Free forever·no account, no tracking, no third-party requests ·open data, free API
showstack / compare / acn-vs-sacn

ACN vs sACN

Are ACN and sACN the same thing? Below is every field we hold on both, side by side. Each one is sourced on the individual entry, and anything nobody could source is left blank rather than guessed.

Side by side

ACNsACN
Categorynetwork-transportlighting-control
FlowControl — small, urgent, must arriveControl — small, urgent, must arrive
PortsUDP 5568UDP 5568
Transportudp, tcpudp
Multicastyes239.255.0.0/16 (IPv4, universe in low two octets), FF18::83:00:00:00/104 (IPv6)
Addressing—512 slots per universe; universes 1-63999, plus 64214 for E1.31 discovery.
Spec availabilityopen-publishedopen-published
StewardESTA (Entertainment Services and Technology Association)ESTA (Entertainment Services and Technology Association)
First published20062009
Statuscurrentcurrent
Our confidencereportedverified

Both are control flows. Latency and loss on the messages that fire once. A repeating level survives a dropped packet; a GO does not. That applies equally to either, so it is not a discriminator here.

Speaks both (1)

These are the safe choices if you have not committed to either, or need to bridge the two.

sACN only (75)

What goes wrong with ACN

When someone on headset says "ACN" they almost always mean sACN (E1.31). They share a port and a root layer but not a purpose, and a device that speaks one does not necessarily speak the other at all.
ACN is a suite of documents plus EPIs that pin down discovery, addressing, port assignment and timeouts. Implementing SDT and DMP from the core documents alone will not give you anything interoperable.
DDL is the whole point and also the hard part. A device is only as self-describing as the DDL file the vendor shipped, and incomplete or stale descriptions are common.
SDT is session-based with membership and sequencing, so it carries state. Dropping and rejoining a session is not the free operation that restarting an sACN packet stream is.
Real-world deployment is narrow. ETC's Net3 and some Shure wireless gear use full ACN; most of the industry only ever met it through the sACN root layer, so third-party tooling is thin.

What goes wrong with sACN

Multicast without IGMP snooping configured on the switch turns into a broadcast storm. This is the single most common cause of "the network died when we patched the second console".
Priority is per-universe (0-200, default 100) and sources with equal priority get merged by the receiver, not by the sender. Two consoles at default priority will fight.
Universe numbering is 1-based here but many products expose a 0-based index internally. Always confirm against the receiving device, not the console.

Sources

Disagree with any of this?

Every claim above comes from one of the two entries and carries a source. If one is wrong, correcting it is a single file and it fixes this page automatically. Edit ACN or edit sACN.