After reading about clocking, consensus seems to be that the device that does the actual AD<>DA conversion should be the master, using it's internal chip as this is least prone to jitter.
The clock connection to the MADI FX will function as accurate over madi fiber as over a seperate WC cable since the MADI FX is not doing any conversion. So far so good.

Now, I've got a SSL Alphalink MADI AX and a SPL Madison.
The SSL is on madi port 1 as internal clock and the SPL on madi port 2 also configured as internal clock.
This seems to work fine and because they are seperate parallel connections (units are not connected together other than that they both end up in de MADI FX) I'm not sure if this should cause problems
but I'm wondering if i should slave the SPL to the SSL with wordclock as well rather than both only syncing with the MADI FX and manually being put on the same samplerate.

are there any potential clock / sync / phase issues that i'm overlooking? I know there should only be 1 masterclock in the chain, but this was before RME invented parallel madi ports
So does this work just fine since it's parallel and not serial? or do i need to configure 1 unit as slave and hook it up with the other via BNC?