What is the best method to get digital audio from your PC to your DAC
May 13, 2020 at 1:48 PM Thread Starter Post #1 of 101
Jun 20, 2019
United Kingdom
So, I should shortly be receiving a Holo May dac, and i'm a bit confused about some of the things that people have said about it/the spring online.

Typically, I'd just connect my DAC to my PC via a decent USB cable (Just a Supra one, not super cheap, but nothing snakeoil)
A lot of people have said that the holo dacs sound better when fed I2S input, rather than USB. And i'm wondering if there could be any truth to this.

As far as I can see there are three options:

Option 1: Connect DAC to PC straight USB
Option 2: Connect a good SPDIF bridge like the Singxer SU-2 or Matrix x-SPDIF 2 via USB, then the dac to the bridge via I2S
Option 3: Use a Pi2AES and connect the dac to the I2S output of that, so that in theory you avoid anything to do with USB altogether.

People have said that option 2 sounds better than option 1, but the only reason I can think of for this to be true is if the USB module or clock in your dac was worse than the bridge, which for some dacs may be true. But the Holo May has a much better USB module so I can't imagine this would be true. (The early spring dacs did have an iffy USB module which was later upgraded so this could be the reason for the subjective opinions about USB vs I2S online)

Am I missing something? Is there some other benefit to using a bridge vs just connecting the DAC via USB?

With Option 3, I'm guessing the benefit here is that you have nothing to do with USB at all. the Pi communicates via I2S to the pi2AES board, and then you output I2S from that whilst utilising the high quality clocks, so there should be no issue at all. With the added benefit of being completely isolated from any activity on your PC. So perhaps a roon bridge or HQPlayer NAA with a pi2aes setup could be the "ideal" option?

If anyone could shed some light on why Options 1/2/3 could be better than the others I'd really appreciate it.
May 13, 2020 at 2:15 PM Post #2 of 101
The less pieces of crap in the middle, the better. For DSD and ease of use USB is the best general solution. A lot of misinformed people like to waste money on gadgets. If you look over some of the threads you will see pics and posts that almost feel like a contest to see who can string the most interconnects and boxes between two points. Just defies common sense and engineering practice.

If you have noise, crappy power etc. then optical is fine but "only" good for 24/192.
Last edited:
May 13, 2020 at 2:41 PM Post #3 of 101
Not specific to the Holo May, but I've had good experience when using optical over USB. To be fair, the USB cables in my specific setup have to run through an absolute rats-nest of power cables. Optical also has the added bonus of isolating your components from each-other

Edit: Just looked up the Holo May, surprised that didn't have optical input as an option. Must have been thinking of a different DAC.
Last edited:
May 14, 2020 at 5:45 AM Post #4 of 101
It's a long time since I have posted on head-fi, but I have in the last few years been developing several CATs (i.e. computer audio transports) for use both in my speaker-focused system and my headphone-focused system.

I have found that there is no single answer to the best way to get the signal from a computer to a DAC. It depends in part at least on the quality of the USB input on the DAC.

For example, I tried feeding the USB signal from one of the CATs direct to a borrowed Chord DAVE, and the sound was less good than when I fed it via my Matrix x-SPDIF 2, using the same USB cable (a Tellurium Q Black Diamond one; which I like very much) to the AES digital input on the DAVE via a Nordost Heimdall 2 AES digital lead.

What I have found to be most important in the signal path is the quality of the USB output from the computer. I have recently installed a (recently-released) JCAT XE PCIE card, having previously used a JCAT Femto PCIE card, and found that using both of those with an external linear power supply has made more difference, probably, than anything else I have done in regard to getting the signal from the CAT to the DAC.
May 14, 2020 at 8:38 PM Post #5 of 101
Stick with a USB to DAC connection.
Optical (or coaxial) would be used if you need to get something like headphone surround sound, from the sound card or have a ground loop.
Otherwise just stick to USB.
Jul 5, 2020 at 4:02 PM Post #7 of 101
Jul 6, 2020 at 5:17 AM Post #10 of 101
When I was experimenting with this, I found that direct connection with USB to DAC delivered unacceptable results. I then messaed aroudn with a plethora of conditioners and convertors, all of which delivered incremental improvement, but still not great (and expensive!). I then went back to basics, and invested ina decent USB audio card for the PC ( in my case https://www.sotm-audio.com/sotmwp/english/portfolio-item/tx-usbexp/) - night and day improvement - when connected with a well-made USB cable, this out-performed any amount of inline processing/conversion. So I agree with Omriff on this - the USB card is the biggest bang for buck in terms of SQ vs $
Jul 6, 2020 at 6:20 AM Post #11 of 101
I use option 3. With Pi2AES you get all output be it COAX, optical, AES ans I2S for fraction of price and you keep laptop free of wire and away from your music source.
But it's your personal preference. I work on my laptop and I don't want my laptop to be connected with DAC with a USB cable while I'm working as I move and work at different places since this WFH has started
Also the DAC you have definitely has a say on this. I use NOS DAC from Metrum and Sonnet and they do sound better with AES or I2S as compared to USB.
I don't have any music beyond 192Khz and neither my DAC is capable of decoding anything beyond that so I'm happy to be away from USB.
Last edited:
Jul 6, 2020 at 6:55 AM Post #12 of 101
None of the above
Jul 6, 2020 at 9:53 AM Post #13 of 101
Curious about this myself. I've used USB with what I believe to be fine results. That being said, I just ordered a optical/coaxial board for my Pi setup, and I'll compare.

