This website is no longer maintained. Its content may be obsolete. Please visit http://home.cern for current CERN information.
Extra BI events - Lars - Dec 07
For BI triggering etc, we’d like to have these also:
1) BLM-capture (study stuff)
2) Fast BCT capture (B1, B2)
3) BSRT capture
4) BWS-fly-the-wire (in-scan movement trigger)
5) Tune measure (B1, B2)
6) Head-tail measure (B1, B2)
Also it would make a lot of sense for me to have an event to be used to restart the PM buffers etc with before the beginning of each fill programmed by the LHC sequencer.
7) New-fill or whatever ..
Fill the LHC [ppt] - including injectors - Julian Lewis April 07
March 07 This morning we connected the new LHC timing in the CCR to the LHC distribution. The old DSC in the MNR can be suppressed (dmcrmtgd). So loading timing tables will really send events out to the control system. Julian
Requirements spec (Latest version - November 2006) Implementation 2006 Timing issues 06
Timing and Post Mortem (November 2006)
Old stuff
What does a slow timing system provide?
Foreseen requirements
Brainstorming sessions: 20th September & 4th October 2000
Fast timing, i.e. distribution of bunch and revolution frequencies.
Beam synchronized events for beam observation will be handled by the BST system. There will be a central gateway for slow timing events into the BST timing.
Schematic (Bruce Taylor)
Phase changes between PCR and P4 (J. Sladen). He did the measurements over one full year on the return loop PCR-P4-PCR, so twice the distance PCR-P4. The frequency he used was 352 MHz. The phase changes were:
1) Standard singlemode fibre = 2116 degrees
2) Temperature compensated singlemode fibre = 86 degrees
From these values you can either deduct the phase changes at other frequencies or propagation delay changes and according to me no other measurements would be justified.
(BT) Many thanks for these figures, from which I see that the performance of the temperature-compensated fibre is a great improvement. The 340 ps variation in one-way propagation delay corresponds to 3 steps of the TTCrx deskew circuits, whereas the 8.35 ns variation for the standard fibre corresponds to about 80 steps! It would be very useful to know the magnitude of the diurnal variation as against the total variation over the full year.
The TTC system requires only one fibre to transmit the encoded 40.079 MHz and 11.246 kHz clocks from S4 to the PCR transmitters (plus a second one for when the RF of the 2 LHC rings are slightly different). I appreciate that there may be several requests for these precious fibres and hope that the TimWG can consider allocating two for the link to the TTC transmitters if these are to remain located at the PCR.
From BT to PB....
Thanks for your interesting observations and I will reply very briefly to the question of the clock phase stability required by the LHC experiments.
The 40 MHz clock broadcast by the PCR TTC transmitters will be received at the experiments by the TTCmi crates and broadcast to the detector electronics through a total of about 16,000 TTCrx timing receiver ASICs. Each TTCrx provides for programmable deskewing of the clock within the 25 ns period in steps of 104 ps. The other signals such as the level-1 trigger and bunch counter reset are also coarse deskewed by the TTCrx in 16 steps of the 25 ns period (i.e., a total range of 400 ns) to allow them to be associated with the corresponding bunch crossings.
Although several methods will be used to set up the initial timing
approximately (databases of fibre lengths, measurements of electronics
delays and calculated particle times of flight to the different
locations in the detectors, distribution of test pulses over measured
paths), the primary method of synchronisation will be crosscorrelation
between the physics event occurrence pattern in each detector channel
and the known bunch structure of the LHC beams. Calculations have shown
that even for the subdetectors (the muon chambers) having the lowest
occupancy this should allow all the TTCrx deskews to be set within a
reasonable time. In order to compensate for phase wander due to local
electronics drifts and fibre length changes, the measurements will be
carried out continually whenever stable LHC beams are available and
individually-addressed deskew adjustments will be transmitted to all the
TTCrxs over the B channel of the TTC network itself.
With this modus operandi, phase drift of the 40 MHz clock received by
the TTCmi crates due to temperature changes of the SR4-PCR and
PCR-experiment fibres is just one of the components of the total phase
wander that has to be automatically compensated for by the TTCrxs. The
only requirement is that the rate of drift should not require undue
activity on the part of the synchronisation processors transmitting the
deskew parameter changes to the TTCrxs.
For the temperature-compensated fibres the total seasonal phase shift of
320 ps is only about 3 steps of the TTCrx fine deskews. So even if a
phase shift of this magnitude were observed on a daily basis it should
be readily accommodated. Of course the smaller the better, and it would
be very nice if the reason why the temperature change of these fibres
appears to be greater than expected can be found and corrected.
