Time-based fx (reverb, delay, etc) sound best after the entire
amplification stage... In other words, post-powertube or post-speaker.
Yes, doing it this way sounds superior even over the useful,
yet still half-assed way of going through an amp's FX loop.
The term "best" here means "clearest" - as in clean repeats, etc...
You can achieve the former (post-powertube) by running the speaker
out to dummy load which can convert the signal to line-level. From
there you go to your time-based FX, then another amp, then to
your final speaker. This is typically a live playing setup. Typical
dummy-load with line-level out would be a device like a Hot Plate.
The way to achieve the latter (post-speaker) is simply the same
way they do it in the studio: adding FX at the console.
Edward Van Heusen is a famous example of the former.