dgburns
software surfer
Ok, This post is for @charlieclouser - I'll take it out of the other thread out of respect for the thread poster. Lemme try and run through.
MacPro Daw machine:
-Focusrite pcie card - this is the main core audio sound card. Solid so far.
-to-
-Focusrite Rednet 2 ( maybe there are better sounding units, but it's ok )
-Verto MX madi-dante ( which I hooked up an SSL Alphalink 24x24 plus adats, mostly for additional inputs )
Protools Cheesgrater ( Running PT10 HD, lol )
-Focusrite Rednet 5 ( 32x32 ins/outs show up as 2x 192's )
-Still have a couple 192's I can hook up if I really need to. 192's play nice with the Rednet since Rednet can match the latency.
Cisco SG200-26 switch 1 gig switch ( CAT5 cable good enough ). I have one for Dante and another for regular internet/VEPro etc. I keep the Dante stuff separate. The Cisco's are older and cheap as chips, and I set them up according to the Focusrite/Dante recommendations. There are a couple things to watch out for, but if I can figure it out. YOU can most definitely figure it out. The one big one is to avoid letting the switch use green settings. There's another thing to setup, I'd have to go back and verify what exactly. Anyway, the basics are that you should get a quality switch that allows you to turn off power green settings. No fancy settings required, an unmanaged switch could maybe get you there depending on traffic. Pretty sure they were recommending the Cisco SG200 amoung others, so I went with that one. They are a pain to update firmware though, it has to be done though a browser, and for some reason I could not get one switch to update. In any case, I use that one for regular internet as well as VEPRO and it's fine across three slaves plus other stuff like NAS etc.
Buffer settings:
Tend to work at 256 in Logic when doing stuff with VePro and basic scoring stuff. If I have a singer or guest I tend to lower it to 64. I don't tend to lower it to 64 when I do my own guitars and stuff, I can deal with that 256 buffer setting, but some ppl really need lower latency. If the session has alot going on, I might just pull a guide mix and turn off all the tracks I don't need, but this is only if I need to layer stuff and feel I need to reduce the strain. Or i might just freeze some tracks and leave most of it live while I have singers etc tracking. Don't usually go to 32 buffer, feel 64 is fast enough.
( post thought -edit- I caught myself on more than one occasion forgetting to switch the buffer back to 256, and noticed a bit more cpu usage, but I could work at 64 if I really felt the need to. Dante does not seem to care one way or the other )
Above is the Dante app, and you can see the latency. Keep in mind I run at 48k and I have 16x16, 24x24 and 32x32 streams of audio here. For a total of 72 each way. I don't think the network is even breaking a sweat. And if I really needed to , I could up the latency, right now it's set to .5 msec for the Rednet Pcie, but as you can see, the peak latency I'm seeing is 187 usec and no late packets. The verto has a bit more latency than the Rednet stuff, they show at 166 usec.
I love the flexibility, because I can monitor through PT, or through my DAW, and any combination of things. You can save presets of all this, so you can switch between using different audio i/o or from any dante source to any other dante destination. Love it.
Wordclock:
Ok, had to dig around on this aspect. IF you use an external wordclock box, attach it to only one dante equipped interface, and make that the Dante system master. All other Dante equipment will follow. In my case, I still use a Big Ben, and it is hooked up to the Verto ( and SSL ), so it is the house clock. Dante goes to great lengths about how wordclock is stable in the entire system, so you don't need to worry too much about word for any dante enabled unit, so long as they are seen and connected to the Dante network.
Addressing:
No real need to assign IP addresses, if you don't the boxes will self assign, and Dante just connects them. My switch for the Dante stuff is not connected to the internet, it sits alone.
Room Correction:
Ok I use the Sonarworks thing, not the Trinnov. What I notice is that the lowest ( fastest ) setting still introduces latency, but that's the plugin, and I don't think the Trinnov will help here. It has to do with the lower frequencies, and the correction that is being applied. If you use the linear phase setting, the phase response is quite flat, but at the cost of more latency. The Sonarworks has a low latency mode, but it does not correct for phase, and a mixed mode, where it mostly corrects for phase, but not like the linear phase setting. The Sonarworks introduces alot of latency to monitoring. But you know, the room correction is pretty important, so I live with it. I can run it either in PT or in Logic, depending on which system I am monitoring through. When scoring to picture, I run it through and monitor through PT right to the Rednet 2.
You can see what's going on for any Rednet box in their app. It's helpful, but I rarely feel the need to use it.
Final Thoughts:
It all just works. I tend to forget about it, sometimes I marvel at how much audio is being thrown around, but mostly I forget about it. It also sounds nice, and maybe a bit more clear then other systems I used in the past. I don't hear a big difference using the Big Ben, but habit keeps me using it. To the Virtual Dante version - I DO NOT RECCOMMEND. I tried it, but the latency is too high. I tend to think you would be dissappointed having been used to AVID and Motu stuff. If you do go Dante, I'd go with a hardware sound card all around. Special use cases aside. The Rednet 5 are really cheap these days.
Dante is kinda like going modular. You can add things later, or even have things simply not connected depending on what you are doing. It boxes you in a bit less, since you can hook up anything to it, if it's got the dante interface. Once the device is on the network, it's available to everything there.
Last edited:
