I don't know if you have followed the discussions from the last years, but FPKM and RPKM are obsolete. Statisticians found that FPKM and RPKM are not the right values that can be used for proper statistics. It is a wrong way of normalizing RNA-seq data. In stead they suggest to use raw read counts, which can be used in tools like edgeR or DESeq2. Also, CPM would be better than FPKM or RPKM, for example you can use (log2) CPM values in limma with trend.
But to answer your questions.
You can't convert, you need to recalculate from raw data.
No not acceptable. To compare studies you need to correct for batch, which you can do with statistical programs such as edgeR and DESeq2.
The best advice I can give is to start with raw data, also from the already published data that you would like to compare your own experiment with. And continue with edgeR or DESeq2 or the like.