Merging NADAC clocking

[Blizzard]

Does anyone know how the NADAC clocking scheme works? I realize that it sends a grandmaster clock over Ravenna, which can be shared with all units on the network. But is it reclocked using a low phase noise clock located close to the DAC chip?

 

If it just uses the grandmaster over Ravenna, wouldn't that be inferior to using a precision clock 1 inch away from the chip?

 

If the clock out of 1 of the DAC's in a multiple DAC setup is assigned the role of grandmaster, wouldn't that DAC be at a advantage by having the clock closer to the chip?

 

If it does reclock, wouldn't this make it impossible to sync multiple units perfectly packet for packet?

 

It's my understanding that the asynchronous buffering that occurs in the FPGA's of modern USB interface DAC's is the best way to do things. It allows all the zeros and ones to align up perfectly in the FPGA, so they can be reclocked with precision low phase noise clocks inches from the DAC chip. But at the same time it makes it impossible to sync multiple DAC's up packet for packet. Does Merging's implementation of Ravenna also take advantage of this method while at the same time allow packet for packet sync between units?

 

Or do they have a superior way of doing things altogether, while at the same time allowing packet for packet sync between multiple units?

 

The reason I'm asking is because I would like to build an active system using 1 NADAC per speaker. Packet for packet sync between DAC's will be a must.

[astrotoy]

I just emailed Dom Brulhart of Merging to make sure he has seen your post.

 

Larry

[Blizzard]

I just emailed Dom Brulhart of Merging to make sure he has seen your post.

 

Larry

 

Sounds great thanks!

[dbrulhart]

Does anyone know how the NADAC clocking scheme works? I realize that it sends a grandmaster clock over Ravenna, which can be shared with all units on the network. But is it reclocked using a low phase noise clock located close to the DAC chip?

 

If it just uses the grandmaster over Ravenna, wouldn't that be inferior to using a precision clock 1 inch away from the chip?

 

If the clock out of 1 of the DAC's in a multiple DAC setup is assigned the role of grandmaster, wouldn't that DAC be at a advantage by having the clock closer to the chip?

 

If it does reclock, wouldn't this make it impossible to sync multiple units perfectly packet for packet?

 

It's my understanding that the asynchronous buffering that occurs in the FPGA's of modern USB interface DAC's is the best way to do things. It allows all the zeros and ones to align up perfectly in the FPGA, so they can be reclocked with precision low phase noise clocks inches from the DAC chip. But at the same time it makes it impossible to sync multiple DAC's up packet for packet. Does Merging's implementation of Ravenna also take advantage of this method while at the same time allow packet for packet sync between units?

 

Or do they have a superior way of doing things altogether, while at the same time allowing packet for packet sync between multiple units?

 

The reason I'm asking is because I would like to build an active system using 1 NADAC per speaker. Packet for packet sync between DAC's will be a must.

 

 

Hello Blizzard,

 

The clock of slaves NADAC on a Ravenna network is regenerated from its own internal clock through a PLL to synchronize with the clock of the master NADAC over the network, exactly as we would do when slaving to a Wordclock, Video or Audio Input signal, but with an even greater precision as offered by the PTP protocol, actually PTP2 in our case (IEEE 1588-2008).

 

Packets sent from a Ravenna source (also clocked by the PTP protocol) leave the sources at the same time, packets have the same size and contain samples starting with the same SAC, but by essence of Ethernet can reach their respective destination at a different time.

 

The Ravenna/PTP2 protocol ensures that they are properly resynchronized and are output by all destination devices exactly at the same moment, with a jitter of a few nanoseconds, whatever the distance, number of switch/routers and number of destination devices.

 

Ravenna is the perfect solution for your proposed system, one NADAC per speaker, as close as possible to the speaker.

 

Best regards,

 

Dominique

[bmoura]

I just emailed Dom Brulhart of Merging to make sure he has seen your post.

 

Larry

 

I sent it to him last night. Looks like he got it and responded.

