Whshhh!!! Hold on…
So many parameters U need to consider. Let’s take it 1 step at a time:
1) When connecting digital components, make sure there is only 1 clock source (master), which all other clients r slaved to. If ur clients don’t support 96Khz (like the vs 880 4 ex.) U’ll need to cut the frequency down.
2) AES SP/DIF and SP/DIF2 are the same basically, however AES and SP/DIF carry an additional subcode containing clock and general data, making it more volnurable to jitter related problems. Both r driven through coaxial cable, I presume COAX is the term the manufactorer used in order to distinguish SP/DIF from optical. That means U can connect the clients. Of course AES code is balanced SP/DIF isn’t.
3) When reducing resolution (24 > 16 bit 4ex.) U need to add dither (small amount of noise added in order to prevent near-zero related distortion).
Going up is no problem – zeros r added. That means that U’ll have to check whether ur card is up to the task. Anyway, compare the two options (24bit and 20 bit at the mic pre), they will sound different. Use a dynamic source 4 checking – I use a piano piece with quite passages.
4) Funny as it may sound, it’s not always “the higher rez – the better”. If I record a PT project 4ex., the material is not too dynamic & I know I’ll b mixing within the program, I’d rather stay in 16 bit and leave more headroom 4 the mix buss. That way the faders r closer to unity gain and less bits r chopped in the process.
5) Choosing the better converter will improve ur sound. Raising resolution when not all components r capable of handling it – not always. Ur chain is as strong as the weakest link in it.
6) Make sure ur cables and connectors r as good (short and hi quality). Monitor at every stage and look 4 sound degradation.

Good luck…

Peace,
Zooot