I thought the benefit of USB is what it allowed your DAC to handle timing, which is likely better equipped to do than your CAS, but I've also read about issues with USB. It would be great to have the definitive word on this.
Jul 7, 2020 at 8:37 AM Post #14 of 101
It would be great to have the definitive word on this.

Here it is :xf_wink:

Legacy protocols like AES/EBU or SPDIF over coax or Toslink use the send rate of the bus to convey the sample rate to the receiver. If the clock of the sender is jittery, you have tons of input jitter at the DAC.
If the designer of the DAC does “nothing” this will result in a jittery performance.
Over the years they invented all kind of tricks like PLL (kind of electronic equivalent of a fly wheel) and probably best, asynchronous sample rate conversion (ASRC). This allows for a free running clock.

USB comes in various flavors of synchronization. Isochronous and adaptive synchronization generates input jitter.
Asynchronous synchronization let de DAC dictated the amount of data send hence the DAC can use a free running clock.
Problem is the electrical noise generated by the PC , it requires galvanic isolation at the side of the DAC. When this is done right, it is probably the best solution.

This of course makes you wonder where ethernet fits in.
Ethernet (the port) is galvanically isolated by design.
It uses large buffers, no DPC latency issues as good happen with USB
Theoretical even better than USB.

In practice Ethernet, AES/EBY, SPDIF, USB, Ethernet, WiFi are protocols.
Big piles of technical papers. Protocols don’t have a sound signature.
It all comes down to the implementation.
This can add distortion.

If the DAC (the chip doing the DA) is sufficiently isolated from the source (noise) and can use a free running clock (ASRC) we might wonder if there is any difference between the various inputs.
Jul 7, 2020 at 11:29 AM Post #15 of 101
Having both looked into this quite a lot, and also trying various products myself recently, I figured I'd post a response to clear things up a bit.

There are two main factors that will affect the quality of the eventual i2s signal processed in the dac:
-Ground plane pollution/power quality.

The latter is a simple matter to resolve. Simply avoid a direct electrical connection to a noisy device such as a PC. If you want to try this yourself and keep things a fair test, just get a $30 raspberry pi3b+, install ropieeeXL and use that. You will likely hear a noticeable improvement not due to any digital change, but simply because the 0v reference ground plane is not being contaminated by the noise from the power hungry PC. Though if using a battery powered laptop this may be less of an issue.
This is the same reason why high quality linear power supplies can bring a benefit to audio equipment.

If you want to step this up, you can get a high quality streaming solution such as a pi2aes, SotM SMS200-ultra, Sonore Ultrarendu, or Lumin U1 for example.

Jitter however, is a bit more of a complex beast and depends on a few things.

- When using USB (with proper drivers), it is asynchronous. Meaning the dac USB module calls for additional data when it needs it, and converts to I2S using the dac's OWN clock. Therefore jitter will be dependant on both the quality of the device clock, and the USB implementation in the dac.

- When using any form of SPDIF (AES, Coax, Optical), the clock signal is encoded within the stream, and so jitter is dependant on the SOURCE (your PC). This is why a high quality digital-to-digital converter such as pi2aes, Singxer SU-6, Matrix X-SPDIF 2 etc will sound better than SPDIF from your PC. Everything is still digital "Bit perfect", but jitter is considerably lower than the output from your PC.
Jitter on isochronous connections like this can also be fought with a high quality PLL and FIFO buffering, such as that found in the holo May. You can see in measurements here that even with high amounts of artificially introduced jitter, the may's PLL when enabled, made the dac almost immune to it. https://www.audiosciencereview.com/...may-probably-the-best-discrete-r2r-dac.10161/
Its worth noting that while optical connections give the benefit of galvanic isolation, they also usually have higher jitter than coax.

- When using I2S from a DDC, again the clock is part of the stream, however i2s uses multiple lines, with the master clock being separate entirely from the data and word clock. Meaning it is in theory better than SPDIF which encodes all of this into one stream and has to be "unpacked" at the other end. Though has the disadvantage of usually only being able to be done over very short runs.

If you want to use the best connection, use the best DDC/clock source you have.
That might well be your dac's internal DDC over USB.
But for $200 you can get a pi+pi2aes which is likely to be better.

However, galvanic isolation will almost always bring a benefit, and you can do that for $30
Last edited:

Users who are viewing this thread