it doesn't say which chip it uses, the isolation is a great idea, but it doesn't work.
The computer streaming is polluted by CPU and video card, it gets into the SPDIF and even optical.
You need a dedicated streamer between the computer and dac which has no switching power supply.
ANDY: i know I used to have a pcm53-K (highest grade x2) and I can recall that. I used the resistor IV.
The computer streaming is polluted by CPU and video card, it gets into the SPDIF and even optical.
You need a dedicated streamer between the computer and dac which has no switching power supply.
ANDY: i know I used to have a pcm53-K (highest grade x2) and I can recall that. I used the resistor IV.
Not really. Having a crappy dac that is too sensitive to computer noise then trying to fix it by getting streamer can be kind of like putting a bandaid on a deep wound. In most cases the problem is with the dac, although sometimes one can come across a really pathologically noisy computer.
It turns out that if there is a weak point in a design then that may dominate what limits potential sound quality. But, it isn't the same weak point in every system or in every dac design. And sometimes there are multiple weak points.
Anyway, a well designed dac should use ASRC and or FIFO buffering for SPDIF/TOSLINK type signals. If a dac is still using the old, and lower cost, PLL clock recovery approach then its not a very good dac. In that case the is going to be much more influenced by any instability of the computer's clock.
Which streamer has no switching PSU onboard? I have never seen one, every board (even for small ARMs) has at least one (but typically many more).
IMO this is mixing ground-loop issues which are not present in proper SPDIF coax (i.e. with a transformer) and optical, also not in USB via isolator.
IMHO no well-designed DAC uses FIFO for merging clock domains. FIFO by principle cannot merge the clocks reliably.
Every SPDIF receiver recovers the incoming clock via PLL. Then the jitter of the recovered SPDIF clock can be with higher or lower success reduced with asynchronous resampling to the precise DAC clock. But IMO best is to avoid the clock recovery completely - i.e. e.g. USB async or slaved I2S directly from the streamer SoC, both master-clocked by the DAC clock.
Understood, yet its been done pretty successfully in practice for consumer hi-fi use. Again in practice, it can sound better than ASRC. The main issue can be the delay.
Also IIUC, FIFO combined with ASRC is used in Tambaqui dac.
Last edited:
Of course. The main issue is if the dac MCLK is PLL derived from the SPDIF stream. Although asynchronous USB is clearly better, some users will still want to use SDPIF. Heck, recording studios are wired up for dacs in closets using AES serial data input. So there is still a need to be able to do it as cleanly as one can manage.
I bought an old streamer from Auralic. Original Aries Femto. Only becuase in the bedroom I have their old Altair and in the shed Mini Aires and I like the platform enough. I hooked it up 10tb HDD and I stream around the house. As I understand , them main difference between this and newer $2k- basic Aries from their offerings is galvanic isolation and better better interfaces and clocks which reportedly offers nice improvement. The reason I bought old Aries is that unlike Aries Mini it offers downsampling option to PCM 24/192 which I need to use with old Theta Gen VIII mark 3 boat anchor. It was the cheapest option.
I don't use computer , actually streamer is a dedicated computer so I use it as a transport and it has linear PSU. The USB/Spdif gizmo is based on Amanero platform whatever it means. Basically I didn't want to spend any money on this but it sadly can't be avoided.
Well, unless users want to use such (often quite expensive) DAC for AV, or even gaming... The initial latency is huge (by principle) and drifts in time (also by principle). My 2 cents quite a few buyers into this method (I would tend to call it a bit hyped up, similarly to long-distance I2S...) have no idea about this and then are disappointed upon finding out the hard way...
IMO recording studios have master clock by their ADC (where it matters for a recording studio 🙂 ) and the PLL clock recovery (if not using centrally-distributed master clock ) is perfectly OK for their monitoring purposes. Just like IMO it's perfectly OK for any consumer hifi, inaudible if done reasonably well. Yes, jitter reduction by subsequent ASRC improves measurements, but IMO is hardly audible. In fact many people complain about "dry" sound of ESS DACs, basicall all with ASRC enabled which in many cases measure superbly. Just like many peple swear by tubes which have nothing to do with fidelity.
In other words - IMO the PLL recovery circuits of modern SPDIF receivers are fine for listening and the ASRC in ESS DACs is mostly for the implementation convenience - the DAC chip accepts any incoming samplerate over I2S automatically, no need for complicated/costly control over I2C + fast smooth rate switching which is hard to make with SPDIF receiver I2C -> controlling MCU + complicated firmware -> DAC I2C.
