• Main INDEX
  • Monthly INDEX
  • PREV
  • NEXT

    User name gilman

    Log entry time 17:48:42 on July19,2001

    Entry number 66761

    Followups:

    keyword=helicity signals chapter III

    We look at the toggle helicity signals and notice a problem.
    Many glitches were observed, with the following scenario:
    + signal is true, goes false, 6 us later the + signal goes
    true for ~80 ns. (It is very asymmetric between the two helicity phases.)

    By changing one of the Phillips latches so that the
    input helicity starts the 500 us delay and resets the latch,
    we determine that the Phillips does not give out the same
    short glitch pulse (10-15 ns) that the LeCroy did.
    (That is, when the start / trigger arrives and the stop / reset
    is present, the LeCroy gives out a 10-15 ns glitch pulse
    while the Phillips has no output.)

    --- differnces in number of helicity pulses counted ---

    This implies that in toggle modes the left arm should
    count only ``good'' helicity pulses, while the right
    arm counts ``good'' pulses and the number of glitches.
    Bob verifies that in toggle mode the difference in (ungated)
    counted pulses is large for right arm, but 0-3 in right arm.
    (Should be 0-2, but a few 3 counts, okay...)

    For the pseudorandom mode, because of the different glitch
    time structure, the left arm / Phillips will only count one phase of the
    glitch, whereas the right arm / LeCroy would count both phases
    for a glitch... thus the asymmetry in number of helicity pulses
    counted by the scalers should be larger for left arm,
    since the glitch rate is different in the two phases.

    --- differences in time scalers / events are gated ---

    For toggle mode, here is what happens when there is a glitch.
    Assume (most often case) helicity + is true, goes false at ``t=0'' for 6 us,
    true for 80 ns, false for ~33 ms. For the + latch, the ginal is
    true until ``t=0'' goes false, and stays false for ~33 ms.
    For the helicity - latch, it is false until ``t=500 us'' when it goes
    true. The glitch thus has no effect on the latched signals and leads
    to no time difference.

    It is different in the pseudorandom mode, where the glitch ends up
    leading to ~500 us being cut out of the 66 ms, e.g., helicity + window.
    If there is one glitch every ~10 s, and almost all glitches are on the helicity +,
    this leads to a time asymmetry of order time+ - time - / sum ~ 500 us / 10s
    ~ 10^-4, much less than the observed 10^-3 time difference. Perhaps the
    effect is a few times larger, but perhaps there is something else
    happening as well.

    (RG, with JP, WK, ROM, BR)