Why I assumed that the managers would choose the new all
optical timing scheme:

 " My understanding of points related to the Timing Distribution
   scheme in random order:

 1. We need redundancy in the Timing Distribution system.
    We can not allow a whole String or Group of Strings to
    fail because of a failure in the Timing Distribution
    system.  The new Timing scheme has a natural path to
    provide redundancy.

    To achieve overall reliability we know that serious
    systems,  e.g. commercial airliners, expensive satellites,
    ...   include redundancy in their vital systems.

 2. We need to control the amount of power used by the
    Timing Distribution system.  I believe that the
    new scheme uses less power.

 3. Something that no one seems to say is that if the new
    timing distribution scheme is used,  then there is
    nothing really left for the TOMCat to do on the DK card.
    Continuing to use TOMCat just wastes power and reduces
    reliability because of the increased parts count.

 4. If the new scheme is selected,  then instead of making
    TOMCat would Michael be interested in making all of the
    components for the new timing distribution system that
    go into the Junction Boxes ?

 5. The new Timing Distribution scheme needs one
    straight forward card in the Junction Boxes and it
    can be the same card at the various stages of the
    Timing Fanout.  This card could also be used as the
    timing source in "stand alone" test setups.

 6. In the new timing system  all links are optical.

 7. In the new timing system,  the ability to measurement
    the delay along each link is a built in feature of the
    components that are used to implement the links.

 8. In the new timing system,  the same timing transport
    components  and  delay measurement system   are used
    for all of the links  including the links to  all  of
    the individual modules.

 9. The old  "some optical plus some copper"  distribution
    system requires two methods for measuring delay  and
    the delay to a given module can involve different
    number of links  and  can change depending on how a
    given module is operated.

10. In my engineering mind it was never proven that the
    copper timing distribution links were going to work -
    especially near the bottom end of the main cables
    were there is the most noise on these 50 meter links.
    The input of each 100 V to 5 V DCDC converter makes
    about 30 mA of noise in the region of 300 kHz.  Will
    a copper timing link work when it is bundled in next
    to  20 of these 30 mA sources of switching noise ?

11. Timing distribution systems work best when they are
    kept as simple and uniform as possible.  They do not
    work well when they include a longer than necessary
    series of links  or  different types of links.

12. The optical components inherently work in the 1 GHz
    range  --->  reasonable expectation of 1 nsec timing.

    The components in the old copper links make it up to
    a few MHz or something like that --->  magic required
    to get 1 nsec timing.

13. Perhaps the worst thing that one could do would be to
    build an overall system that uses both the old timing
    distribution scheme in some sections  and  new timing
    distribution system in the remainder of the detector."