Mozaic trigger system for high transverse momentum physics

Mozaic trigger system for high transverse momentum physics

A.Fulop(ELTE), G.Vesztergombi (KFKI-RMKI)

CHEP

PragueMarch, 2009

Motivation for new measurements below = 20 GeVs

Practically no high or medium Pt data between Einc = 24 and 200 GeV

Mysterious transition around 80-90 GeV: convex versus concave spectra

Energy threshold for Jet-quenching?

Emergence of Cronin-effect in pA interactions is completely unknown

energy dependencecentrality dependenceparticle type dependenceparticle correlations

Production of Upsilon (9.5 GeV) particles near the threshold.

NA49 (CERN) results at 158FODS (IHEP) at 70 GeV

Beier (1978)

Special requirements for Y-> e+e- and high pT

Extremely high intensity - Pile-up

Segmented multi-target - Relaxed vertex precision

Straight tracks - High momentum tracks

DREAM: 109 interactions/sec

z [cm]

x,y

[cm

]

Px=Py = 1 GeV/c; Pz= 5 GeV/c

Px=Py= 3 GeV/c Pz = 10 GeV/c

High ( > 5 GeV/c ) momentum Straight track

MAPS vs Hybrid

Vertex resolution: dz = 1 mm, dx,dy= 0.05 mm

High intensity: radiation hard

Practical 4+ 2 + 3 = 9 planes ( 4 Hybrids + 5 strips)

Selectivity depends on the availability of TOF information

i

i+1

j

j+1

3 dimensional scheme

k=1

k=2

k=3

Mosaic cells in plane “k” : M(i,j,k)

(i,j) Corridor contains: M(i,j,k), M(i,j+1,k), M(i+1,j,k), M(i+1,j+1,k) k=1,2,3

s = sqrt(XX*XX+YY*YY) - delta

delta

Sagitta:

10-20 cm track sections are practically straight fractals

(XX,YY,ZZ)

4 hybrids 2 + 3 strips

2 4 6Basic planes

* * ***

* ** *

Silicon planes

Basic planes: #2 = (x2,y2,z2) pixel , #4 = (x4,y4,z4) pixel, #6 = (x6,z6) strip

Parallel processing: CORRIDOR # corNum

Straight tracking in #2 and #4 planes in space => (mx,bx) and (my,by) Approximation: starting direction is given by (mx,my)

Separate track matching in xand y for planes 5-9

Matching in #1 and #3 pixel planes in space

TUBE definition:

x-tube: xi = mx*(zi-z2) +bx +parabol(x6,z6,zi) +/- deltaxiy-tube: yi = my*(zi-z2) +by +/- deltayi

New algorithm

Mozaic DAQ systemTwo separate systems:

PRETRACKING network: Pixel [#2 , #4] + Strip [#6x]

TRACK-QUALITY TUBE network:Pixel [ #1, #3] + Strip[#5x, #5y, #6y, #7x, #7y, #8x, #8y, #9x, #9y]

In each network parallel CORRIDOR processors: CorID =corNUMNumber of CORRIDOR processors: ndx*ndy

Data select their routes according to plane number and corNUM

In plane „zi” track-hit „xi,yi” calculates its corridor address:

corNum = idx*ndy + idy

Corridor processorsOLD system: consecutive cycling on all „planes”

If only 2 points per plane: number of cycles = 2(4+2*5) = 214 = 16384

NEW system: cycling only on 3 „planes” (for pixels x and y has common cycle)

If only 2 points per plane: number of cycles = 2(2+1) = 23 = 8

The PRETRACKING is producing a list containing:

corNUM, x1,x3,x5,x7,x8,x9, y1,y3,y5,y6,y7,y8,y9

There is NO PROCESSING TIME in the TRACK-QUALITY TUBE network becauseIt is only an ASSOCIATIVE memory which provides YES/NO.

The gain in processing time (if only 2 points per plane): 211 = 2048-fold