S.Monteil PRS timing 1

November 2009 - Clermont team - Calorimetry meeting

Preshower timing / The cosmics and TED results / Collection of plots.

1. Overview of the timing settings. 2. Cosmics results.

3. Analysis of TED events: 64-channels board time. Relative alignment. Timing within a board. Comparison w/ october’s run

4. Fine time-alignment.

Disclaimer: most of the slides were already presented. I’ll move quickly to the addenda.

S.Monteil PRS timing 2

1. Overview of the timing settings:

Croc:

fine delay [0 : 25] ns [+ coarse delay (+ N cycles)]: currently set at 12 ns.

VFE:

t0VFE tunable phase [0 : 25] ns defining the integration

start accounts for both clear and WLS fibre lengths differences. One per board.

t0

Reset tunable phase [0 : 25] ns to sample the VFE reset signal (start at each LHC orbit on the same alternance) tuned empirically to match the t0

VFE.

S.Monteil PRS timing 3

ADC: tVFE

ADC tunable phase [0 : 25] ns to sample the analogical signal. 4 phasers for one board (one phaser deals w/ two ADC) tuned such that the difference (t0

VFE- t0ADC) is constant (there

is a plateau so no accuracy is required here).

FE_PGA: tADC

FE tunable phase [0 : 25] ns to sample the digitized data in the FE-PGA tuned to the same value on the whole detector.

tSPDFE tunable phase [0 : 25] ns to sample the digitized SPD

data in the FE-PGA tuned by Carlos to account for clock cables and fibre lengths differences.

1. Overview of the timing settings:

S.Monteil PRS timing 4

TRIG_PGA:

tECALFE 2 tunable phases to sample ECAL addresses

signal. Tricky point (cf APA story). Minimal authorized range = +/-3 ns around the current value, i.e if one moves ECAL by more than 3 ns w/ respect to PRS, we’re loosing electron and photon triggers.

+ some other phasers for neighbours, irrelevant here,

except transmissions from crate to crate. Cosmics and TED data told us that we don’t need to move PRS crates.

1. Overview of the timing settings:

S.Monteil PRS timing 5

2. Cosmics results: inner region

Comparison w/ ECAL time as given by CosmicsTrackTool.

S.Monteil PRS timing 6

2. Cosmics results: middle region

Comparison w/ ECAL time as given by CosmicsTrackTool.

S.Monteil PRS timing 7

2. Cosmics results: outer region

Comparison w/ ECAL time as given by CosmicsTrackTool.

S.Monteil PRS timing 8

The analyzed runs: Take the runs tagged as LHC/SPS clocks are synchronous. Run #50425: 30 shots. Run #50427: 37 shots. Run #50428: 70 shots. Run #50430: 45 shots. Run #50432: 89 shots. Run #50433: 90 shots. Run #50435: 14 shots.

Particle fluency varied among these shots (to match the tracker requirements)

3. TED events analysis - June 2009

S.Monteil PRS timing 9

Particle fluency distribution

3. TED events analysis - June 2009

Cosmics

Cut

S.Monteil PRS timing 10

ADC distributions in T0 and Next BC:

3. TED events analysis - June 2009

We can compute the asymmetry:

A = (ST0-SNext1)/ (ST0+SNext1).

Model: tPRS = (A-1.7)/0.11 [ns].

S.Monteil PRS timing 11

Typical response for one channel:

3. TED events analysis - June 2009

The typical r.m.s of a channel is better than 1 ns.

The PRS integrates 14 ns earlier than the arrival of LHC particles in wrong direction.

S.Monteil PRS timing 12

Typical response for one board:

3. TED events analysis - June 2009

The typical r.m.s of a channel is roughly the nanosecond.

S.Monteil PRS timing 13

3. TED events analysis - June 2009 INNER DETECTOR

Caution: the error bar displays the one sigma value of the fit, not the uncertainty on the mean value.

Bad fit due to too low statistics : mean value is -14.8 ns.

S.Monteil PRS timing 14

3. TED events analysis - June 2009 MIDDLE DETECTOR

Caution: the error bar displays the one sigma value of the fit, not the uncertainty on the mean value.

Too low statistics : mean values are -14.9 and -13.7 ns, resp.

S.Monteil PRS timing 15

3. TED events analysis - June 2009 OUTER DETECTOR

Caution: the error bar displays the one sigma value of the fit, not the uncertainty on the mean value.

MISSING VFE BOARDS.