Last edited:
I wouldn't expect the $400 Chinese box to necessarily be in the same class. You might be disappointed.
It's just the interface. Femto actually plays pleasantly with Antelope Zodiak Gold mastering DAC from USB and Zodiac shows 384Khz PCM samples on the display while playing DSD files which I have 1300 albums
I would disagree. ASRC changes the sound, mainly because there is always noise, and even a little power supply noise IME will show up as an ASRC PPLL is trying to track. That in turn, degrades stereo imaging (among other things) so that the quality of a recording cannot be accurately judged. However, if the only goals are things like balancing frequencies and adjusting volume levels, then imaging may not be a concern.
Thus when imaging matters AND if SPDIF is required, then I would rather have a well designed FIFO solution.
Of course, for my own dac I use asynchronous USB, since that's the best overall choice 🙂
Last edited:
Frankly, I'm surprised at that comment. Are you one of the people who takes seriously the SINAD/$ theories from over at ASR?
Reason I ask is because it seems to me there are a lot of unproven assumptions baked into that point of view, especially when it comes to the kind of strong nonlinearity that goes on inside a modern sigma-delta dac. Its a very different type of thing as compared to a weakly nonlinear analog amplifier. Yet somehow the criteria used for interpreting measurements which were studied mostly in the days before modern digital are still assumed to apply equally.
Much more can be said on the above subject, of course. I wouldn't be bringing it up if I didn't believe there were good reasons for doing some more thinking about it. IMHO, we need to rethink how we differentiate noise from distortion. Right now most people are still doing it by the eyeballing an FFT looking for spectral peaks and a stationary noise floor. That's not all there is to see.
Last edited:
I am sorry for confusion - actually I was talking about PLL in the SPDIF/AES receiver, which cannot be avoided in any case (eventually using distributed master clock to read from a tiny FIFO to align the PLL'd stream).
But certainly not in a recording studio where low latency of the monitoring is of critical importance. Also when doing digital-audio editing (DAW). I have never heard of a professional equipment using the crude/dumb HW FIFO method.
But ASRC made of a short FIFO combined with some rough buffer-alignment techniques is quite common - I suspect Dante/AES67 receivers use that commonly, to align the incoming PTP-aligned stream to the hardware clock of their output interfaces. I doubt they implement some complex ASRC in the receivers (often low-level FPGAs) since they care about latency a lot. Also gstreamer does it when rendering master-clock streams - dropping or inserting an interpolated single sample as needed in time, basically inaudible. Of course perfectly visible on any precise spectrogram ex post - but that's not an ear 🙂
Last edited:
I would opt for USB and get the dac out of the wiring closet. There is no other way to do it well. For surround systems, unfortunately for now all we have is TDM over UAC2. For hi-res surround it gets more complicated. Maybe a dedicated hardware rack with its own bus.
I mean, forget the recording studio for a moment, how is is the mastering engineer supposed to do his job for a high quality surround symphonic recording?
BTW, don't know if you have been following it but latest rave is a new dac with 170+dB dynamic range. And its for real. Wayne from Pass Labs just reported getting one. Mastering engineers are refusing to return the review models. Pass Lab employee is refusing to return the dac to the office. https://www.diyaudio.com/community/threads/27bit-dac-162-db-dynamics.406498/post-7998112
However, this dac only measures exceptionally well in terms of dynamic range. Yet, people appear to be claiming to hear low level details like never before.
Last edited:
Or PCI-e which has lower latency potential and less CPU load than USB.
UAC2 standard defines 27 channels, and more can be used with a custom driver (the USB isochronous bandwidth limit is three 1,024-bytes transactions per every 125us microframe which yields 3 transactions * 1,024 bytes * 8,000 uframes / ( 48,000 samples * 3 bytes per sample) = 170 channels at 48/24 isch (a bit fewer for UAC2 async one-sample overhead). Or 42- channels at 192kHz. Just getting enough I2S data lines, TDM is not necessary. My 25USD hardware of choice for UAC2-> I2S has 8 standard I2S lines (16 ch/32bit) at 384kHz out and 10ch in.
I have seen that, the technology makes sense. A combination of DAC + ranging reference voltage/gaining for the extra bits is nothing new - but to make it work right was certainly another story 🙂
I wonder why there isn't a thread "are there any excellent inexpensive phono cartridges" -I wonder if it's because the DAC is the only thing CD/digital people have to fiddle with, and we like to fiddle.
Thank you. Will make a note of that. Might have come across it before but then forgot about it 🙁