Kessito

I was wondering if anyone out there has ever tried to communicate with a cobranet digital audio network through the use off the standard tcp/ip socket on the computer?

It should be possible to use the cobranet protocoll without the hardware with maybe a reduced number off channels.

http://www.cobranet.info/en/support/cobranet/faq.html

www.cirrus.com/en/pubs/manual/CobraNet_Programmer_Manual_PM25.pdf

I was wondering if anyone out there has ever tried to communicate with a cobranet digital audio network through the use off the standard tcp/ip socket on the computer?
It should be possible to use the cobranet protocoll without the hardware with maybe a reduced number off channels.

http://www.cobranet.info/en/support/cobranet/faq.html
www.cirrus.com/en/pubs/manual/CobraNet_Programmer_Manual_PM25.pdf

cobranet-external-would-it-be-possible

Have a look at Olaf Matthews' efforts to send audio over IP, although he says that his objects are not stable and they don't run on Intel machines:

I myself had pretty good results with the jitter/audio example that does the same thing, although probably with more latency than strictly necessary:

https://cycling74.com/story/2006/10/23/104657/91

It is not mentioned in the article that it helps a lot to use a metro instead of the qmetro and set the scheduler to overdrive and in audio interrupt in DSP Status. I think this makes sense since the matrices will be sent in sync with the audio buffer update and the send will complete before the matrix is filled with new info.

If you're interested I could post my patches.

Have a look at Olaf Matthews' efforts to send audio over IP, although he says that his objects are not stable and they don't run on Intel machines:
http://www.nullmedium.de/dev/netsend~/

> Have a look at Olaf Matthews' efforts to send audio over IP, although he says that his objects are not stable and they don't run on Intel machines:

Right, I will repeat myself over and over again:

What you need to take in audio signals over network is a way to 

syncronise the incoming data with your internal clock rate. 44.1kHz on 

one machine is not 44.1kHz on another. So when sending from one machine 

to another you either get too much data to or not enough data. But on a 

really tiny scale, like a few samples off per minute, so resampling in 

the classical sense does not work. Additionally, you have fluctuations 

in the network speed, so just checking for one second and then figuring 

out how many samples you get per second does not work.

As long as one needs only Cobranet one could probably make an "external" 

that pretends to be an audio interface. Like ad_asio.mxe and 

ad_rewire.mxe and friends. That way whole Max could sync to the signal 

rate on the Cobranet. But I have no idea (and guess not) whether the API 

to do that is public. And the next thing is, you probably want to listen 

to it, so you don't want to block your dac~ with the Cobranet in/output.

Mattijs Kneppers wrote:
> I am interested in this as well.
>
>
> Have a look at Olaf Matthews' efforts to send audio over IP, although he says that his objects are not stable and they don't run on Intel machines:
> http://www.nullmedium.de/dev/netsend~/

What you need to take in audio signals over network is a way to 
syncronise the incoming data with your internal clock rate. 44.1kHz on 
one machine is not 44.1kHz on another. So when sending from one machine 
to another you either get too much data to or not enough data. But on a 
really tiny scale, like a few samples off per minute, so resampling in 
the classical sense does not work. Additionally, you have fluctuations 
in the network speed, so just checking for one second and then figuring 
out how many samples you get per second does not work.

As long as one needs only Cobranet one could probably make an "external" 
that pretends to be an audio interface. Like ad_asio.mxe and 
ad_rewire.mxe and friends. That way whole Max could sync to the signal 
rate on the Cobranet. But I have no idea (and guess not) whether the API 
to do that is public. And the next thing is, you probably want to listen 
to it, so you don't want to block your dac~ with the Cobranet in/output.

I now that the cobranet timing synchronisation works with "beat packets", these are time-stamped packets that contain the timing information from every client in the network, in this way all the clients are always synchronised with eachother. These beat packets contain the timing info for packets off 64 samples, hence the minimum latency off 1,33ms (64samples/48kHz) in a cobranet network.

I know that the API is available online at cirrus, I got it by accident a couple off months ago, but forgot where I found it and can't find it back yet.

The reason why I would like to have an object like this is because there is a lot of pro-audio cobranet equiped gear out there and it would be nice to have a way off comminucating with it.

I know that the cobranet hardware module (CM-1 / CM-2) consists off a FPGA to convert the data to a SPI buss and some RAM to buffer a sample package. Do you guys think that it (theoretically)would be possible to get a fail-safe replacement without the hardware?

I know that the cobranet hardware module (CM-1 / CM-2) consists off  a FPGA to convert the data to a SPI buss and some RAM to buffer a sample package. Do you guys think that it (theoretically)would be possible to get a fail-safe replacement without the hardware?

I'm an engineer for a commercial audio/video installation contractor, and we design/install systems which include CobraNet all the time, so I'm very familiar with it. Also, I wrote the Wikipedia article on CobraNet; it's fairly comprehensive and it contains a link to Cirrus' programmers manual at the end of the article.

In order to communicate with a piece of CobraNet hardware, you would have to write a program which emulates a CobraNet card. I'm not an accomplished C/C++ programmer, but to me this would be a very difficult and complicated task. However, I believe all of the information you would actually need is publicly available.

However, an easier way to go about this might be to just buy a readily available CobraNet sound card for your computer...

I'm an engineer for a commercial audio/video installation contractor, and we design/install systems which include CobraNet all the time, so I'm very familiar with it.  Also, I wrote the Wikipedia article on CobraNet; it's fairly comprehensive and it contains a link to Cirrus' programmers manual at the end of the article.

In order to communicate with a piece of CobraNet hardware, you would have to write a program which emulates a CobraNet card.  I'm not an accomplished C/C++ programmer, but to me this would be a very difficult and complicated task.  However, I believe all of the information you would actually need is publicly available.

However, an easier way to go about this might be to just buy a readily available CobraNet sound card for your computer...


Yeah, that would be easier, but the problem is that there are no 64 channel cards being sold, only 16 in 16 out, and cascading more off them would be a really expensive solution since they cost 1600 dollar each..

To do CobraNet directly from or to a computer, you'd first need to understand the protocol. The technology is owned by Cirrus Logic and details are not publicly available.

If you did understand the protocol you'd still need to master the art of sending and receiving layer 2 Ethernet packets with strict timing constraints. Standard networking APIs generally do not readily afford the kind of low-level access required as doing so is considered a security risk to the operating system and the network.

In order to get through this and the timing issues, you generally need to work at the device driver level to have any chance of cracking this nut. Is that something you're up for?

Cobranet external, would it be possible?