The measurements which I made on the normal singlemode fibre loop
B4-PCR-B4 are comparable to those which you have observed for the 4.6 km
SPS-CPS link. The total length of this fibre is 13 km and the diurnal
variation in propagation delay was less than 100 ps. I too was quite
surprised by the small temperature variation to which this corresponds,
as the fibre is routed through patch panels in 4 buildings and there is
a length in the corridor from the telephone exchange to my lab.
As there are programmable clock deskews in the modules which will
receive the clock from SR4 at the PCR, and in those which will receive
the clock from the PCR at the experiments, I suggested to the
experiments that I could control these deskews in feedback loops to
compensate for the changes of propagation delay with temperature over
these fibres. The experiments responded that since in any case they will
be using the local TTCrx deskews to adjust the phase of the clock
relative to the beam crossings such compensation is not required.
I fully agree that it would be reassuring to have measurements of the
diurnal variation in propagation delay over the fibres from the PCR to
each of the experiment areas to be sure that some accident of routing
does not result in an unexpectedly large daily change.
Cheers,
Bruce
"
P. Baudrenghien" wrote:
>
>
Dear colleagues,
>
>
following the TimingWG meeting on 20th April, I have gathered some
>
material concerning the fibres originally installed for LEP. (Thanks to
>
Thomas ...). The references are:
>
>
1. E. Peshardt, Transmission of a stabilised RF phase reference over a
>
monomode fibre-optic link, Electronics Letters, 31st July 1986, Vol 22, No
>
16, pp 868-869
>
>
2. E. Peshardt, J. Sladen, RF Synchronization of the LEP accelerator, Int.
>
Conference on Frequency Control and Synthesis, U.K., April 1987, pp
>
157-161
>
>
3. E. Peshardt, J. Sladen, Phase compensated fibre-optic links for the LEP
>
RF reference distribution, CERN LEP RF 89-29, March 1989, presented in
>
IEEE PAC Chicago 1989, pp 1960-1962
>
>
4. L. de Jonge, J. Sladen, RF reference distribution for the LEP energy
>
upgrade, CERN SL/94-34 (RF), 5 th July 1994, presented to EPAC 94 London
>
>
These references are public, but if you have problems to get a copy, I can
>
give you one.
>
>
>From these we get the following results:
>
a. Over one year, the delay variation from PCR to SR4 (9.5 km) is 8 ns
>
(1000 degrees at 352 MHz) for the normal fibers and 0.32 ns (40 degrees at
>
352 MHz) for the compensated ones [from fig 3. of Ref 4].
>
b. This is in good agreement with a seasonal temperature change around 20
>
degrees Celsius (9.5 km at 0.66 c gives 48 microsec total delay. Normal
>
fibers have a thermal expansion coeff. around 7 ppm/degree C -> 24 degree
>
C gives 8 ns. Compensated fibers have thermal expansion coeff of 0.4
>
ppm/degree C -> 17 degree C gives 0.32 ns.)
>
>
Considering that these fibers are laid in (relatively) deep trenches, the
>
20 degrees temperature variation is UNEXPECTED. The SL/HRF group is
>
operating a fiber optic link from the SPS to the CPS (4.6 km, monomode
>
fiber). The total delay is 23.2 microsec. The thermal expansion coeff is 7
>
ppm/degree C. We have tracked the phase variation from 1st May to 31st
>
May: the daily variation was 3.6 degrees at 200 MHz corresponding to 0.05
>
ns. The yearly variation was of the same order (an additional 0.05 ns).
>
This measurement suggests a variation of the temperature ACTUALLY SENSED
>
BY THE FIBER AROUND 1 DEGREE CELSIUS ONLY. We explained this low figure by
>
the fact that the fiber is resting in an underground trench where the
>
temperature stays very constant over the year. Before 1998 the link was
>
ending in a 269 meters long fiber sitting on the surface, in the sun, and
>
this small portion alone was introducing 3 times more phase variation than
>
the underground 4.4 km ... This has been corrected since.
>
>
So, I think we should
>
a. For each client (Experiments, BI, RF, ...) try to specify the required
>
stability of the 40 MHz (200 MHz, 400 MHz) transmitted on the fibers, both
>
short term (daily) and long term (seasonal)
>
b. Explain why the LEP fibers suffer this large temperature variations. J.
>
Sladen told me that the variations were not identical for two equal length
>
fibers going to different buildings. Maybe a small portion of the link is
>
responsible for most of the phase change.
>
c. Measure the daily/seasonal phase variation from PCR to SR4 round trip
>
(and eventually other buildings).
>
>
This problem has a strong impact on the stability of the fast timing
>
signals for the LHC. I think that we should organize a meeting (TimWG +
>
other concerned: ST) soon to get started on it.
>
>
Cheers to all
>
>
Philippe