DACs: Improve your USB (break out PCI USB card with Unison)

I'm glad you're covering this topic. I tried a similar experiment over a year ago. I've been using a 2008 Mac Pro which has the best-sounding USB ports out of any machine I've tried. makes sense as this machine was a studio standard during it's production life.

I purchased a Sonnet card for cheap off ebay: https://www.sonnettech.com/product/legacyproducts/allegro_express_usb.html
and installed it in the Mac Pro. the sound was noticeably worse than the motherboard ports. In this case I think the dedicated 5V was not worth the trade-off of a longer internal signal path through the PCI bus, and the regular ports on the Mac were just better.

I still have the card I'll repeat the experiment with a normal Windows PC and also hook up a cheap LPS and report back. It should be noted that PCIe is worse than PCI for audio applications and most modern motherboards do not have legacy PCI slots anymore. This became clear to me in 2018 or so when we were comparing Lynx and RME sound cards with S/PDIF out.

BTW, the 800 euro JCAT 'audiophile' usb card is just a OEM card using the same USB controller as the $20 amazon USB cards, with some fancier regulators and a DC input jack for an external PSU
https://jcat.eu/product/usb-card-xe/

 

Interesting. Do you think the perceived goodness is a result of a dedicated USB bus for the audio output, or that cad specifically? Maybe the improvement comes from the power not being routed through the MB, as it is with the on-board USB ports.

 

It's definitely a crapshoot, some mobo USB ports are better than expansion cards, some are worse, but it's a pretty cheap experiment to try. With cheap SoC Atom mobos I didn't find a generic USB card which sounded better than the best of the on board ports.

Another point to note is that Yggdrasil is very sensitive to HF hash on the ground plane. Putting masking tape over the outer metal shell of the USB B connector breaks the ground connection from the PC to Yggdrasil which I found reduces sibilance and digitus quite noticeably (though no change in focus). If you like the effect you can make up or get made up (eg by Ghent Audio) a cheap cable with the shield disconnected at the USB B end.

 

In my main two channel system I switched from

Windows PC -> Unison USB -> Yggdrasil A2

to

Windows PC -> Pi2AES -> AES -> Yggdrasil A2

This was a step up, especially in terms of clarity and less listening fatique.

But the real game changer for me happened, when I switched the Pi2AES for the SotM sMS-200ultra Neo:

Windows PC -> SotM sMS-200ultra Neo -> Unison USB -> Yggdrasil A2

WOW! Enter depth galore and for the first time I understood the term blackground that I was reading here for several years. Before I had just understood 'blackground' in theory, but had no clue how more or less 'blackground' would sound.

I got the SotM used and I just got it out of curiosity. I actually wondered if I would hear any difference at all.

Boy, was I surprised. You can hear the difference in seconds. No critical listening necessary at all.

A litle anecdote: Gushing I had to show my girfriend the difference. When I switched back to the Pi2AES she was shaking her head, so bad it sounded in direct comparison.

I figure that the difference is bigger with speakers than with headphones (soundstage not headstage).

Many thanks again to @rlow for his tale of three transports. Elevated my music listening experience by several margins!

 

FWIW, that nuts ORFAS 20k USD server from Taiko Audio until recently (now it’s some custom OCXO card) used its on board mobo audio for USB out because the designer claimed it sounded best. Some folks on other forums figured out the dual CPU Xeon ASUS board they were using has a ASMedia 3142 for its controller chip, which you can find in PCIe cards for about as cheap as the Renesas cards. The 800 buck JCAT card uses ASMedia as well, I think (although not the cheaper “Femto” card, it has a Renesas chip).

 

Does anyone know if Unison USB needs 5V on the USB connection for handshake or communication?

Just found this USB card and many raving user reviews.

One can easily switch off power delivery on the USB and let the DAC just use its own PSU. The USB bridge chip from Texas Instruments opens up independent data channel for USB audio stream. It helps signal bypass the USB controller integrated in the motherboard, avoiding the sharing of data channels with other USB devices in the computer.

So if Unison only uses data and not power this might be worth a try. Talking only about self powered Unison DACs, of course.

Above quote from here:

 

Just found the answer by @Garns in a post from 2019:

 

I'm not sure how it's implemented in commercial DACs, but I researched lots of different DIY interfaces recently.

By default they almost exclusively draw power from the USB connection to power the receiver. Then, the decoded digital signal is usually sent through galvanic isolators that are powered by the DACs internal power supply. This keeps power supply and ground noise from the computer out of the DAC.

If the receiver is powered by the DACs internal power supply, that can actually create more problems. This is because you end up tying your DACs power supply ground to the USB data ground. This is bad, no matter how clean the new power line is. The only way to avoid this is to have an entirely separate power supply just for the USB receiver, isolated from the DACs power supply, which is just nuts from a design point of view.

So, I'm not sure what my point here is? Maybe other than cutting the USB power from the computer might not always be the best solution. Especially if your USB interface has galvanic isolation.

 

is this a PI2AES versus SoTM comparison or an Yggdrasil USB versus AES comparison? Does the SoTM have AES output?

And yes, I am not unbiased. But I wonder if the results were fair.

 

No, the SotM only has an USB out.

I don't judge if this is fair. Mind also the huge difference in price. Also the SotM was run by a LPSU and the Pi2AES was not.

So in total one has to say: SotM USB on LPSU vs. Pi2AES AES into Yggdrasil.

I am not trashing the Pi2AES, I was happy with it and did not have any complaints. But I also encourage people to read @rlow 's comparison, take it seriously and experiment themselves. I read his assessment as an understatement now. The difference in my system was huge.

This goes for two channel. Don't know about headphones.

 

My memory is that Unison is actually powered from the Yggdrasil power supply (unlike Gen5 which is bus powered as you suggest above). However it definitely doesn't work without the +5v pin connected (from experiments with masking tape). But as @purr1n says it might only use the +5v for handshaking so be happy for it to be disconnected after a session is established.

 

Interesting but I have always think that this works more or less the same as USB isolators like Intona and similar units.

 

With that many differences it does not seem reasonable to conclude anything other than the entirety of one chain is better than the other entire chain. You could just as easily have concluded that Yggdrasil USB versus Yggdrasil AES is much better.

Not sure why this bugs me so much!

 

I tried that but my only spare PCI-express is right above the GPU which is about the most adverse environment imaginable.

I'm also using nearly all my sata power cables for watercooling so that powerline is dirty as hell which doesn't help.

 

I have tried MANY different approaches. Including the several audiophile USB cards. The best I was able to do was using the ADNACO fiberoptic USB extender system. Of course all of this requires a "B.A.PC" running all the time. I still want to test some more but I never found nirvana and a bigger electric bill... Two Pi's and Two Supplies.....

https://www.adnaco.com/products/s3a/