Some thoughts on CESR bunch structure

dm smilgies 5/03

Basic facts

CESR has 183 buckets with a spacing of 14nsec.
The ring period, i.e. the time for one bunch to travel around the ring is hence 2.56usec.
The actual bunch length within a bucket is 65 psec.
 

Ultra-fast experiments

- using the 65psec bunch length -

The brute-force approach would be single bunch mode -
note: everyone else around the ring will hate you !!

If the timescale of the sample is less than the 2.5usec ring period,
e.g. all sample dynamics is over after 0.5 usec, you may want
to fill 3 buckets equally spaced (61 bunches or 854 nsec) which ups
the current somewhat.
 

Fast timing mode

- matched to the ORTEC MSC resolution of 100 nsec -

Now you do not need a source that is faster than your time resolution
(minimum dwell time) of 100nsec. Hence you can work with a bunch train
rather than with single bunches:
5 buckets get you 70nsec pulses, 10 buckets 140nsec pulses

Next question to ask: what is the time scale of my sample.
One train gets you a time window of 2.56 usec between pulses.
Two trains get you a time window of 1.28 usec between pulses.
Three trains get you 854 nsec between pulses.

If the latter is combined with a 10-bunch train, we would have 30 buckets in the ring,
which is not so bad compared to the usual 45 bucket pattern, and the other users may
be more willing to put up with fast-timing experiments this way.
 

Tricks and traps

First of all, big lasers have a lousy duty cycle of maximum 1000 Hz.
The second problem is the jitter of the laser trigger, i.e. when you send the "shoot" pulse, when will the laser respond.
(value for the KrF excimer laser in G3:
 typical value for a Ti-sapphire laser: )

Third problem is to make a clean fill of the storage ring, in order not to populate supposedly empty buckes.
Fourth problem is to prevent the ring from losing particles into empty buckets during a fill.

Fifth problem is suitable x-ray detectors:
The currently fastest detector is an avalanche photodiode (APD) which can achieve a shaping time of about 5nsec.
i.e. within a 65psec bunch, only one photon can be detected. This limits the maximum possible count rate to 1 count
per 2.5usec in single-bunch mode, i.e. 100,000 counts per second. For weak scatters (10e-6 efficiency) you still would
want to get as many photons on the sample as possible, so that you get at least 1 photon per sec. Good news is that an
APD can just about get a pulse from each bucket (5nsec versus 14nsec). A top-of-the-line conventional bicron detector
with a 0.3usec shaping time (Cyberstar) would only see 1 count per 10-bunch train maximum (300nsec versus 140nsec), so it may be worth while to switch to an APD for strong scatterering signals (small-angle scattering, reflectivity). A plastic scintillator with fast electronics may also do the trick.