Re: [Jack-Devel] netjack1 / netjack2

PrevNext  Index
DateWed, 05 Oct 2011 06:50:03 -0700
From John Swenson <[hidden] at comcast dot net>
To[hidden] at lists dot jackaudio dot org
In-Reply-ToTobias Hoffmann [Jack-Devel] netjack1 / netjack2
Follow-UpTobias Hoffmann Re: [Jack-Devel] netjack1 / netjack2
Follow-UpRobin Gareus Re: [Jack-Devel] netjack1 / netjack2
Hi Tobias,
I actually did this a while back. I used netjack2 implemented in an 
FPGA. I used a wiznet ethernet chip to deal with the low level protocol 
stuff, the only thing it doesn't do is DHCP, so I just used static IP to 
make it easier. I did implement the master in the FPGA. For details I 
just read the source code and figured out what parts of it I wanted to 
implement. If you are just building a simple one to one system you can 
leave out a LOT of the jack stuff itself and just implement the netjack 
protocol. One advantage of doing the master is that you can set the 
sample rate to what you want and build a low jitter oscillator for that 
frequency. Of course its best if that is the same frequency used by 
whatever is the slave.

Things only get tricky if both master and slave have their own clocks. 
For example if the slave is an ADC plugged into a mic, it can get 
tricky. But if the slave is playing back a file there is no "local 
clock" in the slave. In the system I put together the slave was purely 
software so there was no problem with clock syncing, the master just had 
the main clock driving the DAC and the FPGA.

I haven't looked at netjack1 in detail. When I started this project 
netjack1 was not being supported and netjack2 ran on multiple platforms 
so that is the way I went.

John S.


On 10/3/2011 3:17 PM, Tobias Hoffmann wrote:
> Hi,
>
> as I've seen, there has already been a lot of discussion involving jack,
> ethernet and fpga implementations.
> AFAICS good timing (word-clock, low jitter) is hard stuff.
>
> My current issues are much more mundane, so here they go:
> 1. It is my understanding, that netjack (as slave) is one of simplest
> audio-over-ip protocols (others being dante, ravenna, etc,...),
> especially for fpga implementations, is this correct? (or would it be
> better to start with something else?)
> 2. Is there any concise ("RFC-like") documentation of the netjack1
> and/or netjack2 protocol (i.e. control packets, data packets)?
> 3. "Is netjack1 or netjack2 the way to go?"
> (Beyond knowing that netjack2 has discovery and advanced fragmentation
> support?)
>
> Regards,
> Tobias
> 
> Jack-Devel mailing list
> [hidden]
> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
>
> ---------------------------------------------------------------------------------------------------
>
> Text inserted by Panda IS 2011:
>
> This message has NOT been classified as spam. If it is unsolicited mail
> (spam), click on the following link to reclassify it:
> http://localhost:6083/Panda?ID=pav_11221&SPAM=true&path=C:\Documents%20and%20Settings\JohnNChris\Local%20Settings\Application%20Data\Panda%20Security\Panda%20Internet%20Security%202011\AntiSpam
>
> ---------------------------------------------------------------------------------------------------
>
>
>
PrevNext  Index

1317822637.20391_0.ltw:2,a <4E8C608B.7070606 at comcast dot net>