Dante card audio sync

[tutu13]

Hi everyone,

I installed an optional M-Waves card mainly to record my lives in multitrack (the USB is already taken) and, later, to test the Soundgrid plug-ins.

But the fact is that the card does not automatically sync to the console frequency.
I work primarily with the SQ, it seems logical to me to leave it sync master. Working with an AR2412, sync clock is set at Internal 48kHz.

While for the M-Waves card:

“1-The clock source is set to Digital. i.e. locked to the console, and the sample rate is 96 kHz. It is NOT POSSIBLE to work with 48kHZ when the card is locked to the console.
2-The clock source is set to Internal, i.e. the console is locked to the card, and the sample rate can be 96 kHz or 48 kHz.”

This explains that if I want to make the card run at 48kHz, I must open the Qrec application which ALWAYS opens on the frequency of 96 kHz and leave it open in the background. Of course at this frequency the audio can’t transit, I must first open the card’s settings to change it, assuming that my settings will be taken into account for each future use, that these settings will be memorized in the card like with all sound cards. Well no !
Each time I use it I have to open the application then change the frequency.
Okay very good, so to go faster I saved my Qrec config on my desktop to recall it with a single click,
this opens Qrec but still at the default frequency (96kHz)
and I have to click the gearwheel again to reload my parameters,
but “LOAD” does not offer the last path traveled, I still have to go find my shortcut…

… a whole ritual that I would like to avoid in moments when all must go fast.

In short.. when I took this card I had no idea that it could not automatically set itself to the SQ at 48kHk, nor that it would not memorize its last settings..

Return to sender,
Then.. I aim the Dante card hoping that it works on its own, we turn on the A&H console and Dante automatically syncs to the SQ internal frequency;
Is that the case?

THANKS

[Scott]

The Dante cards can be set to sync to host (SQ) or from any other source using Dante controller. Once set in controller, the change will remain. So YES it can be synced to the SQ.

[ian.hind]

SQ will ALWAYS run internally at 96k. Even when using 48k stage boxes, the console will SRC the stage box itself up to 96k.
Eg. A SQ-5 local inputs will still run at 96, even with a 48k AR2412 patched in.

The SQ-Waves will automatically grab clock from console on boot, which is 96k.

You can SRC down to 48k with the driver software but I’m not aware of a way to make this happen automatically.

Not an answer to your question but maybe it’ll help you understand the architecture?

[tutu13]

> “you can SRC down to 48k with the driver software”

wich driver software?

[Brian]

As noted, the SQ console will always run at 96k. The I/O cards are responsible for any reclocking that needs to happen. For example, a Dante network has to run with all connected device at the same clock speed. If even a single connected device is limited to just 48k (and there are a lot of Dante devices that don’t run faster than 48k), then the entire Dante network must run at 48k. The is generally done automatically, but I believe you can adjust a device’s clock setting with the Dante Controller software if you had the need to do it manually.

Long story short, the A&H Dante I/O cards have no problem reclocking the 96k signal from the console down to 48k or whatever clock speed is needed for the larger Dante network. But this occurs on the I/O card hardware itself, not in the console. This is also why it is critical for the console’s I/O card to be set to “Enable Sync to Ext” in Dante Controller (meaning the I/O card is getting it’s signal from the console’s clock which is external to the I/O card’s hardware) and that the console’s I/O card be set as the “Preferred Master Clock” to ensure that all the other Dante devices use the I/O card as their clock since it is the only device that can communicate with both the console’s clock and the rest of the Dante network.

[ian.hind]

When you install the SQ-Waves driver, it comes with a small application for clock and sample rate selection.

[tutu13]

So initially, I preferred to set the console at 48kHz rather than 96kHz because I thought that this would generate less delay in the signals coming from the AR1224 due to the 96>48 conversion, but I was probably paranoid!
Does this add extra milliseconds or not at all?

Yes Scott, the problem is that you have to redo the settings each time on the M-Waves, but if you record at 96kHz (default setting) you just have to click on the Qrec app.
Multitrack recording at 96kHz doesn’t take up too much CPU, so I figure “why not?”

But then a question comes to me: if later I embark on the adventure of Soundgrid plug-ins on server, will running at 96kHz rather than 48kHz change the latency?

[Flo84]

Difference between 48/96 KHz stagebox are very low.
For AR/AB I had read 1.1 ms input to output (with no processing like dynamic eq) in the Allen&Heath doc.

Florent

[tutu13]

Well, I just did some latency tests with a MacBook connected to USB B:

Logic plays a Kick sent to an SQ channel routed to an AUX,
this AUX returns into Logic always via USB B
then is recorded on a second track.

On the recorded track I notice a delay of 604 samples (in 48kHz) which I already find enormous compared to other USB sound cards.

Then I do the same thing again, this time using the M-Waves as an I/O device, still at 48kHz.
Recording delay: 688 samples, that’s more than via USB.2 !

I find that weird. On paper the M-Waves card is supposed to transfer signals faster than USB.2, right?

Can I hope to have a better flow rate with the Dante?

[Brian]

The Waves card has sub 1ms latency when used with the Waves Soundgrid servers. That’s because Waves has spend the time and R&D to create not just custom drivers, but they have actually customized the entire OS from the ground up with a singular goal – no minimize the audio latency as much as possible. If you are using it with a normal computer along with the Waves drivers for your OS, then it is going to have much higher latency because the overall OS (Windows or Mac OS) is not optimized for audio latency.

You should not expect a lower latency using Dante DVS, but it might be possible to achieve lower latency if you use a hardware I/O interface (like a Dante PCIe card, RMS Dante Digiface, etc). However it is still going to be MUCH higher than the sub 1ms that Waves can achieve with a proper Soundgrid Server.

You can DIY your own Soundgrid Server. The software is readily available directly on the Waves website and there is a very active user group on Facebook (Waves DIY Soundgrid Server group). You still need a valid Superrack license and valid licenses for your plugins, but it is very easy to set up a machine that can take advantage of the super low latency offered by the Waves Soundgrid system.

[tutu13]

Thank you for your detailed response.
I’m disillusioned, so it’s the programs & drivers that make the bridge between these cards and the computer that create latency higher than that of USB, does the Ethernet > Thunderbolt conversion of the Mac also add it?

[Adethefade]

Our show MacBook (multi-track audio and video) runs DVS and operates at 6ms. The rest of the network operates at 1ms: I believe it’s a limitation of the OS, not Dante or DVS. There’s no discernible issues during operation though.

[Brian]

Our show MacBook (multi-track audio and video) runs DVS and operates at 6ms.

Is that latency number roundtrip or only one-way? Have you measured the real latency using a program like SMAART or are you relying on the self reported latency of the software? I’ve never tested an Apple system myself to know for sure, but I would be shocked if it actually measures 6ms round trip latency using DVS. Honestly that is getting close to total latency times seen when using hardware interfaces (usually around 3-4ms). Using a software interface like DVS is always much slower. I’ve normally seen total latency times in the 12-18ms range (roundtrip) when using DVS. FYI – the “reported” latency shown in software is not accurate because it does not include all the potential points of latency in the overall system. Total latency will always be higher than what the software reports.

[Author]

Posts