I run Audirvana Studio on a MacBook Pro M2 and I have a Silent Angel M1T Network Streamer connected to a Gustard R26 DAC.
I am evaluating three different scenarios:
a) Macbook directly connected via USB 2.0 to the DAC
b) AS interfaced (UPnP - Ethernet) to the network streamer
c) AS interfaced (UPnP - Ethernet) to the DAC
According to your experience what could be the best scenario in order to have the best audio quality ?
Thank you
The very short answer is that since you have this specific equipment and the means to evaluate it, while others are unlikely to (if someone else has all these pieces, please step in), plus only you know your own personal preferences, you are in by far the best position to answer this question for yourself.
If you don’t find sound quality differences between these three scenarios in your listening, then you are free to choose whichever one gives you the most reliability and convenience.
Now I haven’t answered your question, and I (and the other people here) would like to be helpful if possible. Can you tell us more about why you asked the question? Do you want to save the time of listening to all three setups, or are there specific questions you have about one or more of these scenarios?
It is quite difficult to judge these scenarios because switching from one to another takes some time (cables, device configuration). To better understand differences I should have the three scenarios available in parallel, for a real time evaluation but it is not feasible. For sure the differences are not huge.
My question implied evaluations at a higher level, not depending on the specific equipment I have. Maybe AS was designed to give better results with a direct USB connection rather than UPnP (it’s only an hypothesis). Maybe AS ‘prefers’ a direct connection with a DAC instead of a network streamer as an intermediate node (it’s just another hypothesis).
Sharing the experience of other members of the community might help me to focus on a scenario or at least, to have e better comprehension of the phenomenon.
How far away is the DAC from the computer?
They are very close, less than 1 meter
Do you have the companion LPS for the Silent Angel? If so, it would seem that using the I²S output of the Silent Angel M1 to the Gustard with a very short I²S cable would be your best application… If you don’t have the LPS then this is where the true qualitative test is… My perspective is from general reviews of this DAC.
Edit: Use the shortest good quality USB interconnect to the DAC and/or the Silent Angle that is possible… @Jud would recommend this:
Your question is very reasonable. I think you’ll find members of the community have many different answers! 
Let me have a look at the Silent Angel, with which I’m not familiar, before I say anything.
Thank you !
Jud, I am not an expert, I entered this wonderful world quite recently and I am in a phase of continual learning. That’s why the experience of skilled contributors like you have a big importance to me.
You may want to center your focus on the Silent Angel as your primary DDC/streamer for the reason that you may eventually want to try a different DAC… The Silent Angel generally seems to be worthy of that place in your playback scenario…
@Mahatma63 …What computer are you using…? The reason I ask, is because it is not clear how the Silent Angel M1T handles DSD over PCM (DoP) signals… This may make a difference in the approach to connectivity from the computer to the streamer. (PC/Windows, Mac/macOS)… I’m presuming the M1T handles DoP signals.
I use a Macbook Pro M2. I cannot answer your technical question about how M1T handles DSD over PCM…
So… given the stellar development expertise of the Silent Angel network products, it would be safe to presume the Ethernet interface and UPnP protocol support is very good… however, you will be the final judge…
The MBP M2 is certainly a fine platform, and I presume you have more than 16GB of System RAM to work with. MacOS supports transmission of DSD via DoP (up to DSD256 in Audirvana) on both the USB and Ethernet buses, and will transmit native DSD via Ethernet bus using UPnP protocol.… The maximum DSD sample-rate supported by the M1T is 11.2MHz (DSD256) and the maximum PCM sample-rate it supports is 768kHz (DXD).
I suggest contacting Silent Angel regarding DSD via DoP support… I am almost certain it does, however there is no description of this support in their overview/specs and quick-start guide.
Some of the issues with UPnP transmission are centered around how UPnP is supported in the device (DDC or DAC) in concert with Audirvana, and then the nature of the network configuration, regarding QoS (Quality of Service) and other technical electro-mechanical gremlins that affect signal-integrity.
If you utilize a very short Ethernet cable from the MBP to the M1T, you may find this just as easily workable as USB…
USB protocol is codified and most reliable… the quality is intrinsic to the USB interface component architecture and the synergy with the circuit topology of the M1T…
The optical USB audio cable, provides isolation (used to connect the MBP and M1T USB input) however @Jud has found that a good LPS powering the DAC side of the cable makes an audible difference… (There are other ways to isolate USB wire interconnects and this may be already done in the M1T platform design) … Ethernet provides a limited level of isolation inherent in its design protocol, so this may be the least expensive interconnection (keep it as short as possible)… It is imperative to keep the computer and the playback component power-feeds on a common power/ground/earth circuit… This should be very easy for you to do since the components are in close proximity to each other…
I hope this helps…
After having read your post, I’m definitely amazed by how you could explain with incredible details. I will contact Silent Angel to clarify the point you have arised and I will perform further tests with different cables in order to get the best solution.
Thank so much for the time you spent helping me.
Marco
Hi Marco -
As I mentioned, members of the community here will have different thoughts. 
Mine are a bit different than @Agoldnear, and that’s fine - you are the one who’s listening, so you get to be the final judge, and whichever way you decide I’m sure things will sound very nice.
Because there are several factors that may have an effect here, for clarity I want to set them out one at a time, moving from the simplest to those that are slightly more complex.
Fan noise: Later on I will be mentioning using Audirvana to upsample to DSD. The M2 MacBook Pro should be able to handle this without audible fan noise. However, if you do hear fan noise from the MacBook Pro, then you will want to move it somewhere that you can’t hear it and use UPnP rather than a direct USB connection.
Reliability of the UPnP connection: UPnP has been referred to as “the most non-standard standard.” Different manufacturers implement UPnP in their own ways. So if you have difficulty with the UPnP connection - either Audirvana can’t make the connection, or it drops often enough to be irritating - then use the direct USB connection.
Those are the very simple and clear situations, now on to the slightly more complex ones.
Whether to use the Silent Angel or not: It appears from the Silent Angel documentation that the maximum resolution is 384kHz PCM or DSD128. Your DAC has all the network capabilities of the streamer and will accept higher resolution signals, which will be lower noise (important for an R2R DAC like yours), and it will simplify your system because you will have one less box and power supply. Therefore I’d suggest you may experience best sound leaving the Silent Angel out of the system and connecting via USB or UPnP to the R26 DAC.
Leave the DSD Direct setting off: Your DAC has a setting that eliminates a filtering step and passes through DSD signals directly. However, it measures substantially better in terms of distortion and noise with that setting (called DSD Direct) off. As noted on this page that tested your DAC:
However, the built-in DSD direct interpretation module is a bit too magical
On the contrary, it is better to turn it off (it is turned off by default)
This comment plus measurements are from Measurements & Review of Gustard R26 Network R2R DAC - L7Audiolab
You will find information about the DSD Direct setting starting on page 2 of your R26 owner’s manual at http://www.gustard.cn/qfy-content/uploads/2022/07/b669e827d55d7c660317cf9860670ce0.pdf
So now my suggestions are narrowed down to USB or UPnP directly to the R26 (assuming neither fan noise nor dropping the UPnP connection are factors) with the DSD Direct setting turned off. Two more factors to consider:
Transmission of electrical noise: If both USB and LAN connection are copper, the LAN connection will transmit slightly less electrical noise because isolation (via small transformers) is built into Ethernet, but not USB. You can isolate either one by making the connection through optical fiber, though this runs into some expense (switch with SFP+ connection for Ethernet, the optical USB cable @Agoldnear linked plus power supply for USB).
Resolution to send to the DAC (upsampling): As mentioned above, it’s important to reduce the noise sent to the DAC. I think a DSD256 or DSD512 signal using the upsampling in Audirvana is likely to provide the lowest noise and distortion. I would suggest selecting the B7 or B8 modulator in Audirvana’s upsampling settings for the lowest noise results.
I know this has been quite long, but I hope it is helpful. Happy listening!
This seems to be the specification for the S/PDIF and AES/EBU output…
Why would they restrict the I²S output when they support 768kHz PCM and DSD 11.2MHz signals?
Okay… here’s the answer to how DSD (DoP and native) is handled in the M1T question:
So I’m presuming the I²S output supports native DSD256 output. Which can be obtained through the Up-sampling algorithms applied to PCM files and output via UPnP in Audirvana.
No sweat… always tricky with some of these Chinese overviews and documentation… 
You know, after listening to Ted Smith about the number of taps used in sample-rate conversion, he made specific comments about the size of the registers (buffers) needed for 11.2Mhz modulation, where if not adequately proportioned or allocated, buffer under-runs and over-runs will create more noise and the clock-timing must be very tightly controlled… I’ve always understood this, but he reinforces the issues of working at such high sample-rates (large files).
Jud, thanks so much ! Now I’m a bit confused (!!!) but I have many elements to consider and above all I must study a lot, it is challenging, but I am ready.
