plissken Posted July 28, 2017 Share Posted July 28, 2017 Here's all anyone need do on their computer: Que up a track over the network. Start playback. Have someone pull the plug and if you still hear audio playing out of buffer but you can't tell when the cable was pulled then don't spend the $$ on this type of product. louisxiawei 1 Link to comment
plissken Posted July 28, 2017 Share Posted July 28, 2017 6 hours ago, Marcin_gps said: Just to clarify: the switch is not our design. It's a D-Link switch with a modified clock and capacitors. It comes from Paul Pang. We may have our proprietary design next year. Best regards, Marcin What does a modified clock do to the data in buffered systems when I can pull the Ethernet cable and still have audio for 7-20 seconds on a lot of properly designed pieces of gear? Same question goes for your $400 Intel 520 NIC. Link to comment
plissken Posted July 28, 2017 Share Posted July 28, 2017 30 minutes ago, davide256 said: On switches I do find that LRC Ethernet cable interaction and power supply noise interactions exist. And there are some older switches (FE only ports) that don't run fast enough for audio. Using fiber optic connection was sufficient, no fancy switch required I'm dubious to any system that would exhibit perceptible noise from an Ethernet cable. This should be doable without any said noise. I've managed this dozens of times. Link to comment
plissken Posted July 28, 2017 Share Posted July 28, 2017 4 hours ago, davide256 said: I've had hum from a wallwart powering a switch cause downstream renderer interference in a wired Ethernet solution, very noticeable when playing music... switched to LPS power and the problem went away. Wired Ethernet sounds strained compared to using fiber optics between switch and network streamer, an effect similar to using a marginal power amp with speakers.. 99% of the time its okay but that repeatable 1% makes you want the next step up. And my experience with buffers has always been turn them to lowest setting compatible with your hardware as they degrade the sound. I'll try pulling the Ethernet connection on my microRendu some time, I'm curious how much it buffers. It may be that I use switches with 3 prong power. Who knows. If I have a repeatable 1%... Well that is easily solved: I don't repeat it. On the buffer thing about minimal buffers = better sound I ain't buying it. There is no difference on 3 seconds vs buffering the entire track. Arguably buffering the entire track should be the most optimal. Link to comment
plissken Posted July 29, 2017 Share Posted July 29, 2017 5 hours ago, Marcin_gps said: Same thing it does for USB Audio. It improves sound quality. If I setup J River and setup a 1GB RAM buffer take a track make copy A and copy B, plug in your Switch, queue up A track and let it start playing, then swap out the switch with a normal D-Link, queue up the B track you are telling me that there is going to be an audible improvement with A track vs B track while there is no switch connected? Link to comment
plissken Posted July 30, 2017 Share Posted July 30, 2017 1 hour ago, Marcin_gps said: If you don't hear the difference, good for you! I'm wanting some insight into your thought process for how tighter tolerance 25Mhz clock that PHY's used to sync up each other so they can transmit makes through the input and output buffers on the Intel PHY, then the buffer on the PCIe bus, then the DMA set aside on the RAM bus, then the buffer set aside by the CPU then the buffer set aside by the playback application, then the buffer set aside by the USB PHY, then the buffer on the DAC where the data finally has an audio clock placed on it. So my question is: Buffers can be RAM or Disk. Volatile vs Non-volatile. So are you maintaining that your tighter tolerance clock some how makes MD5 hash identical data that has been copied 5 or 6 times, crossing several clock domain boundaries via FIFO buffering, sound different? Do you have an instrumented measurements that you have done along the way while you were developing modifications for the Intel 520 card? Would you sit for a blind session where I supplied a stock i520 and placed both into a network team using Intel's utility allowing for your modified card and a stock card to have the cables swapped out at will with no interruption in playback? We would run your modifed i520 on your modified D-Link. I'll bring a Cisco Switch. What airport is closest to you assuming you are in the U.S. (Sorry don't know where you hail from). Link to comment
plissken Posted July 31, 2017 Share Posted July 31, 2017 7 hours ago, Speed Racer said: Hold on a second. You folks do realize the clock is in the Ethernet packets themselves, right? If you use UTP, you are getting galvanic isolation due to the transformers used on each end. The clock is a 25Mhz oscillator on the PCB that allows the PHY's to sync up and start transmitting packets. Link to comment
plissken Posted July 31, 2017 Share Posted July 31, 2017 7 hours ago, Marcin_gps said: It's intel350. And we're not modifying - it's a proprietary low-noise design. Years of listening experience and intuition is where I find ideas for improvements. I'm in Wroclaw, Poland. I'm happy to participate in a blind A/B session to prove what I already know. Best regards, Marcin You understand that Ethernet isn't real-time and that with today's systems a serious audiophile can setup a system (like I have) to cache entire tracks to play back out of RAM leaving the NIC in an extremely minimal use state. I shot a video of this on a $200 laptop playing back 24/192 over 54MBit wireless showing the transfer going down to 0Kbps. With a 1Gbps NIC and hitting a wire speed of 107MB / second I could queue up an entire 16/44.1 ALBUM in 7 seconds. 24/192 in about 30 seconds. What good is this going to do me again? Link to comment
plissken Posted July 31, 2017 Share Posted July 31, 2017 27 minutes ago, Marcin_gps said: I think further discussion is futile. EOT for me Regards, Marcin What discussion besides your disheveling and evasion? I asked a straight forward question and you couldn't answer it. So how does a switch port or Ethernet card, when not in use, improve SQ? Just trying to understand what mechanisms the switch or NIC applies to data that was buffered 5/6 times and anywhere from 6 seconds to ~50 minutes ago. Link to comment
plissken Posted January 1, 2018 Share Posted January 1, 2018 You could get a switch like the Cisco SG 200-8. It has a 12v barrel connector and just drive it with a linear PSU. Link to comment
plissken Posted January 2, 2018 Share Posted January 2, 2018 If your router is bridged what is your firewall? What router do you currently have? Typically the cable modem (or DSL) is either in bridged mode or routed mode. If in routed mode the cable or DSL modem provides the firewall and IPv4 NAT. If in bridged mode you have something operating as your firewall/router. Then from there you would cable to something like the SG 200-8 and then hang WAP, NAS, TimeCapsule, etc directly off of it. The switch becomes the hub and everything else a spoke. Link to comment
plissken Posted January 2, 2018 Share Posted January 2, 2018 Here is a 3 prong power, TP Link Router with 4 GBe Ports. Link to comment
plissken Posted January 2, 2018 Share Posted January 2, 2018 If you're out of ports on the TimeCapsule then you will simply toss in another switch. Wire your audio device to the TimeCapsule and then other connections the switch. Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now