GUSTARD DAC-R26 Balanced Decoder R2R+1Bit Dual Native Decoding Music Bridge

[ray12345]

I'm still not convinced that it doesn't use the clock synthesizer (and therefore external clock, if connected) when you use oversampling. Or is oversampling disabled with I2S?

The external clock is indeed used. U18 will pass on the clock to the DAC when it is connected via I2S, this has nothing to do with whether oversampling is used.

 

[Nada]

I'm still not convinced that it doesn't use the clock synthesizer

Well lets have a think about this.

The R26 uses an internal clock synthesizer to generate the needed clocking rates for 44.1 and 48KHz intergers eg 88.2/96 etc. But none of these are freely divisible by a 10MHz clock whether its internal or external.

So whether using an internal or an external clock seems immaterial. The clock is essential processed by the clock synthesizer. It doesnt matter if its running NOS or OS. The original clock is heavily processed either way.

This raises the question of why use an external clock? From an engineering design perspective its daft. From a marketing perspective its brilliant.

 

[Alcophone]

The external clock is indeed used. U18 will pass on the clock to the DAC when it is connected via I2S, this has nothing to do with whether oversampling is used.

I'm not sure whether you're saying the R26 uses the external clock if it's directly connected to it even with I2S (which is what I'm wondering about), or whether you're saying it's indirectly benefiting from the external clock because that is connected to the U18, which will pass on its (word) clock to the R26. The latter is definitely true, but technically nothing special about I2S, the word clock is just right there as its own signal rather than being mushed together with everything else like with S/PDIF.

The R26 uses an internal clock synthesizer to generate the needed clocking rates for 44.1 and 48KHz intergers eg 88.2/96 etc. But none of these are freely divisible by a 10MHz clock whether its internal or external.

So whether using an internal or an external clock seems immaterial. The clock is essential processed by the clock synthesizer. It doesnt matter if its running NOS or OS. The original clock is heavily processed either way.

Yes, if the R26 needs to generate a clock signal for a certain frequency, the synthesizer is involved, we're in agreement on that. I'm stipulating that it might not need it at all in NOS mode with the I2S input, but would likely need it with oversampling.

My point is just that in NOS mode it could just use the word clock that's part of the I2S signal and ignore the clock synthesizer (regardless of whether the master clock is internal or external) because in NOS mode that's the speed at which it has to do things, but if it wants to do something more quickly (like with upsampling), it has to generate its own timing, I would think. And so it would use the clock synthesizer again, which should be able to do a better job with a higher quality external master clock than its internal master clock, meaning it should theoretically be beneficent even with I2S when upsampling is involved.

This raises the question of why use an external clock? From an engineering design perspective its daft. From a marketing perspective its brilliant.

Well you get the best results with oven controlled oscillators, and you need a lot of careful circuitry to not mess up that precision. That would increase the cost of the DAC. Instead you can put a reasonably good master clock inside the DAC and add the option to simply replace it with a better external one, either way you leave the rest to a synthesizer.
Why they chose a single 10 MHz master clock instead of one for 45.1548 MHz and one for 49.152 MHz, I don't know - you could generate any real world PCM or DSD frequency from that by merely skipping beats. Possibly because that's been an established standard for master clocks (see MUTEC REF10 and others for pro use cases).
I suspect it may have to do with all the other components that need some sort of clocking (USB, Ethernet, etc.), but that's just a hunch. Not sure a single clock synthesizer can generate multiple different frequencies at once like that.

On that note, is there a menu option to disable that "2nd order PLL" it has? Maybe it's reclocking I2S regardless of NOS or oversampling. The Holo Audio May DAC has an option to disable its PLL.

Can someone translate the following for me?

Clock synthesization: Replace the input signal' s clock with precision oscillators (local or external). Interference from the input source' s clock is avoided. Function off: IIs, AES, COAX, OPT, BT; PCM NOS, 1 Bit Function on: USB & LAN K2, ultra-low noise clock synthesizer: provides precision clock from local oscillators or external clock in.

Does that mean it's only using the synthesizer for USB and LAN audio, while otherwise being externally clocked? Would not be totally surprising since with asynchronous USB and everything it streams there's nothing doing the ticking for it already.

 

[Alcophone]

Had a closer look at I2S, and surely still have a lot to learn, but for one there's not just the word clock (when the sample for one channel is complete so it can switch to the buffer for the other channel), but also the bit clock (basically telling the destination "I've written a bit, go copy it to the current channel's buffer") and an MCLK pin pair. Now that looks suspiciously like master clock, but in the X26 manual it has this paragraph:

In BYPASS mode, the highest acceptable sampling rate for coaxial or AES is 192 kHz.
Meanwhile, the audio main clock (MCLK) frequency input by IIS must be 90.3168M or 98.304M.
Otherwise, it may cause silence when playing music with a relatively high sampling rate.
If you want to connect U16 to use this mode, because the default output frequency of MCLK of
U16 is 22.5792 or 24.576M, you need to unplug the J35 jumper of U16 to use this mode normally

That's 2048x 44.1 kHz or 2048x kHz, so very convenient for upsampling. Let's say you upsample 48 kHz to 768 kHz, then you just count when the MCLK has ticked 128 times and you got yourself a 768 kHz clock signal, no need for the synthesizer.
I don't know whether the U18 works that way, but it wouldn't be surprising.
How one accurately generates a 98.304 MHz signal from a 10 MHz master clock... that's a mystery for another day, I suppose.

