As I said: IIRC.
But for your sake I dug a bit.
The 80% memory came from this: https://www.nationalreview.com/news/china-supplied-faulty-coronavirus-test-kits-to-spain-czech-republic/
Though in the Czech case they perhaps included false positives as well as false negatives in the 80% figure.
That 30% figure goes back to some early Chinese studies, also mentioned here: "Early research out of China stipulated the false negative rate could be around 30%"
For that 1-2% figure, go here: "Large test manufacturers, like Roche, Abbott Laboratories, and Cepheid, all have PCR tests that achieve 100 percent sensitivity and specificity"
So, even better than I thought, but of course, in the field it depends on handling them properly.
you first need a better test, then much testing and many statistics to THEN know how many false positives or false negatives there were
also time goes on and things like covid come and go - so good luck with that