That would also presume you can get the datasheet level of performance from LT3045 by wiring it up haphazardly. Take a look at the evaluation board manual to see what they did with the layout to get those numbers. Hopefully the layout is done carefully and your noise is differential mode and not common mode.
You mean like AES67?
Last edited:
Yeah, I have not looked deeply into what these apps / devices are using.
I have worked on some products where we had to create a custom protocol on top of UDP to minimize latency. TCP isn't always suitable for real-time latency sensitive data like VOIP or video calls because it's guaranteed delivery. Retransmissions, large window sizes, and Nagle's algorithm make for a laggy experience if there is any packet loss.
Even if these things were streaming via UDP with nothing on top, I've heard the impact of a missing datagram and out of order datagrams in an audio stream. Hard to mistake the garbage and pops / clicks.
If these nuts were really serious, they would use 10GbE switches with SFP+ ports and make an all fiber system.
I was wondering when someone would mention this.
But then does the extra bandwidth from the OM4 cable provide a better dynamic range over OM3? Come on guys, nobody's saying that data traveling on the network gets lost or corrupted and that is why there are audible differences.
Many of us are in fact IT professionals. We know all too well how all this works, L2 through L7.
Still, different ethernet cables somehow sound different in the end. It is obvious that it is not due to packet loss or (network) jitter.
But audio jitter is in fact influenced by electrical noise, among other things. And this noise does eventually get coupled into our dacs. Remember that galvanic isolation essentially means DC isolation, not AC. Depending on the bandwidth of the isolation transformers, all kinds (and spectrum) of noise easily gets through. This would explain why some people claim that GbE sounds worse than Fast Ethernet. GbE transformer bandwidth (at least 80MHz) >> FE transformer bandwidth (at least 25MHz).
Think about it.
Many of us are in fact IT professionals. We know all too well how all this works, L2 through L7.
Still, different ethernet cables somehow sound different in the end. It is obvious that it is not due to packet loss or (network) jitter.
But audio jitter is in fact influenced by electrical noise, among other things. And this noise does eventually get coupled into our dacs. Remember that galvanic isolation essentially means DC isolation, not AC. Depending on the bandwidth of the isolation transformers, all kinds (and spectrum) of noise easily gets through. This would explain why some people claim that GbE sounds worse than Fast Ethernet. GbE transformer bandwidth (at least 80MHz) >> FE transformer bandwidth (at least 25MHz).
Think about it.
I was wondering when someone would mention this.
I have half a mind to design a matching set of RPi & Ethernet switch that get connected by OM3/4 fiber. Just to see if different fibers sound differently.
(hint: I honestly see no reason why they should, since they provide both true galvanic isolation and isolation from electrical noise)
But then different SFPs might make for different sound. That would be a major bummer.. Especially if an expensive Cisco SFP sounded a lot better that an el-cheapo part.
What's the bandwidth of an RCA or XLR interconnect without any transformer? I assume you connect those to your system at some point.
The theory that the difference in bandwidth and interwinding capacitance between two different types of pulse transformers is creating audible differences needs some measurements, to say the least.
If this really keeps you up at night, the great thing about Ethernet is that you can go fiber.
That would make sense and might be the case for my home-brew USB receivers and dacs.. But then these audible differences have also been observed in very expensive dacs, many of them considered to be very well designed.
But like I said, "considered to be".
I'm happy to report that it does not keep me up at night, since I enjoy building stuff more that I enjoy using it.. In fact I do most of my music listening either in my car, in my lab (through a disgracefully cheap system) or on my PC listening through JBL LSR305s. But when I eventually sit down to listen to music on my main system, I remember why I go through all of this. But it's just not practical to do all of my listening there.
But like you said, fiber is tempting.. see my post just above yours.
Nobody seem to say anything about cause.
Yes - use a few meter of toslink opto and move your DAC away form the rest of the equipment and you are good to go. Using toslink, you need a DAC that does buffering/reclocking.
//
Yes - use a few meter of toslink opto and move your DAC away form the rest of the equipment and you are good to go. Using toslink, you need a DAC that does buffering/reclocking.
//
The cause is the same one that causes an RPi with a cheap SMPS to sound like crap compared to how it sounds with a good quality (linear or otherwise) power supply.
Toslink is fine, as long as you buffer and reclock, so you'll need an FPGA. Not many people can dabble with FPGAs (yet), so not worth much for DIY. (Ian's and Allo's gear excluded, but I don't consider that to be "hardcore diy" since it's essentially a finished product).
Toslink is fine, as long as you buffer and reclock, so you'll need an FPGA. Not many people can dabble with FPGAs (yet), so not worth much for DIY. (Ian's and Allo's gear excluded, but I don't consider that to be "hardcore diy" since it's essentially a finished product).
Audio is sent the same way as any other data sent across an IP network and encapsulated the same way using the same defined international standards, so how can the physical interface determine whether its audio data or data belonging to a website ?
How is it even possible for the isolation transformer to influence only audio encapsulated in data packets but untouch all other types of encapsulated data ?
The balanced isolation transformer along with an ethernet cable forms a transmission line that is resistant to noise.
Last edited:
Why would it presume this? And is it even important?
Just groundless speculation adding noise and garbage to the thread.
The physical interface is working just fine. It's doing what it was designed to do.
Unfortunately, a dac is not immune to electrical noise, like a computer would be.
The transmission line is resistant to common mode noise. The noise we are talking about originates in the equipment, and is not necessarily common mode.
Yes, but also a terrific source of RFI, plus pollution and modulation of power rails due to relatively high current demands.
Plus OS overheads. Easy to hear the improvement in windows once the wifi service is stopped.
- Status
- Not open for further replies.