square wave source, IHMO.

I am actually more concern about other issue. Low frequency jitter is more important than high frequency, as a source is always weak in this area. I think we should focus on a low frequency jitter while these analogue filters are completely transparent to the acoustic frequencies, and placing inductive elements on a cable make it very sensitive to the environmental noise which carries a lot of power lines noise. This is a huge issue in this idea.

Cheers sajunky. I defer to your greater knowledge here - I just find the whole area fascinating and am reading what I can to get a better mental model of the many variables in play. Your comments re PLL tuning, filter frequencies and inductance make clear that I'm still barely scratching the surface. Btw to clarify the filter is only being proposed with a sine wave output, not a square wave.

I posted this here as it has been one of the more sustained and friendly discussions around this topic (alas not the case on all threads). Felt his theories may help explain some of the varying results people get with clocks/clock cables. But perhaps the master clock thread is a better spot.

Cheers,
Jake

 

[bodiebill2]

Audio GD has 75ohm output for a word clock, it is compatible with professional audio and 50Ohm inputs for 10MHz clock, it matches most of 10MHz sources. I don't see any deviation from a common practice.

Very useful post, thanks!
Although the DI20HE WCLK output is standard 50 Ohm as confirmed by Kingwa and Magna Hifi. I had it converted to 75 Ohm.

P S Sorry, this was already mentioned by @gimmeheadroom above...

 

[iFi audio]

Afterdark Trifecta Emperor Double Crown

Just googled that. Not having heard it I can't comment what that clock does, but it sure looks good all in carbon!

 

[sajunky]

Btw to clarify the filter is only being proposed with a sine wave output, not a square wave.

I posted this here as it has been one of the more sustained and friendly discussions around this topic (alas not the case on all threads). Felt his theories may help explain some of the varying results people get with clocks/clock cables. But perhaps the master clock thread is a better spot.

Yes, It is expected. I just mentioned it as most of people following Kingwa recommendation will purchase devices with square wave clock output. Users may be confused with this project, now they won't. Filters are not a right solution for a square wave output, it is plain and simple.

New ideas are emerging. While in this case I have a reservation even in the sine-wave output setup, but members and myself will be interested in your feedback, thanks in advance.

[EDIT] On a second thought, filter on the transmission line is a complete nonsense, including a sine-wave clock output. If you look at the input impedance of a narrow-band pass filter, it has a peak at the nominal frequency. While it can (?) be tuned to the transmission line value, all other frequencies will bounce, it should be surpressed as it happens on the pure resistance terminator.

A solution is to terminate a line correctly with a pure resistor, then buffer the signal and apply filter afterwards. It means an active buffer. The best place for such design is inside receiver.

 

[DecentLevi]

I have reboot 10 times, reinstalled firmware 20 times and burnt in unit for over 1000 hours. I have reinstalled Amanero driver many times as well. I have not specifically burnt in using the spdif output functioning as all burn in has been with the HDMI output- and this may be the last possibility. All input is USB from PC only...

Not sure if you've solved your problem yet, but here's my 2c with a similar problem that may help. I used to get a loud screeching sound from my DI20 (non-HE) when navigating within, or stopping DSD playback. After making the following cabling upgrades, the problem never came back. In the last few months I have added an iFi iDefender from the PC connected to a Fibbr Alpha USB-optical-USB cable, iFi iPurifier 3 (which goes into my existing Wyrd), then a new much thicker gauge short-run USB cable, (which goes to my existing iFi iPurifier 2), then out of the DI20 I'm using a much better quality RCA coax cable to my DAC. What I believe to be making the difference is thicker USB and digital RCA coax cables which happen to be made of pure sterling silver - these are cheap Chinese knockoffs of multi-thousand dollar cables that are actually made of said material, which you can read about here. In terms of the iFi products, these seem to be adding more depth, purity and detail but I believe it's these better conducting cables that are fixing any data transmission issues.

 

[DecentLevi]

My system is quite transparent. I use optical isolation, and all the gear is powered with good LPSs.
At the beginning, my interpretation was similar to yours. I thought that with the DI-20HE the transparency of my system was even greater, and that was the reason for which I can hear an improvement when iGalvanic is placed between the Allo USBridge and the DI.
But my interpretation changed after thinking to what happens in the system of @bodiebill2 . He told me that with his After Dark Rossana streamer + his DI-20HE, the sound was not as good as with his SOtM + DI-20HE, despite the fact that Rossana benefits from the Diretta protocol, and is configured as a Diretta Target. He concluded that the USB port of the SOtM was better implemented than the port of Rossanna. That's why I reached the conclusion that the galvanic isolation of the USB port of the DI-20HE was not that great.

Pegasus has a good USB implementation by AMR. I just presumed that it has a galvanic isolation, but I did not see any technical information about this port, and as far as I know the Musician did not release it.

I don't have iFi iDefender, but I have iFi iPurifier 3. When I place it alone between the USB ports of the Allo and the DI-20HE, it improves the sound, but not as much as iGalvanic. When I use both iGalvanic and iPurifier, the improvement is very slightly better than with iGalvanic alone.

I bought the Gustard R26. When I will receive it, I will proceed with additional tests. I will report how its USB input behaves VS the input of the DI-20HE.

After some research and familiarizing myself with many obscure products mentioned, I was finally able to understand this post. It seems you're basically saying the DI20's USB input benefits from clean, regenerated USB input. That definitely agrees with my experience, being that I have a series of 5 USB purifiers/defenders/regenerators between my laptop of the DI-20, and I could hear an improvement every time, usually for the better depending on the placement. Additionally I found that thicker cables and premium conductors made a difference, as well as both specific ferrite clamps and copper foil tape together on the outside of the USB and RCA coax cables.

Do you or anyone have an opinion on any difference between the iGalvanic and the Schiit Wyrd? Although not quite the same category type, jut curious about the sonic differences.

 

[sajunky]

Do you or anyone have an opinion on any difference between the iGalvanic and the Schiit Wyrd? Although not quite the same category type, jut curious about the sonic differences.

How your output device name is changing (from Amanero digital....) in the Device Manager when inserting reclockers or isolators like iGalvanic? Report back, it is important for asynchronous data delivery. DI-20 assumes that transfers are asynchronous. When asynchronous transfer is broken, it may generate noise as there is no reclocking on USB port! Some other brands do always reclock, quality drop may be unnoticed.

USB cables are important and placement of a ferrite on the cable is critical. Also order of devices in the chain. By example iDefender, must be the first device in the chain, iFi recommends direct plugged to the USB port - not using the cable, and redirecting ground loops with external power supply is most effective. Extra devices are optional and order is not so clearly defined as many have a mixed function. Reclockers (if used, I really don't like these) or isolators should be placed closer to a DAC.

Now about chosing the right cable in the right place. As mentioned before, for devices with ground loop redirection function, the best is no cable between USB host. If using cable, it must be a thick one (thick wires, not the outher jacket) and sans ferrite. Having a multimeter, check resistance of the wires. Poor cable may measure 1Ohm on a distance 1meter. Power supply of the ground loops redirector should be plugged close to the computer's outlet.

Cable coming out of ground loop redirector may have a larger resistance, but should be well shielded and have a ferrite (either moulded or added a clamp). Closer to a DAC, better shielding and bigger ferrite. But ferrite choke do not block mains frequency noise, only HF. It is why ground loop redirection (if properly made) is the most critical part of installation.