Clock Monitoring at Green Bank
[as of July 2008 -- by F.Ghigo, NRAO]
Revised April 2017 (F.Ghigo)
Revised June 2017 (F.Ghigo)
Herewith a description of the clock measurement system used
at the GBT in Green Bank to record offsets between various clocks.
How To Adjust the ST Maser Rate
Sources of Time
- NEW as of Feb 2014: CNS GPS receiver (descendant of the TAC)
- the GPS antenna from the roof goes to both the CNS and the TAC. CNS 1 pps output goes to the time logging system.
- The monitor and display computer for the CNS is windows machine "GBTIME", located behind the main operators console.
- note: the antenna cable delay is set to 57 ns and the internal receiver delay to 37ns (data-->set timing parameters)
- GPS receiver: Tom Clark's "Totally Accurate Clock"
- The TAC detects up to 6 GPS satellites and provides a 1 Hz pulse on the UTC second at the input to its antenna.
- "SiteTime" : the observatory master clock (TRAK Systems model 6464), driven by a 5 MHz reference signal from the Sigma Tau maser.
- The TRAK provides a 1 Hz pulse which is distributed to many backends (refer to 1pps cable delays ).
- The TRAK also provides IRIG time signals which are distributed to many devices including the ntp time server on "yed".
- The Sigma Tau Maser provides stable reference frequencies at 5, 10, and 100 MHz, distributed around the site for stabilizing LOs.
Clock Logging system
The 1Hz pulses
- Site1Hz : the reference 1Hz from the TRAK master clock.
- TAC : the UTC 1Hz from the TAC GPS receiver.
- CNS_GPS : the UTC 1Hz from the CNS GPS receiver.
- GBT_VLBA : The 1Hz generated at the VLBA formatter.
- RA-return: The 1Hz return from the RadioAstron receiver on the 140-foot
- GBTRtn : The 1Hz return from the GBT receiver room (This is twice the one-way delay; typical value 28.9 us).
The various 1Hz inputs can be selected as inputs to a RACAL-DANA 1991 counter.
The counter measures the time delay between the arrival of a pulse at its "A" input until a pulse arrives at the "B" input.
|Logging System Inputs (revised April 2017) |
|A ||B |
| Site1Hz || Site1Hz |
| Site1Hz || TAC |
| Site1Hz || CNS_GPS |
| Site1Hz || GBT_VLBA |
| Site1Hz || RA |
| Site1Hz || GBTRtn |
| TAC || Site1Hz |
| TAC || TAC |
| TAC || CNS_GPS |
| TAC || GBT_VLBA |
| TAC || RA |
| TAC || GBTRtn |
The inputs to the counter are selected from a switching matrix by the sitetime computer.
The inputs to the matrix are as follows:
| 1pps input || signal |
| X0 || Site 1Hz |
| X1 || TAC |
| X2 || not used |
| X3 || CNS_GPS |
| Y0 || GBT_VLBA |
| Y1 || RA (round trip to 140') |
| Y2 || GBTRtn (round trip to GBT) |
The "SiteTime" computer logs the time offsets, and also logs the maser monitor points, and the round-trip delay
The logging may be viewed in real time with Cleo: cleo sitetime
Typical cleo sitetime display
SiteTime records these clock offsets every minute and stores them in FITS files in the directory /home/gbtlogs/SiteTime-OnePps-OnePpsDeltas/
The maser monitor points are stored in: /home/gbtlogs/SiteTime-Maser-MaserList/ -- logged once per hour.
The round-trip delays are stored in: /home/gbtlogs/SiteTime-Rtpm140-Rtpm/ -- logged once per 10 seconds.
Several cron jobs run as fghigo on bratac (as of June 2017)
- monmaser (~fghigo/CLOCKS/monmaser.c) runs once a day
- reads the latest maser monitor points from the fits log file (SiteTime-Maser-MaserList) and updates the web page: monmaser.html
- today's maser monitors are appended to a log file : http://www.gb.nrao.edu/~fghigo/timer/monmaser.logs
- maser monitors may be ploted with an idl program: ~fghigo/CLOCKS/plotmaserreg.pro
- monacc1pps runs once a day.
- Read the OnePpsDeltas fits files
- Take hourly medians of the clock offsets.
- Write an archive file with hourly medians for all data in the current month up to now.
- These archive files are written into the directory:
- file names of the form: yyyy_mm.t00new
- Uses the program "rfclock.c" (~fghigo/GBT/fits/)
Pulsar timing files
Pulsar observers use a "time_gbt.dat" file with the clock offset for each day.
These are generated from the "yyyy_mm.t00new" files once per day.
Note: Jan 2018: As of Oct 3, 2017, the CNS receiver 1pps is used instead of TAC,
because the TAC has gone haywire. The "time_gbt.dat" file is now based on the
1pps from the CNS. There were jumps in the time_gbt.dat due to changing to the
new CNS and from adding time offsets in the CNS to try to connect to what the
continuation of the old TAC values would be. So as of Jan 2018, the time_gbt.dat
file has been modified to remove all the non-real offsets. All time jumps
should be real and due to re-synchonizing the 1pps pulse.
- flippsrtime.csh creates "time_gbt.dat" and stores it in "/home/pulsar64/tempo/clock", and also in "/home/ftp/pub/fghigo/timer"
- flip3psrtime.csh creates "time3_gbt.dat" and stores it in "/home/pulsar64/tempo/clock", and also in "/home/ftp/pub/fghigo/timer"
- The "flip" is because the sign is flipped from the way we store Site1Hz vs GPS to the way the pulsar people like it.
- "time3_gbt.dat" uses the pulse from the new CNS GPS receiver. This has replaced the TAC receiver.
- some day we will drop time_gbt.dat and store only time3_gbt.dat.
- These scripts use programs "psrtime2.c" and "psrtime3.c"
- "psrtime2.c" adds a time offset of -167 ns after 15 Dec 2009 to account for having moved the TAC to a different location and changing the length of cable from the antenna.
- "psrtime3.c" : no time offset is added; the CNS internal timing parameters for cable delay and receiver delay should make the results close to time_gbt.dat.
- CNS : antenna cable delay = 57 ns
- CNS : internal receiver delay = 37 ns.
- (set these delays with menu item "data-->set timing parameters")
- to match up the CNS delay with the continuation of the previous TAC record,
the "intentional extra early offset" was set to 0.089 on Dec 14, and then to 0.019 on Dec 21, 2017.
- Oct 9: "psrtime2.c" no longer adds the -167 ns offset.