NO PARTICLES

S.Monteil PRS timing 16

3. TED events analysis - June 2009

So far so good, a r.m.s of 0.3.

Looking at the channels distribution within a board:I am taking the average answer of each of the 64 channels and plot it. Let’s scrutinize first a typical board comprising 32 channels:

S.Monteil PRS timing 17

3. TED events analysis - June 2009

Not so bad, a r.m.s of 0.5 ns.

Looking at the channels distribution within a board:Let’s look at a « typical board » comprising 64 channels:

S.Monteil PRS timing 18

3. TED events analysis - June 2009

Clear double structure and a r.m.s of 0.8 ns.

Looking at the channels distribution within a board:Let’s look at another « typical board » comprising 64 channels:

S.Monteil PRS timing 19

3. TED events analysis - June 2009

An explanation:

Two clocks are distributed to the VFE boards. One for 8 chips.

A unique correction for 64 channels (so far) is applied to the integration time.

The effect is consistent with a skew effect (difference of signal propagation from one pair w.r.t to the other) of the 27 meters long clock cable.

Looking back at the data sheet of the cables, it happens that they are sold with a maximum skew of 7ns / 100 m (I had in mind 0.3 ns, gloups) . Actually close to what we observed.

This misalignment can be corrected by adjusting at the ns level the dedicated half-board clock phaser

S.Monteil PRS timing 20

3. TED events analysis - June 2009

The r.m.s is a good estimate of how bad we are:

Lack of statistics:

All these boards are in PRS_6 or PRS_7.

The effect is at maximum at the level of 0.8 ns r.m.s, close to the intrinsic limitation of the tuning : 1 ns step.

S.Monteil PRS timing 21

3. TED events analysis - June 2009

A conclusion:

Overall picture for the PRS channels/boards/crates timing interalignment is satisfactory.

There are board misalignments at the level at most of 3 ns.

For some boards, the channels are misaligned by about 1-2 ns. This is related to the bad skew of the corresponding clock cable. At the level of the nanosecond correction we can bring.

Corrective factors from TED events are ready to be set in the electronics.

S.Monteil PRS timing 22

3. TED events analysis - June 2009

Timing comparison ECAL vs PRS:

t_PRS=-13.5 ns in average. We integrate 13.5 ns before the arrival of LHC particles.

Define asymmetries as A = (ST0-SNext1)/ (ST0+SNext1) and suppose an ECAL zero asymmetry when signal is picked up (-12.5) ns w.r.t the particles. ECAL asymmetry = -0.2 => t_ECAL=(-12.5-1.6)=-14.1 ns. ECAL further delayed (w/ CAT) by +8.5 ns. => t_ECAL=-5.6 nsFine delays => (t_PRS-t_ECAL)=(12.0-16.0) = -4.0 ns. Time of flight : 0.90 m/ c => (t_PRS-t_ECAL)=-3 ns.t_PRS=-12.6 ns, in pretty nice agreement w/ the standalone estimate.

S.Monteil PRS timing 23

3. TED events analysis -

Timing comparison 0ctober vs June:

t_PRS=-9. ns in average.

ODIN was moved by 5 ns, w.r.t June.

S.Monteil PRS timing 24

4. A PRS/SPD scenario for fine time-alignment

Sequences and actions:

1) Currently, the PRS/SPD channels are inter-aligned with our best knowledge of the detector (Cosmics + TED events).

2) The first LHC data we’ll get will be splashes of particles from beam1 all arriving in time in all regions of the detector (set HV physics settings). There we have two actions:

1) Check the inter-alignment. Correct whether necessary (hopefully, it won’t be).

2) Measure with the asymmetry method the difference between the particle time and the Preshower VFE integration time. ODIN can be moved accordingly.

S.Monteil PRS timing 25

4. PRS/SPD scenario for fine time-alignment

Sequences and actions:

3) Prior to collisions, we set the cos(theta) correction, i.e the particle flight correction. This is at the level of VFE phase (3 ns max).

4) Comments:

If PS is giving the time to ODIN, we do not touch anything. If XCAL gives the time to ODIN, we’ll have to adjust the TTCrx fine delay to account for t_XCAL-t_PRS. We have there to be very careful to the impact on the electron and photon trigger response. We know the phasers of ECAL addresses to be critical (+/- 3 ns for the most critical channels).

PS might be suitable to give the time shift to ODIN since we’ll work with the Physics HV on splash events. Even if we are below, the time transit difference in the tube is negligeable. No further time correction.