 

[stoicar]

Hi i’m new here . Referring to 10M clock it is considered as a discipline clock. Is correcting and referring the clock in the dac. Is a norm for labs and measuring equipment. And yes is best to be ref with dac stremer and rooter if possible. But not a must. Everything have a buffer and a small clock. Some bad some ocxo or rubidium or gps. But all have someting
The most crucial thing in music besides resolution is timing human are extremely sensitive to timing and the skin goosebumps are usually when timing and resolution happens right

 

[comzee]

So whether using an internal or an external clock seems immaterial. The clock is essential processed by the clock synthesizer. It doesnt matter if its running NOS or OS. The original clock is heavily processed either way.

This is based. Also I partly alluded to this concept in an early post about reducing conversion steps.
I still believe the absolute best way to feed R26 is straight Ethernet.

Introducing U18 (or any other DDC) is adding #1 USB (which I've gone over has troubles) #2 An entire box that double converts digital (from USB in, to i2s OUT).
You can cut that all out with Ethernet direct. Also just as Nada said, it all goes thru the internal synthesizer anyway...

You guys see my crazy post about DDC early in this thread, I've tested a lot.
I currently own DI-20HE / Hydra-Z / U18 / Lynx AES16e / Singxer SU-6.

You know why I've used them in the past, and still own so many, it's because any DAC older than lets say, 5-6yr ago, greatly benefited from them.

I have a 15yr old 192 Dacute, a 10yr old Zodiac Gold.
I've also cycled many other DACs that have not been super modern.

Traditionally, hifi DACs internal USB was baaaaad. Really bad. Traditionally hifi DACs internal clock was very meh.
This is the case for both my 192 Dacute and Zodiac Gold. DDC immensely improve them.

The R26 already has a gigaclock AND a gigachad clock synthesizer INSIDE it.
As I said this is something relatively new for hifi DACs, and I'm all for it.

Boiling this all down, I still stand by my statement. I think Ethernet direct is the best method.

I mean, if you want to spend $2100 on C18 + U18 got for it, I've said my peace.

 

[Forgisound]

I used a SoTM SMS200 Ultra with their power supply. This was much better than directly connecting the mini PC with a USB cable to the DAC. The SoTM has an ethernet input and a USB output to the DAC and works as a reclocker and some sort of noise cleaner. Now R26 will replace SoTM and DAC. I hope the sound will be even better. As I already wrote, the quality of the cable is also very important, in this case the ethernet cable, and so far also the USB cable.

 

[EMINENT]

Why am I getting clicks and pops upsampling PCM-DSD with U18-i2s-R26?
NAA/USB direct does not.

 

[GP123]

Hi I’m new here. From the original Chinese description, R26 will always use internal K2+ext.10M for LAN and USB input. If PCM NOS is enabled it will only use the source clock. Same for DSD direct. In other words if OS is enabled it will always use internal K2+ext.10M, even if I2S is used. Hope this helps.

 

[Knobstler]

how does the u16 clock differ from the u18?

 

[Highfive99]

how does the u16 clock differ from the u18?

The U16 doesn't use the K2 that the U18, X26p or the R26 use. The U18 is a newer generation. I have both and the U18 made a noticeable improvement over the U16

 

[Alcophone]

Why am I getting clicks and pops upsampling PCM-DSD with U18-i2s-R26?
NAA/USB direct does not.

Faulty or too long HDMI cable, maybe?

 

[BShaw]

This is based. Also I partly alluded to this concept in an early post about reducing conversion steps.
I still believe the absolute best way to feed R26 is straight Ethernet.

Speaking of ethernet, there was some discussion earlier in the thread about the Supra cat8. I just wanted to convey the response I got from them with some of the construction details:

"It is the same Copper OFC 5N. It is stranded but pressed together to make the wire round and the impedance stable. The PE is foamed. I have updated the data sheet.

With best regards

Jorgen"

In another email he told me they were made in house 100%

 

[Nada]

Buying an "audiophile" Ethernet cable is irrational. This is a data cable. A sensible choice is to buy a well designed industry standard Cat cable and benefit from the best design and construction standards. As an extra benefit save dollars.

As a community we need to be aware that many posts saying buy XXX brand cable are, as a far as we know , just marketing shills. The research into social media reviews reveals a sizable number are shilling.

Obviously I have nothing to sell or gain posting this opinion. As always we are free to waste out money as we like

 

[BShaw]

Obviously I have nothing to sell or gain posting this opinion.

Me either, FWIW. Just bringing closure to an earlier question.
I have little reason to doubt you Nada, I'm really not that knowledgeable about some of this minutia. But I have read some convincing arguments, and not being sure, I dropped an extra $100 to remove any doubt. My experience has shown me that with other cables, even USB, the cable matters. Just my opinion, not looking for an argument here. Could very well be that ethernet cables are different, but I'm going to err on the side of caution. Hell, there are 1m ethernet cables over $1k! That's where I start to be very suspicious. But at $14/m, the Supra isn't such a huge risk. Plus my mate is a Swede and from very near where they're made, so she got a kick out of it.