markmcconnell / mai

Mark's AES67 Implementation
BSD 3-Clause "New" or "Revised" License
95 stars 22 forks source link

Unstable initial PTP source #10

Open QRabbani opened 1 year ago

QRabbani commented 1 year ago

First of all, thanks for making this package! This is one of very few packages I've found that demonstrates proper PTP-synced and SDP-broadcasted AES67 stream generation that also seem to work with my Dante-enabled speaker.

I have both a Dante mic (normally the leader) and a Dante speaker (both AVIO adapters) on my local network. When I start up your package, within a few seconds, a weird back-and-forth oscillation between the two devices begins as your PTP clock alternately displays each of my devices as the new PTP synchronization source only to eventually settle after a few minutes.

During this period, the latency stats in Dante controller (routing the AES67 stream to my speaker) are abysmal but are totally fine after the weird back-and-forth settles. Also, both become primary leaders until it stops as well. Lastly, the first response to a delay request only seems to happen afterward (not even in the few seconds before).

Do you know why this could be and if it might be fixable?