G. RizzoMarch, TS resolution needed in L0 striplets Assume want to have a cut of 50 ns on offline time window (  t) to keep the occupancy <= 10% (2 MHz/striplet from background in LO) In BaBar the time window cut  t was 200 ns and the TS_clock (BCO) was 67 ns (15MHZ). Just estrapolating from this one could require a BCO clock at 60 MHz to reduce the time window cut to 50 ns. I tried to understand the width of the BaBar time window cut to see if we can accept a slower BCO. The following assumptions should be verified!

G. RizzoMarch, BaBar offline time window cut This time window  t depends on the resolution of the t0 of the SVT hits:  t0 –t0 is the corrected time of the SVT hit calculated from the timestamp (threshold crossing time) and the information on the pulse height (from TOT). –Two contribution to this t0 resolution Typical  t0 extracted from BaBar data (next slide)  t0 ~35ns for inner layers with 200 ns shaping time  t0 =50ns for outer layers with 400 ns shaping time In BaBar the time window cut was  t~4-5  t0 In BaBar  TS =67ns/sqrt(12)=19 ns and from data one can extract the contribution from TOT correction –  TOT_corr =30 ns for 200 ns shaping –  TOT_corr =46 ns for 400 ns shaping

G. RizzoMarch, BaBar data SVT hit Corrected_time_Phi The time of SVT phi hits relative to the event time. This time is corrected for time-walk and the relative DCH-SVT offset. One plot for each SVT layer. TS_clock=67 ns (15MHZ) L1-2-3 RMS~0.5*TS_clock~35ns L4-5 RMS~0.75*TS_clock~50ns

G. RizzoMarch, SuperB offline time window cut In SuperB, with 25 ns shaping in Layer0, the contribution of the TOT correction is probably very small  TOT_corr =5 ns (is this a reasonable scaling from BaBar? Need to be more conservative?) Assuming for SuperB a given TS_clock the SVT t0 resolution can be calculated from The TS_clock required to get 50 ns cut (>4-5  t0 ) is reported in this table. –TS_clock > 30MHz seems required. BCO (TS) f MHZT BCO (TS) TS resolution TS/sqrt(12) SuperB TOT correction resolution (with 25 ns shaping) SVT t0 resolution Time window cut = 50 ns Time window/t0 resolution E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E+00

G. RizzoMarch, Analog info for SVT hits Analog info on pulse height is useful in SVT: –to improve spatial resolution w.r.t digital response –to correct timewalk and improve time resolution on SVT hits –to measure the dE/dx for low momentum tracks (high energy deposition --> high dynamic range required MIP) –Other less critical motivations Two options could be evaluated: –Time over threshold (as in BaBar) –Flash ADC (today not so difficult) Initial thoughts in next slides

G. RizzoMarch, Time over threshold (TOT) Assume shaping CR-RC 2 (as in the ATOM chips) with  p varying from 25 ns (L0) to 1 us (L4-5). Which TOT clock period is needed? –The ratio  p /T TOT will affect the spatial resolution and the SVT hit time resolution How many TOT bits do we need? –This determines the dynamic range with a given TOT clock –For pulse height of 10 MIPs TOT= 4.5  p Assume no saturation and theoretical TOT response!

G. RizzoMarch, Which time over threshold (TOT) clock? Spatial resolution From studies performed for BaBar TDR (BaBar Note-161): –Spatial resolution depend on the ratio between the shaping time  p and the TOT clock_period:  p /T TOT –The gre.ater is the ratio the more closely the time- over-threshold method should resemble a perfect analog readout. –With  p /T TOT =1 the resolution is already fairly good and there is a 20% improvement having this ratio =10. In Babar was = 3 in all layers. (skip control set T_TOT/T_TS) Lower limit on TOT clock frequency imposed from spatial resolution=1/  p –Huge variations among layers –L0=25 ns L1-3=50 ns, L4-5=1us shaping

G. RizzoMarch, Which time over threshold (TOT) clock? Dynamic Range (dE/dx) If we want to have a dynamic range of 10 MIPs (probably ok for dE/dx) the max TOT ~ 4.5  p With N bit the max value of TOT is 2 N TOT with N bits, with a given dynamic rage, imposes an upper limit on TOT clock freq. BaBar experience: F_TOT= 3 x1/tau_peek in all layers. This ratio is ok for resolution and dynamic range with 4 bits. In SuperB Layer0 TOT could be also 60 MHz (no dE/dx) –Is that ok for time walk correction? =3.56 with N=4 bits layer tau p ns Min TOT freq MHz (1/taup from spatial resolution )N bit Max TOT/tau p for 10 MIP Max freq TOT MHz (for dE/dx dynamic range req) F TOT MHz sugge sted from BaBar

G. RizzoMarch, Analog info with flash ADC Pros Better calibration w.r.t TOT(?) TOT clock is not needed (different among layers) Hit data is ready (pulse height info) at tau_peek while with TOT hit data is available later (for high ph ~ 10 MIP hit ready ~4xtau_peek Cons More difficult implementation? More space w.r.t the TOT logic? To have high dynamic range (10-15 MIP) with a reasonable resolution (0.2 MIP) ADC need 7 bits, while 4 bits are needed in TOT, assuming no deformation from logaritmic response of TOT vs charge. (worse resolution for high charge with TOT). N bit/hit (25-28) –7 strip addres, 9 TS, 1 data valid, 4 chip address, 4 TOT or 7 ADC –Larger (+15%) memories inside the chip (maybe even for Layer 0 this is not a big problem, still need to evaluate the area needed for buffers. NEED INPUTS FROM Pavio