[Blizzard]

Hello Blizzard,

 

The clock of slaves NADAC on a Ravenna network is regenerated from its own internal clock through a PLL to synchronize with the clock of the master NADAC over the network, exactly as we would do when slaving to a Wordclock, Video or Audio Input signal, but with an even greater precision as offered by the PTP protocol, actually PTP2 in our case (IEEE 1588-2008).

 

Packets sent from a Ravenna source (also clocked by the PTP protocol) leave the sources at the same time, packets have the same size and contain samples starting with the same SAC, but by essence of Ethernet can reach their respective destination at a different time.

 

The Ravenna/PTP2 protocol ensures that they are properly resynchronized and are output by all destination devices exactly at the same moment, with a jitter of a few nanoseconds, whatever the distance, number of switch/routers and number of destination devices.

 

Ravenna is the perfect solution for your proposed system, one NADAC per speaker, as close as possible to the speaker.

 

Best regards,

 

Dominique

 

Thanks for the reply! Sounds like the Ravenna protocol solves the problem of syncing multiple units on a network.

 

 

It will just need to be determined if the implementation of the ESS chip is a superior one to DAC's like the $4995 Resonessense Mirus. It uses Dual Sabre 9018 chips in monoblock mode, with Crystek 950-20-50 clocks within inches from the chips. I'm currently using that DAC along with a modded Exasound E20 as my reference.

 

2 other DAC's that will be used in the comparison will be FPGA based. The MSB Analog DAC, and the Meitner MA-1. I'm hoping the NADAC has sound quality more in the league of these units.

[bmoura]

Thanks for the reply! Sounds like the Ravenna protocol solves the problem of syncing multiple units on a network.

 

 

It will just need to be determined if the implementation of the ESS chip is a superior one to DAC's like the $4995 Resonessense Mirus. It uses Dual Sabre 9018 chips in monoblock mode, with Crystek 950-20-50 clocks within inches from the chips. I'm currently using that DAC along with a modded Exasound E20 as my reference.

 

2 other DAC's that will be used in the comparison will be FPGA based. The MSB Analog DAC, and the Meitner MA-1. I'm hoping the NADAC has sound quality more in the league of these units.

 

Sounds like an interesting DAC shoot-out/comparison.

[Blizzard]

Sounds like an interesting DAC shoot-out/comparison.

 

Should be. I'm also making some mods to one of my Mirus's. First up is master clock with 10 HZ phase noise specs of -115 DB. Second is Dexa special edition discrete opamps to replace the stock AD 797's

 

http://newclassd.com/index.php?page=55

 

I'll have under $6000 into my Mirus with these mods so hoping Dual 2 channel NADAC's in a fully active setup with 4 amp channels will be superior to a single modded Mirus with 2 amp channels and passive crossovers.

[bmoura]

Should be. I'm also making some mods to one of my Mirus's. First up is master clock with 10 HZ phase noise specs of -115 DB. Second is Dexa special edition discrete opamps to replace the stock AD 797's

 

NewClassD Discrete OPAMPs

 

I'll have under $6000 into my Mirus with these mods so hoping Dual 2 channel NADAC's in a fully active setup with 4 amp channels will be superior to a single modded Mirus with 2 amp channels and passive crossovers.

 

Having heard the NADAC recently at The Show Newport - in both Stereo and Multichannel - I think it will be up to the challenge.

[Blizzard]

Excellent news! Another comparison will be single 8 channel NADAC using 2 of the channels for each speaker vs dual 2 channel units. Using dual units will allow the use of 6 inch long XLR cables to the amps that will be built into vibration isolated sub enclosures within the speaker stands. Each stand will have to be designed to hold a NADAC in the base pedestal.

 

 

This will allow for a streamlined sleek setup with only Ethernet and AC cables connecting into the each of the speaker stand pedestals. It's the next best thing to wireless, while still being able to take advantage of PCM up to 24/384 and DSD 256.

 

Using a single unit, long analog XLR cables will need to be run from a unit located on a equipment rack in between the speakers. This is the kind of setup I'm looking to avoid.