- Thread Author
- #1
first off, i am not an electrical engineer nor a computer scientist and would greatly welcome any corrections and/or further insights into this topic.
______________________
in the recent discussions here about audiophile networking equipment in general and ethernet switches in particular, their ability to reduce phase noise has been of particular interest. not having run across that term before in the audio world and wondering if it is something that i might need to address, i did some research into the subject...
if my understanding of the subject is correct, it turns out that with respect to computer networking, phase noise results from an imperfect oscillator [clock] used to time the transmission of data over the network. imperfect clocks cause random fluctuations in the cycle of a waveform.
these fluctuations can be measured as the difference in frequency (cycles/sec) of the imperfectly generated cycle from the theoretically perfect cycle. here, fluctuations are measured in the frequency-domain which quantify how much of the actual signal lies within the theoretically perfect frequency. this is the measure of clock phase noise.
these fluctuations can also be measured as the difference in the time (sec/cycle) necessary to receive one full cycle from the imperfectly generated cycle from that of the theoretically perfect cycle. here, the fluctuations are measured in the time-domain which quantifies clock jitter.
thus, phase noise and jitter are the exact same thing being measured in two different ways. in fact, phase noise and jitter are transformations of each other (via the fourier transformation which decomposes a function of time such as a signal into its constituent frequencies).
...so, returning to the discussion of audiophile networking devices several things stand out here:
the purpose of clocks in networking devices such as routers, switches, FMCs, etc. is to time the transmission of data over the network and ensure that it travels at the appropriate speed such as 10/100/1000 mbps. so, as long as the clock in the device operates within networking standards, thereby, ensuring interoperability and efficient network data transmission, its job is done — period. whether the device is a $25 consumer grade switch or a $1,900 tricked-out audiophile switch is irrelevant to network data transmission — again, provided each operates within standards.
in the paradigm of an asynchronous DAC receiving a usb or ethernet signal, the purpose of networking devices and their clocks is simply to transmit data over the network to the DAC. once the DAC has received the data it is then the clock within the DAC which is tasked with the timing the digital audio stream for analog conversion. the clock in a network switch has absolutely nothing to do with the timing of the audio stream received by the conversion chip within the DAC and, thus, nothing to do with system sound quality.
to the extent audiophile networking devices have designs and/or parts which reduce electrical noise and any environmental noise such as RFI and EMI, these devices can then have a material impact on system sound quality. however, this is a completely different subject than clock noise.
so… why all the recent buzz around network device generated phase noise?
______________________
in the recent discussions here about audiophile networking equipment in general and ethernet switches in particular, their ability to reduce phase noise has been of particular interest. not having run across that term before in the audio world and wondering if it is something that i might need to address, i did some research into the subject...
if my understanding of the subject is correct, it turns out that with respect to computer networking, phase noise results from an imperfect oscillator [clock] used to time the transmission of data over the network. imperfect clocks cause random fluctuations in the cycle of a waveform.
these fluctuations can be measured as the difference in frequency (cycles/sec) of the imperfectly generated cycle from the theoretically perfect cycle. here, fluctuations are measured in the frequency-domain which quantify how much of the actual signal lies within the theoretically perfect frequency. this is the measure of clock phase noise.
these fluctuations can also be measured as the difference in the time (sec/cycle) necessary to receive one full cycle from the imperfectly generated cycle from that of the theoretically perfect cycle. here, the fluctuations are measured in the time-domain which quantifies clock jitter.
thus, phase noise and jitter are the exact same thing being measured in two different ways. in fact, phase noise and jitter are transformations of each other (via the fourier transformation which decomposes a function of time such as a signal into its constituent frequencies).
...so, returning to the discussion of audiophile networking devices several things stand out here:
the purpose of clocks in networking devices such as routers, switches, FMCs, etc. is to time the transmission of data over the network and ensure that it travels at the appropriate speed such as 10/100/1000 mbps. so, as long as the clock in the device operates within networking standards, thereby, ensuring interoperability and efficient network data transmission, its job is done — period. whether the device is a $25 consumer grade switch or a $1,900 tricked-out audiophile switch is irrelevant to network data transmission — again, provided each operates within standards.
in the paradigm of an asynchronous DAC receiving a usb or ethernet signal, the purpose of networking devices and their clocks is simply to transmit data over the network to the DAC. once the DAC has received the data it is then the clock within the DAC which is tasked with the timing the digital audio stream for analog conversion. the clock in a network switch has absolutely nothing to do with the timing of the audio stream received by the conversion chip within the DAC and, thus, nothing to do with system sound quality.
to the extent audiophile networking devices have designs and/or parts which reduce electrical noise and any environmental noise such as RFI and EMI, these devices can then have a material impact on system sound quality. however, this is a completely different subject than clock noise.
so… why all the recent buzz around network device generated phase noise?
