Mar 23, 2006

Kicker Timing Requirements for the Two LCLS Kickers:
(1) LI25 Kicker (BXKIK) which kicks the beam into the bunch length 
    measurement screen.
(2) LTU Kicker (BYKIK) which kicks the beam into the safety dump 
    instead of the undulator.

General Timing:

(1) For stability, kickers must be triggered at the maximum rate of the beam 
    for a particular run.  For example, if we have a maximum rate of 30hz for 
    many months (or maybe weeks), then we trigger the kicker at 30hz.  If we 
    have a maximum rate of 120hz but we want to run at 10hz for a few days, 
    we still trigger at 120hz.  

(2) The capacitor bank has a maximum number of discharge/recharge cycles 
    before the caps need replacing so we must minimize kicker pulsing when 
    possible.  For example, if we stay rate-limited on purpose for more than 
    a few days, then timing should be reconfigured to trigger at the lower 
    rate.  Once we go back to full-rate, we then put the full-rate 
    configuration back.   Timing configuration changes are done on user 
    request.  However, when the kicker transitions from a lower rate to a 
    higher rate (and maybe vice-versa), x minutes must pass before allowing 
    the beam to run at the new rate to allow the kickers to stabilize.  

(3) BEAM MUST NEVER BE ALLOWED TO HAPPEN AT A RATE FASTER THAN 
    THE KICKER TRIGGER RATE.

(4) Even though the LI25 kicker normally kicks at 1hz or less onto the screen, 
    it must still run at full-rate since sometimes the user will request 
    consecutive bunches at full-rate for diagnostics reasons.   ???Do we need
    to support two modes of operating the LI25 kicker - one for when there is 
    no expectation of multiple bunch measurement (ie, during normal operation) 
    and the other for commissioning or testing periods.  And allow a short 
    amount of time to switch between one mode and the other on user request.???

Pulse-to-Pulse Timing:

(1) When the kicker trigger is coincident with the beam pulse, then it'll 
    kick it into the dump (for LTU) or screen (for the bunch length 
    measurement screen).  When a kick is NOT desired, the trigger is set 
    later than the beam by a small amount.  The trigger has to be sent right 
    after the beam has passed through the kicker magnet so that the next beam 
    pulse is not disturbed by any fields rising inside the kicker magnet.  The 
    kicker system needs about 7.3msec to pulse and then recharge.  At 120hz, 
    if one pulse DOES NOT need to be kicked and the next pulse DOES need to be 
    kicked, the time between those two triggers will be somewhere between 
    7.3 and 8.3msec.

(2) For the LTU kicker, once the kicker pulser has been fired, any other 
    triggers to it within the time it is still ringing (~2.5 msecs) will have 
    no effect.

(3) To kick the beam, the beam must arrive at the same time as the high flat 
    top of the kicker pulse which is around 25% of the total pulse time 
    (ie, 0.25 msec of a 1 msec pulse).  The flag top is >1us.  Depending on 
    the kicker pulser operating point (mainly its peak current) and on other 
    factors that change dynamically with time and temperature, the trigger 
    delay will need to be tuned (ie, knobbed) for each kicker to find the 
    right spot.  Tuning will be required periodically while the machine is 
    running.  It's probable that a slow feedback loop is needed to 
    automatically adjust this delay.  (The kicker pulse peak current will also 
    need to be adjusted periodically or automatically using slow feedback but 
    that'll be done by the magnet system, not the timing system).  The ability 
    to save and restore trigger delays is required.

Decision when to Kick:

(1) The LI25 kicker is driven by the needs of the user who can choose one the 
    following options:
    (a) Kick one or more consecutive pulses (usually one) at a rate less than 
        or equal to the beam rate until told to stop.
    (b) Kick one or more consecutive pulses in a single shot as soon as 
        possible (do we need a user-settable delay for when the single shot 
	actually starts?).

(2) The LTU kicker is driven by the needs of the user and MPS - when either 
    requests a kick, the kick is done.  Two triggers are provided to MPS 
    which then decides which to pass through to the kicker pulser:
    (a) One trigger is provided at full rate to MPS coincident to the beam 
        pulse.  MPS will allow this trigger if a fault was detected on the 
	previous beam pulse or an old fault is not yet reset.
    (b) A second trigger is provided to MPS to satisfy user kick requests.  
        MPS will allow this trigger if all faults are cleared.  The user can 
	request the following: 
	(i)  One or more consecutive pulses kicked as soon as possible. 
	     ??Is this really required??
	(ii) Rate-limiting in the undulator (including zero rate).  
	     Some beam pulses are kicked at a rate less than or equal to the 
	     beam rate until told to stop.