The oscilloscope trigger is more reliable at slower time bases.
   Measuring at 20uS, I can massage the trigger to detect the first
   voltage drop:
   [1]https://ipfs.io/ipfs/bafybeiggijiystskmrgdocvczfoqqulvfnzuktle7cuwy7
   3kqpjyo54tkq
   It turns out the voltage scale is wrong in the pictures, because I
   tried to zoom out by setting it to 10. I think it only affects the next
   recording, unsure. The gradiations were set to 1V for the recording.
   Farther to the right there are visual corruptions that go off the scale
   and report multi-ten p2p voltages:
   [2]https://ipfs.io/ipfs/bafybeibyzmeqcqa24rkib6gleg74jmg44ztny4vev23qg6
   ua7jthk5b4ym
   This seems to happen associated with viewing a signal farther into the
   buffer. I might guess some kind of memory corruption. It makes sense
   that memory corruption would appear this way.
   My first idea was that the signal could be oscillating so fast that it
   misleads the analog-to-digital converter. I'm guessing this isn't the
   issue, because after measuring a couple times it seems more associated
   with buffer position.

References

   1. https://ipfs.io/ipfs/bafybeiggijiystskmrgdocvczfoqqulvfnzuktle7cuwy73kqpjyo54tkq
   2. https://ipfs.io/ipfs/bafybeibyzmeqcqa24rkib6gleg74jmg44ztny4vev23qg6ua7jthk5b4ym