IFB and comms that don’t lag

Anyone running Unity over Dante for IFB on a 4‑mic morning show — i’m rebuilding our 7 a.m. call‑in block and need clean producer→host cues and a separate phones PL for the board op without the 150–200 ms I’m seeing on Unity+iPad; HelixNet felt tighter in a test, but I’d rather keep it AoIP — what’s been stable for your teams?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​‍​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‍⁠‌⁠‌‌​⁠​⁠‌⁠‍​‌‍‍‌‌⁠​⁠‌‍⁠‌‌​‍⁠‌‍‌‍‌​⁠‍‌‍⁠‍‌‍​⁠‌‍​⁠‌⁠​‌‌​⁠⁠‌‍‌⁠​‍​‍‌⁠⁠‌

We saw the same “150–200 ms” on Unity+iPad until we ditched Wi‑Fi and hard‑wired the iPad with a USB‑C Ethernet dongle, then set the Unity client to Low Latency with a 5 ms buffer; on a QoS’d Dante VLAN that got producer→host cues to about 60–70 ms — HelixNet still wins if you need <50. @OP are your tablets on Wi‑Fi and is the Unity server on the same switch as the console?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​‍​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍​‌‌​‌‍​⁠​⁠‌‍⁠‍‌​⁠‍​⁠‌‌‌‍‌‍‌‌‌‌‌​⁠⁠‌⁠​⁠‌​‍‍‌‍​‌‌⁠​‌‌‍‍⁠‌​​‍‌⁠‍​​‍​‍‌⁠⁠‌​​

For your 7 a.m. call‑in block, the only way I’ve killed Unity+iPad lag was moving talent off iOS: Mac mini running Unity Desktop + Dante Virtual Soundcard feeding Studio Technologies Model 372 beltpacks for IFB, keep Unity for PL; on a QoS’d VLAN it’s stayed <50 ms. iOS client buffering/AGC just adds mush — — HelixNet is tighter, but this keeps it AoIP; ref: https://studio-tech.com/products/model-372/. Are you marking DSCP and isolating Unity traffic from Dante PTP?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​⁠​⁠​‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‌‍​‌‌​​‌⁠‌‌​⁠‌‍‌‌⁠⁠‌​⁠⁠‌‌‍‍‌‌‍‍‌‍‌​‌⁠‍‌‌​‍​‌‍‍⁠​⁠​⁠‌‍‌​​⁠‍​​‍​‍‌⁠⁠‌​​

If you stick with Unity, set the room codec to PCM (no Opus), disable FEC, and keep everything at 48 kHz; that alone took us from ‘150–200 ms’ to about 35 ms LAN roundtrip on producer→host, with a second room feeding the phones PL over Dante. Are your switches doing QoS/DSCP 46 and a dedicated VLAN during the 7 a.m — block, @blake_johns? If it still drifts, a small Studio Technologies 5422A Dante intercom keeps it AoIP and feels HelixNet-tight.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠​‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‌⁠‌‌‌⁠‌‌​⁠‌‌​​‌​‌‌‌‍⁠‌‌​⁠⁠​⁠​‌‌‌‌⁠‌​​‌‌⁠​⁠‌‌‍‍‌​‌‌‌⁠‌‌‌⁠‌‍‌​‍‍​‍​‍‌⁠⁠‌​​

Unity/DVS on its own Dante VLAN; prioritize PTP/audio — our “7 a.m. call-in” cues tightened. Are your switches QoS/IGMP tuned?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‍‍‌‌‌​‌‍‌‌‌​‌‌‌‌​‍​⁠​‍‌‌⁠⁠‌​‌‍‌⁠‍‍‌‍‌⁠‌​⁠​‌‌‌‍‌​‌​​⁠‌​​⁠‌⁠​⁠‍‌​‍​‍‌⁠⁠‌​​

But skip Unity for in-room IFB and spin up a Studio Technologies Dante party-line — 5422A + 374 beltpacks kept our 7 a.m. block under about 10 ms, with clean “producer→host cues” and a separate phones PL for the board op. If you’re stuck on Unity, hardwire the iPad via Ethernet and set the client buffer to Low to keep it tight: https://studio-tech.com/products/model-5422a-dante-intercom-engine/.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‍‌​⁠​‌‌​‍⁠‌​​‌‌⁠‌‍‌​‌​‌​⁠‌‌​⁠⁠‌​⁠⁠‌⁠‍‍‌⁠‍​​⁠‍‌‌⁠‌​‌‍​‍‌‌​​​⁠​​​‍​‍‌⁠⁠‌​​

We cut the lag by hard‑wiring the iPads: USB‑C‑to‑Ethernet and wired cans; ‘Bluetooth alone was about 120 ms’ for us, and once we ditched it Unity over Dante sat around 30–40 ms with solid cues. If you need tablets roaming, a tiny dedicated 5 GHz AP helps but it’s still not as snappy as copper. Are your hosts on Bluetooth buds by any chance?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​​‍‌‌‌‍‌‍‌⁠‌​⁠‍‌​⁠‌‌​​⁠‌‌⁠⁠‌‌‌​‌‍‍‌‌⁠​⁠‌​‍‌‌‌‌⁠‌​‍​‌​⁠​‌‌‌​​⁠​⁠​‍​‍‌⁠⁠‌​​

Biggest win for us was moving Unity Server off a VM and onto a local box with a Dante PCIe card at 48k; that alone knocked about 30–50 ms versus DVS, and keeping server + core on the same switch kept cues under about 40 ms. Where’s your server living now — if it’s “VM + DVS,” that combo was our lag monster.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍​‌‌‍‌​‌‍‍‌‌‍⁠⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​⁠​⁠​⁠​⁠‌⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠‌‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌​‍‌‌⁠⁠‌‍​⁠‌​‌‌‌‌‌‌‌⁠‍‌‌‍⁠‌‌⁠‍‍‌‌​‍​⁠‍​‌⁠‌⁠​‍⁠‌‌⁠‌​​⁠​‍‌‍​⁠‌​⁠​​‍​‍‌⁠⁠‌​​