All of the entries in the quantity option of the combiner are incorrect in some way.

First off, any text values you enter are generally enclosed in quotation marks ("text"). So the Column field should read "S".
A source's arrival schedule can not be accessed directly by name. You have to refer to the respective node like the trigger on the source does. Tables that can be accessed by name are Global Tables, Statistics Collectors, Calculated Tables and State Tables. Again, the name would have to be enclosed in quotation marks.
"item" is a reference to an object and can not be a row identifier. This has to either a numerical or text value (so in theory, "item.name" could be used to identify the row, if the schedule of the source was set up to assign the row header as the item name.
With all that said, there is no need to even use the source schedule to access this information. Since the values are written to the items as labels you can simply use those.

It looks like you copied the expressions from the source's creation trigger to the exit trigger of the separator. This does not work for two reasons:
"current" is a reference to the object the trigger is defined on. So for the source, this can be used to find and access the schedule table node in the source's attribute tree. To access it from the separator's trigger you would have to refer to the source directly ("Model.find(...)") instead of using "current".
The parameter "rownumber" is only defined for the creation trigger of the source, where it denotes the row of the schedule table the created item corresponds to. This information is lost after the item creation. So to use it in the separator's trigger, you would have to store it on the item as a label.
Again though, there is no need to actually use the schedule table, since all needed values are assigned to the items as labels.

separator-test_1.fsm