Help building Rasberry Pi4 with Ropieee as USB Roon Bridge

One advantage is that AES is meant to be used with longer interconnects in noisy environments.

Yes, that’s why I asked about the advantages at the DAC. That’s usually a short connection that is not subject to significant noise pickup.

I don’t think there’s much advantage, except in rare edge cases. If you’ve got a bad USB handler in your DAC, AES might be useful to solve ground loop problems. AES is an older digital signalling protocol, and it’s source-clocked. USB Audio class 2 offers a lot of things you can’t get with AES, mainly in the realm of improved control over the DAC over the USB channel. I’d prefer USB over AES, and AES over coax S/PDIF, myself.

1 Like

Depends on the DAC but very few DACs have “good” USB inputs. AES is more consistently a very good input on all DACs that have it.

Really? What an interesting claim! What are the numbers? Who did the counting? Which are the good ones?

1 Like

You can go to your favorite search engine and type the name of the dac and " best input" or " best input for sound quality" or something like that. What you’ll discover is across a lot of DACs, I’d even say the majority of DACs, USB is not the recommended input to use for best sound quality.

What you want to find is a DAC that is immune to the USB source. That is, find a DAC where no one says buying a “better” streamer influences the sound. Find a DAC where people are not running out to replace a standard Rasp Pi USB with something like the ​PI2AES or buying the Matrix Audio X-SPDIF. These things exist because the SPDIF input on the DAC sounds better than the USB.

Look, you’re the guy who said it. Don’t tell me to look at a miscellaneous collection of search results with some unknown algorithm surfacing random posts by various dogs on the Internet. Tell me what the numbers are, or admit you don’t know. And if you don’t know, why post in the first place?

Maybe. Or maybe they exist because sufferers from audiophilia nervosa buy them because they’re anxious types and they listen to outmoded audio memes because they just don’t know better.

4 Likes

It’s actually an interesting claim. Not sure anyone tests these things properly, but some variant of it could be true. Here’s the claim:

very few DACs have “good” USB inputs.

Not “few” or “relatively few”, but “very few”. What would that be? One in ten, maybe? A substantial minority of all DACs, in any case.

What can we reason from? There are some sources of presumably solid information. The compact disc format, the first readily available source of digital music, came out in 1982. The AES3 standard came out in 1985. The S/PDIF standard, a variant of AES3 adapted for consumer use with cheaper connectors on the cable ends, came out in 1985 as well, apparently. USB came out eleven years later, in 1996. And USB Audio Class 1 wasn’t finalized until two years after that, in 1998. So any DAC produced before 1998, including all the DACs in the initial round of CD players, had really poor USB inputs, because it didn’t exist yet. So we can count those in the “not good” input category.

Additionally, the S/PDIF and AES3 protocols had 13 years “head start” on connection engineering. So while DAC developers puzzled over the new USB spec, S/PDIF was hardening into an audio community standard. Probably few of the DACs from the 1998-2003 period had USB inputs that bested their by-now-industry-standard S/PDIF inputs.

But the USB standard wasn’t sitting still, either. In 2006, UAC 2 was published. Apple and the Linux community immediately started adding drivers for it to their platforms, but Microsoft dragged its feet. Drivers for UAC2 weren’t added to Windows until 2017, with Windows 10. So most DAC developers with USB inputs could sit back and figure out how to implement UAC 1, and they got to be pretty good at it. Inevitably, a standard way of handling it was packaged onto a chip, and debugged until it was solid. So you’d expect most DACs from, say, 2005 to the present, to have pretty good USB inputs.

But that’s reckoning without boutique audio shops. Some of the stuff that comes out of those shops seems to be really badly engineered in one area and really wonderfully engineered in another area. Limited experience by the designers, I expect. And sometimes the area that falls down would be the USB input, I imagine, because it is complicated, and it’s not what these “lone designer” shops are really interested in.

So my guess, and it’s purely an educated guess, would be that the majority of DACs sold in, say, the last ten years, have good USB inputs. The state of the practice is there. But older DACs existed, and DACs made before USB existed, and DACs with no USB inputs still exist, and some modern DACs will have bad USB inputs. And you add those last four categories up, it may vastly outweigh the DACs with “good” USB inputs. Or maybe not.

