Digital Signal Chains & Networking

1. The OSI Model in Professional Audio To troubleshoot modern audio systems, one must understand the Open Systems Interconnection (OSI) model. Professional Audio over IP (AoIP) primarily operates at Layer 2 (Data Link) and Layer 3 (Network).

• Layer 2 (MAC): Systems like AVB (Milan) operate here. They use hardware MAC addresses to move data. This is extremely fast but requires specialized "AVB-capable" switches.

• Layer 3 (IP): Systems like Dante operate here. They use IP addresses, allowing audio to be routed through standard IT infrastructure.

2. Clocking and the Precision Time Protocol (PTP) In a digital network, every device must be "in sync." Dante and AES67 use PTP (v1 or v2). One device is elected the "Grandmaster." It sends a sync pulse (a "sync packet") hundreds of times per second. The other devices use this pulse to align their internal sample clocks. If a network is congested, these sync packets can be delayed—this is called "PDV" (Packet Delay Variation). If the PDV is too high, the device loses sync, resulting in "mutes" or "digital pops." Understanding how to configure Quality of Service (QoS) on a switch to prioritize these PTP packets is a core skill for any systems engineer.

3. Latency, Buffers, and the "Human Threshold" Latency is the time it takes for a signal to enter a system and exit it. In a digital network, latency is caused by:

• Packetization: Gathering samples into a "packet" to send over the wire.

• Processing: The time the FPGA/DSP takes to calculate EQs and Comps.

• Buffer: A "safety net" that holds audio to account for network jitter. For a vocalist using In-Ear Monitors (IEMs), the total round-trip latency (Mic -> Console -> RF) must remain below 7-10 milliseconds. Exceeding this causes "comb filtering" in the performer's head as the "bone-conducted" sound of their voice meets the delayed electronic signal, causing them to pitch incorrectly.

4. Redundancy: Primary vs. Secondary In high-stakes environments, we utilize "Hitless Switching" (ST 2022-7). This involves sending two identical streams of audio over two completely separate physical networks. If the Primary cable is unplugged, the receiver simply starts using the packets from the Secondary stream. Because the packets are timestamped with PTP, the switch is sample-accurate and audible glitches are non-existent.

5. Audio-over-IP Standards: Dante vs. Milan vs. AES67 Dante is the "market leader," known for its ease of use and massive device ecosystem. However, it is a proprietary protocol. Milan (based on AVB) is an open standard that guarantees "Deterministic" behavior—meaning the network reserves bandwidth for audio so it can never be interrupted by other data. AES67 is a "translation" standard, allowing Dante devices to talk to Ravenna or Q-SYS devices. Understanding the interoperability between these formats is essential for festival "system drive" positions.

II. Practical Lab: Network Integration

• Objective: Route 16 channels of redundant audio across two virtual switches.

• Tasks:

  1. Open Dante Controller. Identify the "Primary" and "Secondary" IP ranges.

  2. Configure a "Multicast" flow for a 16-channel drum kit. (Theory: Why use Multicast instead of Unicast for a broadcast feed?)

  3. Setup a Waves SuperRack host and observe the "System Latency" meter while changing the buffer size from 32 samples to 128 samples.

  4. The Failure Test: Unplug the "Primary" network on your virtual bridge and observe if the audio continues on the "Secondary" side.

III. Daily Assessment

1. Math: If your sample rate is 48 kHz and your buffer is 32 samples, what is the "Buffer Latency" in milliseconds?

   The buffer latency is 0.67ms | Latency = Samples/Sample Rate

2. Theory: Explain the difference between Layer 2 and Layer 3 networking in the context of professional audio.

   Layer Two uses hardare mac address to send the data between units (Milan, AVB etc) whilst the Layer Three uses IP addresses to sent the data from point to point (Dante etc).

3. Application: Your Dante network is showing "Sync Errors." Name three things you should check in the network switch configuration.

   The "Big Three" Switch Settings for Sync:

   EEE (Energy Efficient Ethernet / Green Ethernet): This MUST be disabled. It puts ports to "sleep" during silence, which destroys clock synchronization.

   IGMP Snooping: Ensure an IGMP Querier is active so multicast clock traffic is handled correctly.

   QoS (Quality of Service): Ensure the switch is prioritizing PTP (Clock) traffic using DSCP values (usually 56 for Clock) so timing packets aren't delayed by audio data.