It's often easier to find primary sources by searching for something specific, rather than searching by topic.
You may not be able to find good primary sources by running keyword searches in JSTOR or Proquest. Instead, do a bit of background research on your topic first. You might be able to identify:
Doing some basic background reading can help you identify good primary sources.
For example, let's say you're interested in the US women's rights movement during the mid-20th century, particularly the intersection of race and advocacy for women's rights as workers. You do some background reading, and:
Caveat: This is (obviously) a 20th century US example. So although this case offers one model of thinking about and finding primary sources, consider how your approach might vary depending on the time period, geography, and topic in which you're interested. To discuss this and get advice, reach out to your professors or librarian Simon Elichko.
Original Sources: Working with primary sources in their original form often means visiting an institution dedicated to preserving those sources - typically, a special collections library or an archive.
Reproductions of Sources: You might read scanned copies of historical newspapers using an online database, or consult a book from McCabe consisting of transcribed letters between historical figures.
A digital scan of a letter gives you more detail than a text-only transcription, but viewing a high-quality full-color scan is still different than holding a physical paper letter. An edited collection of letters reflects choices made by editors. It don't mean that certain kinds of sources are automatically "bad" - just that it's worth keeping these considerations in mind as you select and analyze sources.