This is an output transformer. If I use it the "otherway" around, won't it then become a "step down."? I would just like to make it a 1:1 and therefore could wind it 900:900 or 1100:1100.
But I think Gyraf's inductance idea is the correct way to think... The source impedance is very low. The output of a discrete opamp, like a JH990 or something
Following gyraf's comments, inductance has two main consequences.
It must be high enough to provide good LF response
in conjunction with the source that drives it.
It must be high enough to limit LF distortion at an acceptable value.
In both respects, one has to decide what LF is and what deviation of frequency response and what distortion level is acceptable.
A transformer designed for 40Hz LF response is very different than one designed for 20Hz.
If you designed the transformer from scratch, you would have to choose the core first and then determine the turn numbers, but since you have an existing transformer, you have to measure it first.
Using the circuit that will drive it, what is the LF frequency response? What is the LF THD level? Note that driving the 1100 turns side will give slightly better results than driving the 900 turns side.
Depending on the results, you may decide that the LF performance is adequate from both sides, so you can use 900 turns as the reference, or if both sides do not give good performance, you may have to increase the numer of turns, until you get proper performance, but remember that increasing the number of turns may result in HF loss. All xfmrs are a compromise between LF and HF response.
Now you may find that driving the 1100 side is good, but the 900 is not, then you would choose 1100 as the reference.