Hi TrigKeeper,

welcome to the RME forum.

There is no quality difference whether using digital connections like MADI or Word Clock (WC) for synchronization.
A WC star topology is in most cases not needed; synchronizing through MADI and ADAT saves you the additional WC cabling.

WC cabling has an advantage if you intend to change the sample rate often between single, double, or even quad speed
and if you want the slaves to automatically follow the clock of the master.

AES and SPDIF are the only digital protocols where it is possible for a slave to detect the sample rate from the digital data stream. Therefore, some devices had AES sync ports or AES I/O in general.

For ADAT and MADI there is no header information so that a slave could differentiate between single, double, and quad speed.
Therefore, you need to configure the sample rate at the slave(s). By this configuration, the device knows whether sample multiplexing between channels is needed. For single speed, no sample multiplexing (SMUX), for double speed, between two channels (SMUX/2), between four channels (SMUX/4). Note, all sample rates use SMUX protocol. SMUX/2 only means sample multiplexing between two channels, and this needs to be configured on each slave because it can't be detected automatically.

For MADI we have a special situation based on the history of the standard definitions.
Only for double speed they invented 96K MADI frame. Important to know is, that the 96k frame format is only in use when the devices are at double speed. If you change the sample rate to single or quad speed the frame format is again SMUX.
This leads to the "unfortunate" situation that a MADI slave can detect double speed and by a special bit in the protocol header it can differentiate between 88.2 and 96 kHz. But when the clock master changes to single or quad speed, the slave is only able to detect, that he receives again data in SMUX format. What he can't deceide is whether he should change to single or to quad speed. RME devices behave standard conform and do nothing, as it can't be decided.
So you can change from single or quad speed to double speed as 96k frame format at double speed is recognizable, but from there its not possible to detect single or quad speed properly, as in both cases the SMUX protocol is being used.
And the sample multiplexing you can only get by manual configuration on each of the slaves.

But also in this case there are exceptions; I want to mention them briefly.
1. Ferrofish seems to have drawn the design decision to change from double speed (96K frame format) to single speed once frame format changes to SMUX. Those devices can change as a clock slave from single to double and back to single speed automatically. But this is no standard; it's a design decision by Ferrofish. If you use quad speed, you will see, that they fail to recognize it properly.
2. Network devices that also support AVB or Dante can't switch automatically between sample rates at all. It would be nice if those devices could detect if AVB/Dante is not in use and then behave like a normal MADI device.
So for those devices, Word Clock doesn't make it possible to automatically and properly detect the sample rate.

If you want to have sample rate detection and that all clock slaves automatically follow the master, the best solution is to have either SPDIF or AES as a sync port (which can be a waste of ports) or to use Word Clock.

In your case, Word Clock would enable automatic detection of sample rate for
- the device connected to ADAT and
- the Ferrofish Pulse 16 MX and possibly also for the M32 Pro (if this is a pure MADI device)

So I would use the UFX III as clock master (driver settings, clock source internal).
Then you can set the sample rate in the DAW, and the UFX III gets the sample rate through the driver
and the slaves will follow the sample rate automatically, if they can (see above).

You have two choices for a setup, with and without word clock, depending on whether the automatic change between sample rates is needed or not. If not, then save the efforts of WC cabling.

MADI settings on each of the devices: 64ch, 96k MADI frame.
[cm] clock master, [cs] clock slave

Option A) without word clock; auto switching capabilities of sample rate very limited

Computer---USB2---ARC USB (optional)
|
| USB3
|
Fireface UFX III [cm]---ADAT1 OUT------------->ADAT IN---Arturia 16Rig
|         |                 \---ADAT1 IN<-------------ADAT OUT---/
|         |
|         + Ferrofish A32 Pro [cs]
|         |
|         + Ferrofish Pulse16 MX#1 [cs]
|         |
|         + Ferrofish Pulse16 MX#2 [cs]
|         |
+--<--+
MADI
OM4 or OM3

Option B) with word clock and automatic switching of sample rate (as good as possible)

Computer---USB2---ARC USB (optional)
|
| USB3
|                     
|                           WC OUT-----------------------------------------+
|                          /                                                                 | WC IN
Fireface UFX III [cm]---ADAT1 OUT------------->ADAT IN---Arturia 16Rig
|         |                 \---ADAT1 IN<-------------ADAT OUT---/       | WC OUT
|         |                                                                                  | WC IN
|         + Ferrofish A32 Pro [cs]                                            A32 Pro
|         |                                                                                  | WC OUT
|         |                                                                                  | WC IN
|         + Ferrofish Pulse16 MX#1 [cs]                                   Pulse 16 MX#1
|         |                                                                                  | WC OUT
|         |                                                                                  | WC IN
|         + Ferrofish Pulse16 MX#2 [cs]                                   Pulse 16 MX#2 (WC Termination enabled)
|         |                                                                                         
+--<--+
MADI
OM4 or OM3

I would use MADI optical to get galvanic isolation between the devices as an additional feature.
As OM4 multimode cables do not cost much more than OM3, I would use OM4 and stick to this one optical cable standard.
If you have in-house cabling, then you need to stick to the standard that is given by in-house cabling.
On one connection between two MADI devices, do not change the type of optical cable (OM3 or OM4), as this will negatively impact the signal by higher dampening. I would use the SFP multimode transceivers from RME (or Ferrofish) as they use the proper wavelength needed for MADI and they have a "neutral" firmware (not vendor specific like eg for Cisco, Brocade, Intel, ...).

Regarding clock synchronization in general, see my blog
https://www.tonstudio-forum.de/blog/ent … ios-en-de/