USB interfaces and clocks

[Nick]

I am not saying that any of it is a issue or worth worrying about. I was just pointing out that if you believe that something measured in ps matters when it comes to low speed data like audio, then there are more places to point that belief at than whatever you are being sold by the invocation of that belief.

[Tony Moore]

I used to get very focused on I2S jitter and clocks/reclocking and I have tried a couple of different implementations over the years but now I realize that it's less about jitter as such and more about noise and grounding. Introducing more electronics into the chain sometimes just makes matters worse, or at least different enough that it muddies the waters and you're left not knowing what change had what effect on the sound.

Good power supplies, short signals runs and good matching of the signals to not ring or other bad effects and that's most of the battle. If you need to send I2S any distance not on short tracks on a pcb then LVDS works a treat. I just tried it between my Transporter and DAC and I'd say if there's any degradation in signal I cannot hear it in the sound. I used ready made boards for 4 channel send and receive but if I were doing it again I'd probably use the TI chips and do my own pcb.

[vinylnvalves]

LVDS cable - same as ribbon cable, or is it something different. I am using some ribbon cable for raspberry pi stuff with the pin connectors on each end (7 cm)

[Nick]

I am assuming Tony means LVDS as in the "low-voltage differential signaling" interface. So a transmitter/receiver pair with a pair between them.

https://en.wikipedia.org/wiki/Low-volta ... _signaling

[Tony Moore]

I am assuming Tony means LVDS as in the "low-voltage differential signaling" interface. So a transmitter/receiver pair with a pair between them.

https://en.wikipedia.org/wiki/Low-volta ... _signaling

Yes, exactly. I used these modules as it was simpler at the time:

https://magnahifi.com/audio-gd-hdmi-lvd ... ut-module/
https://magnahifi.com/audio-gd-hdmi-lvd ... ut-output/

Cable is standard HDMI. The chips in these things are made for very high data rates for video so I2S is not breaking a sweat.

[vinylnvalves]

Okay think i am understanding now, so you basically put the USB converter in the streamer, convert to HDMI and feed those twisted pairs to the DAC via an HDMI signal. Putting 2 more devices in the path though I see a benefit if we are saying the USB cable is the weak link in the digital chain.

[Tony Moore]

Well, in my case I don't have any USB as I am using Squeezeboxes, which have I2S available. However, if you're using a USB to I2S converter then yes, then that is the source of the clock(s) and data for I2S so its from there it is the "weak link" as they are logic level signals, not intended to be transferred very far.

I don't know much about how USB transfers audio streams but I'd imagine that it is far more resilient than logic level signals. The biggest hurdle would be keeping out HF noise transferred into the I2S signals from it and the PC/streamer.

[Nick]

Okay think i am understanding now, so you basically put the USB converter in the streamer, convert to HDMI and feed those twisted pairs to the DAC via an HDMI signal. Putting 2 more devices in the path though I see a benefit if we are saying the USB cable is the weak link in the digital chain.

Just to be sure there is no confusion, just because the modules Tony linked to uses a HDMI plug and socket, its not HDMI, its just using the fact that a HDMI interface contains four twisted pairs. so its a convenient way of getting what we need. You would only do that is you were going any distance between boxes. For point to point its not adding anything extra. But for external use it does as logic level i2s is not designed to be used in such a way.

Tony is talking about: i2s (from whatever) -> LVDT -> HDMI Cable -> LVDT -> i2s (to whatever)

I don't know much about how USB transfers audio streams but I'd imagine that it is far more resilient than logic level signals.

Only in that the protocol and hardware definition is designed for long (few Mtr) distance use. There is no error correction or retry for audio use, though there are checksums. The assumption is that the cost of loosing a packet is not that great, it will only be a click in the audio stream.

[Ray P]

You can do LVDS using RJ45 connectors and ethernet cable - that's what Twisted Pear's Teleporter modules used - IIRC Twisted Pear tested it to close on 100metres without any issues.

[Nick]

You can do LVDS using RJ45 connectors and ethernet cable - that's what Twisted Pear's Teleporter modules used - IIRC Twisted Pear tested it to close on 100metres without any issues.

As I said "its just using the fact that a HDMI interface contains four twisted pairs", fully wired cat5 and up does as well.

[Nick]

HDMI does have the possibility of using the extra wires to remote power the receiver via the connection.

To add to that, looking at a LVDS receiver, you can either have a common ground, or need to use cap or transformer isolation. Once you do that, you start having to use a encoding that ensures DC balance. How did Twisted Pair get around that problem with rj45 Ray?

[Ray P]

How did Twisted Pair get around that problem with rj45 Ray?

To be honest Nick I only had a quick superficial look at the Teleporters - didn't actually progress with LVDS as I went with the local Beaglebone Black renderer route so my main DSD DACs all connect to HQ Player over the home network.

There may some information here

https://www.diyaudio.com/community/thre ... er.201106/

it was the RJ45 approach that caught my attention when I looked, mainly because it seemed to be a much more flexible solution than HDMI, both literally and figuratively.

[Nick]

I read the first page and then lost the will.

[Ray P]

I read the first page and then lost the will.

Yeah, I didn't get any further.