This AVB Bandwidth Calculator helps you determine the maximum number of audio streams and channels that can be carried over an AVB (Audio Video Bridging) network link. Enter your network and stream parameters to see real-time calculations of bandwidth utilization.

Provided by Jeff Koftinoff <jeff.koftinoff@gmail.com>. See avb.statusbar.com for more information.

Understanding the Results

The calculator shows:

• Total channels per network link — Maximum audio channels the link can carry with your settings
• Max stream count — How many streams fit within your AVB bandwidth allocation
• Efficiency — Ratio of useful audio data to total packet overhead (higher is better)
• Largest Frame Payload — The Ethernet payload size; must be ≤1500 bytes for standard Ethernet

Network Speed

The physical link speed of your AVB network. Common speeds:

• 100 Mbps — Legacy networks, limited to ~50-100 channels
• 1 Gbps — Most common for professional audio; supports hundreds of channels
• 10 Gbps — High-density installations; supports thousands of channels

AVB BW Allocation

The percentage of network bandwidth reserved for AVB Time Sensitive Streams. The default of 75% means AVB traffic can use up to 75% of the link capacity, leaving 25% for other network traffic (control protocols, IT traffic, etc.).

This is configured in your AVB switches via SRP (Stream Reservation Protocol). Lower values leave more room for non-AVB traffic but reduce the number of audio streams possible.

Stream Format

AVB uses two audio stream formats defined in IEEE 1722:

AM824

The original AVB audio format, compatible with IEC 61883-6. Supports:

• 24-bit linear audio
• IEC 60958 encoded audio (S/PDIF and AES3)
• SMPTE Time Code
• MIDI

AM824 has three packetization modes. All AVB listeners must support receiving any mode:

AM824 (non-blocking, synchronous)

The most common mode. Sends one Ethernet frame every observation period (125μs) with a fixed number of samples (6 samples at 48 kHz, 12 at 96 kHz). Predictable and efficient.

AM824 (non-blocking, async)

Allows occasional frames with one extra or one fewer sample, accommodating devices with multiple clock domains sharing one packetizer. Requires slightly more bandwidth reservation to handle the occasional larger frame (7 samples at 48 kHz). Apple Macs use this mode.

AM824 (blocking)

Originally used by FireWire audio devices. Sends fixed-size frames (8 samples at 48 kHz, 16 at 96 kHz) but may skip transmission periods when no frame is needed. Simpler to implement but requires reserving bandwidth for the larger frame size.

AAF (AVTP Audio Format)

A newer, more flexible format with several advantages:

• Lower overhead than AM824
• Supports 16, 24, or 32 bits per sample (including floating-point)
• Configurable samples per frame for optimizing efficiency vs. latency
• Fixed frame size and format throughout a stream

Sample Rate

All AVB stream formats support standard audio sample rates:

• Base rates: 32 kHz, 44.1 kHz, 48 kHz
• Double rates: 88.2 kHz, 96 kHz
• Quad rates: 176.4 kHz, 192 kHz

Higher sample rates require proportionally more bandwidth. AAF additionally supports NTSC pull-up/pull-down rates (×1.001 and ×1/1.001) for video synchronization.

Bits/Sample

The bit depth of each audio sample:

• AM824: Always 24 bits (fixed by the format)
• AAF: 16, 24, or 32 bits per sample

Use 32 bits for floating-point audio. Lower bit depths reduce bandwidth but may not be supported by all devices.

AAF Samples/Frame

For AAF format, you can configure how many audio samples are packed into each Ethernet frame. This creates a trade-off:

• Fewer samples/frame — Lower latency, but more overhead (more frames, each with headers)
• More samples/frame — Higher efficiency, but increased latency and larger frames

The "Default" setting uses the observation interval to determine optimal packing. Custom values let you fine-tune for your latency requirements.

Channels Per Stream

The number of audio channels bundled into a single AVB stream. A stream is the unit of bandwidth reservation on the network.

• Typical values: 2 (stereo), 8 (common multi-channel), 32, 64
• Maximum: Limited by Ethernet MTU (1500 bytes payload)

More channels per stream means fewer streams needed, but each stream becomes a larger "chunk" of bandwidth. Fewer channels per stream gives more routing flexibility but increases overhead.

AES-SIV Encryption

IEEE 1722 supports AES-SIV (Synthetic Initialization Vector) encryption for secure audio transport. When enabled:

• Adds 16 bytes of authentication tag overhead per frame
• Provides authenticated encryption (confidentiality + integrity)
• Reduces maximum channels per stream slightly due to overhead

Observation Interval

The time period at which AVB streams transmit Ethernet frames. The standard interval is 125μs (8000 Hz), meaning each stream sends one frame every 125 microseconds.

Longer intervals (e.g., 1ms) pack more samples per frame, improving efficiency but increasing latency. This setting affects bandwidth calculations for all stream formats.