[An aside: The problem with counting posts on the Internet is that complaints about someone’s obsolete DAC made 15 years ago are still going to show up. And complaints about poorly engineered boutique DACs that cost an arm and a leg are still going to show up, and it may be that people complain more about something they spent a lot of money on that doesn’t work right.]

6 Likes

My Benchmark DAC1 Pre is limited to a 96khz sampling rate to its USB input. The sp/dif inputs–1 optical, 3 coax–all do 192. I think that was true of most DACs from that era. Often to do more than that would require drivers. We can debate whether 96 v.s 192 is actually audible, but those are the numbers.

I haven’t looked at DACs in a while (I currently use an Oppo UDP-201 with its own DAC for Roon and physical media). Benchmark has newer ones, as does everyone else. But for a long time USB was sample-rate limited.

Of course, the issue with pre-RPi4 was that the USB port was shared and tended to sound crappy, but not the DAC’s fault.

There should be enough bandwidth for audio on the USB bus, even at 192/24.

There was. People ran in trouble at DXD(+)/DSD resolutions though.

It also didn’t help that early device drivers were poorly optimised – things improved over time.

This is the USB Audio Class 1 (1998) versus Class 2 (2006) issue I mentioned. Class 1 was limited to 96kHz, but Microsoft didn’t include Windows drivers for Class 2 (much more functional, and supporting 24bit/192kHz) until 2017 in Windows 10. So DAC manufacturers who wanted to include Class 2 support often wound up having to provide their own proprietary drivers for Windows. Many decided to just stick with Class 1.

So one thing to look for in a DAC is whether it supports USB Audio Class 2.

There’s also a Class 3 (2016), which includes support for things like hotword detection, but I don’t know if it provides any significant hi-fi improvements.

Yes, the Raspberry Pi 3 used a shared bus for Ethernet and USB, which led to problems. Fixed in the Pi 4, but even the Pi 3 is OK if you don’t use Ethernet.

RPi4 is now connected to the CXA61 and after a short listen I’m well pleased. Just as good as direct connection to the core. Now I have the flexibility to have the core back in the office and controlled by remote on ipad/iphone OR when I have friends over for a ‘pick-a-track’ session and I want the ease of using a computer it can sit on the coffee table.

One last noobie question: Do you leave your RPi4 on or do you shut it down? Is there a point to scheduling a daily or weekly reboot?

Many thanks folks!

I leave it on 24x7, and have it set up to reboot every night at 3:04 am. But you don’t need the reboot.

1 Like

Check out Archimago’s blog post on building a raspi streamer with display. Linked here. Same instructions apply to a raspi4. I have a few running throughout the house. I wouldn’t get worked up about any specific cases. I have one in the basic Raspberry Pi plastic case I bought through Microcenter. They don’t get too hot and don’t need fan or heatsink. Yes, a case with a heat sink will run cooler but no difference in performance. The metal cases will impact wifi range but if you’re going to use wifi then you’ll probably need a USB wifi antenna instead of relying on the on-board.

On Archimago’s site you can also find measured results on how the streamer performs and also on comparison of USB to other methods of connecting to a DAC. USB has been demonstrated to be the best method to connect to a contemporary DAC.

Building a raspi streamer is a very good choice. They’re inexpensive, reliable and perform as good or better than any of the commercial products. Ropieee works really well. You’ll find support on some Ropieee threads here in the Roon community. If you go that route then consider donating on the Ropieee site if you value that software.

1 Like

I should add to this that when selecting the Raspi, the smaller RAM configurations seem to run cooler. I have 4 raspi devices running and their respective operating temps seem to correlate with memory size. The smallest ram configurations are plenty to run as a streamer.

I’ll keep that in mind when I put together an Rpi for upstairs. I ended up with an 8gb Rpi4 simply because that’s the only model available in New Zealand at present. The stores said they’re having real issues with supply at the moment. I’ve been streaming through USB to my main system for a couple of weeks now and it all works a treat.

1 Like

I just put myself on a waiting list for a Rpi4 2GB and 4GB.

If the intention is to run RopieeXL as Roon endpoint plus occasionally other streaming services, maybe also adding a HAT later on, is 2GB still safe for future proofing or it’s better to get 4GB?

If it’s only to run RopieeXl 2GB will easily do it. But if you want to add other streaming services, add a hat or potentially use it for other purposes down the track I’d go for the 4GB.

1 Like

2 posts were split to a new topic: Help with M-DAC Settings required