I wonder if they're using some kind of EDFA or other amplifier for that delay spool. For 350µsec we're talking at least 14.3dB of loss going through that thing

I wonder if they're using some kind of EDFA or other amplifier for that delay spool. For 350µsec we're talking at least 14.3dB of loss going through that thing, in other words, only 3.7% of the signal is coming out the end in the best case scenario.