You can do it either way. Some people choose to split their data into multiple experiments, some people put everything together. So this is up to you. Here, we include a few notes that will hopefully help you decide.
FlowRepository contains datasets that include several thousands FCS files. While this is possible, please note that if you put too many data files in the same experiment, if may become harder for users to navigate, and it will take a bit longer for the experiment to load. Therefore, if you have a very large dataset that can be logically (and meaningfully) spit into a few smaller pieces, you may consider creating separate experiments for these parts. However, this is assuming that these parts are more or less independent. If you want to share a gating strategy, compare data from different parts in a shared figure, etc. then you should put them all into the same experiment.
Using different panels or tubes typically does not warrant the creation of separate experiments. Please see the topic about resolving conflicting channels if you are facing this issue.
If you are using different cytometers, then you can still put all the data in a single experiment; however, proper instrumentation annotation will take a little bit more effort if you do that. This is because you will not be able to describe the instrument once and apply the same settings on all FCS files. Instead, you may have to first apply the settings of the mainly used cytometer to all the data files, and then correct the description for all the files generated by a different instrument.
If you decide to split a dataset in several parts, then these cannot be directly linked to each other; however, they can all point to the same manuscript (and your manuscript can point to all the parts).