Thanks gab. I don't want to add any negative schema, but it really does seem for me that there is usually a reasonably long period of non-REM prior to each REM, even at the end of the night. I'm basing that on many nights of recordings using my Zeo headband. I found some consistency where I would wake about every 90 minutes at the end of REM, and then the Zeo would show about an hour before the next one. That seems to match my few attempts at WILD that were successful, where after a WBTB I found it took about an hour to work.
I will try exploring it again though.