Bill Christie [christie@bnl.gov]

To:

 Visser, Gerard‎; Oleg Tsai ‎[tsai@physics.ucla.edu]‎ 

Cc:

 rfatemi@pa.uky.edu‎; Eleanor Judd ‎[egjudd@lbl.gov]‎‎; Tonko Ljubicic ‎[tonko@bnl.gov]‎‎; Jacobs, William W‎; Jack Engelage ‎[JMEngelage@lbl.gov]‎‎; Hank Crawford ‎[HJCrawford@lbl.gov]‎‎; Stephen Trentalange ‎[trent@physics.ucla.edu]‎‎; Zhangbu Xu ‎[xzb@bnl.gov]‎‎; Suvarna Ramachandran ‎[suvarna.r@uky.edu]‎‎; James Adkins ‎[kevin.adkins@uky.edu]‎‎;
 

Hi All,
     RHIC got the their clocks back up and running for STAR this
afternoon, and Oleg, Gerard, and I looked with a scope at clock signals
at various points in the system. What we found was that there was a
wider width in the distribution in the clock ticks if we used the RHIC
Clock than if we used the local clock.
     To measure this width Gerard triggered on a rising edge of a clock
pulse at the BEMC electronics crate. He'd then vary the time that would
pass  from 107 ns (one clock tick) up to about 4 us, and then plot the
distribution of the arrival times of the clock tick. For the local clock
the width of this distribution didn't change noticeably over this whole
range. For the RHIC clock the width of this distribution was about two
to three times wider after ~ 1 us, and then didn't grow noticeably for
times up to the 4 us.
     Running the trigger with the Jet patch trigger the trigger didn't
fire with the local clock, and with the RHIC clock it fired at the
maximum trigger rate (~2700 Hz).
     We then looked at these clock distributions at the TCD front panel
output, and the RCC2 front panel output and observed the same effect,
with slightly varied actual widths, but qualitatively the same
conclusion. We then looked at the distribution from the output of the
optical module that the RHIC clock arrives on at the South platform and
saw the ~same wide width.
     The next thing we did is change the RHIC clock so that it came from
one of the channels on the Blue ring V124 module in the DAQ room, as
opposed to the Yellow V124 module. All other links  in the clock path
were the same. We observed a reduction in the width of the distribution
by about a factor of two (I have pictures for exact values), as measured
out at the platform. Running the Jet patch trigger with this Blue RHIC
clock we saw that the trigger rate dropped from the trigger limit pegged
value of 2700 Hz to a few Hz.
     We take these observations as evidence that we now understand the
source of the clock source dependence on the trigger rate of the Jet
patch trigger! I'll ask C-AD to change out or fix this Yellow V124
module, and then we can check this again at a future date.
     Earlier in the day Oleg took fresh BEMC pedestals using his
standard pedestal technique. We collected ped_as_phys runs with the
local clock, the Yellow RHIC clock, and the Blue RHIC  clock. Oleg will
pass the file names to Rene so that we can see how the Offline generated
pedestals look for these three cases, and for comparison to the new BEMC
online pedestals that Oleg generated. The results of this analysis will
determine if we have, as suspected, a separate issue related to the
determination of the pedestals.
     Greetings,
         Bill

---
GV adds:
This is the 9-cycle period jitter as measured at a BTOW crate controller clock monitor output, with normal RHIC clock connections (V124 Yellow channel 8):

This is the 41-cycle period jitter, same conditions:

This is the 9-cycle period jitter measured at BTOW crate controller clock monitor output with RCC2 local clock:

When we switched to "alternative" RHIC clock (V124 Blue channel 7) the jitter was improved by roughly a factor of 2 I think. However, that comparison was made at the RHIC optical link receiver front panel monitor output (on platform). I don't have a picture of it, sorry.

-----
Bill's pictures from the clock at RHIC opto receiver front panel output are in the attachment.