Daisy Chain issue

[JohnnyRoehn]

Hello,

I am having an issue with the XMOS AVB Daisy Chain Kit when connecting to a talker stream. My setup consists of 4 Daisy Chain Boards connected in a daisy chain to a XMOS talker. When I try to connect one of the last 3 Boards in the daisy chain as a listener, they return a successful connection but there is no sound coming out of the analog outputs. Only If I then connect the first Daisy Chain board of the chain, all the others also output audio normally. After that I can also disconnect the first Daisy Chain board again and the other boards of the chain still can playout normally. It seems like the three other boards in the daisy chain need something like a sync once from the first daisy chain board to operate correctly.

Does anyone have the same problem or any solutions?

[JohnnyRoehn]

No idea anyone?

[larry]

Which node is media clock source (LOCAL_CLOCK)?

[JohnnyRoehn]

Do you mean the Grandmaster Clock of the AVB Network?

[larry]

One node runs its media clocks off a local clock source (the local PLL driven by XCore itself, based on XCore's system oscillator), this one prints "Setting clock source: LOCAL_CLOCK". All other nodes recover media clock (local PLL is driven by XCore based on incoming timestamps) for a given stream. They print "Setting clock source: INPUT_STREAM_DERIVED". If you have an LC endpoint alongside your DC boards, the LC will be media clock source. If there are multiple LCs, they will elect a source amongst themselves based on GUID (MAC address), see "simple demo controller".

[JohnnyRoehn]

It seems like all of the devices show INPUT_STREAM_DERIVED on startup. I have one LC board and 4 DC boards in my setup. So what could be the problem?

[larry]

In AVB, one endpoint's media clock is locked to by all other endpoints. That way all endpoints can play synchronised audio.

[JohnnyRoehn]

Yes, I know that concept in AVB. I forgot that there is a Marvell Firefox AVB Switch between the 4 Daisy Chain Boards and the LC XMOS Talker. The Marvell Switch should be the Grandmaster Clock in this case. So could this clock be the problem?