If I only use I2S, should I still upgrade to DSD MK2?

I want to be sure that this would be a substantial upgrade. Recently, I set up an Optical Rendu to Ultra Digital, connected to the I2S input. That was a huge, obvious improvement over my NUC connected over USB, as I think i previously posted. Both optical isolation and I2S combined is a home run with the MK1. Am I really going to get a noticeable improvement with that setup on the MK2? If so, please explain why. I’m not in the mood for yet another upgrade if I can be comfortable that I’m not missing much, if anything.

I think you’d likely find that running USB straight to the MK2 would be your best bet, and a considerable improvement.

Yes, the MK2 is that good.

I have been using I2S link only with DS and with MK2 until last week, and the improvement has been amazing from DS to MK2. Now I am using a USB.

What I find is the I2S and USB inputs sounded different. Simply put the I2S has a more solid-state sound whereas USB has a more tube sound. On MK2 the different cables make more sonic difference than DS. So, you want to use the best cable no matter which link you prefer.


The very fact that you describe such a difference between I2S and USB on the MK2 is somewhat alarming and I think merits a technical explanation.

This could simply be explained because for many, they removed an USB to I²S DDC from the signal path.

1 Like

I am the last person who can explain a technical term. :laughing:

The experience was based on five USB cables and three AQ HDMI cables that I tried on MK2, but there are other users in the MK2 thread that had similar experiences.

I may have been the first to report USB was better.

Perhaps a streamer with native I²S out would sound the same as USB.


I think you nailed it Vince, the Matrix does harden the sound a little with adding a bit of distortion.

1 Like

Correct me if I’m missing something, but you said “Recently, I set up an Optical Rendu to Ultra Digital, connected to the I2S input. That was a huge, obvious improvement over my NUC connected over USB”. What is surprising about USB being better (or equal to) I2S on the Mk II? We changed the USB implementation to use a much quieter, lower powered, and more featured chip as well as adding galvanic isolation so USB should have gotten proportionally better on the Mk II compared to “just” adding galvanic isolation to the I2S inputs. FWIW as far as the FPGA and the rest of the dac are concerned USB is I2S, but its I2S doesn’t have to run thru a long cable outside of the DAC.


I only have my own experience to reference.
I use a Lumin U1 Mini to a Matrix and to a DS Mk1 via i2S.

I went the transformer upgrade and VOCM upgrade route.

It is like a new DAC. Sounds fantastic.

Some say it is as good as the stock Ds Mk2 when they do a side by side.
I cannot confirm that but am very happy.

Thanks, Ted. I had the understanding, perhaps mistaken, that at least on the MK1, that I2S was the best way to go because it was a direct connection into the architecture and bypassed longer and noisier routes taken by all the other connection types.

The theoretical advantage of having a separate clock with I2S compared to TOSLink, AES3 or S/PDIF doesn’t matter with the DS or the DS Mk II. Neither looks at the edges of the clocks at all. Input jitter doesn’t affect the sound out of the DS or DS Mk II.

On the Mk I you can get any input to sound great if you work at it hard enough. The biggest differences between inputs is caused by factors external to the DAC, ground loops, shielding, electrical noise etc. I2S uses HDMI cables which are much better designed for the purpose and have better grounding than other inputs, except TOSLink. The best input (ignoring bandwidth) on the Mk I is definitely TOSLink - it’s the most direct input - a single wire from a trivial receiver to the FPGA and it doesn’t have grounding issues. USB is the most convenient input, higher bandwidth that the non I2S inputs, but it has more noise which is hard to get around without explicit galvanic isolation. With galvanic isolation all of the inputs are on much closer footing.


Agreed. A streamer with galvanic isolation on the USB output like the Signature Rendu is fantastic on both the MK1 and MK2.


Don’t you need additional work to strip I2S signal from USB protocol? Isn’t that going to emit extra digital noise inside the DS MK2? It is done outside of DS MK2 if we use I2S input, so lesser noise w/galvanic isolation?

The USB audio receiver chip is on the outside of the isolation barrier in the Mk II. There isn’t any I2S involved with USB, but the USB receiver can output the audio in multiple ways. We chose I2S for the best bandwidth over the isolation barrier.


To do DSD 256, must I still use I2S?

No. The USB input also supports DSD 256

1 Like