1. March, 11, 2006 LCWS06, Bangalore, India Very Forward Calorimeters readout and machine interface Wojciech Wierba Institute of Nuclear Physics Polish Academy of Science Cracow, Poland

2. March, 11, 2006 LCWS06, Bangalore, India Contents Instrumentation of Very Forward Region Mechanical design of Calorimeters Space for electronics & cooling Pads & strips designs Digital Calorimeter Channel occupancy & dynamic range Machine interface requirements Readout based on analog transmission Digital data transmission Readouts based on digital transmission

3. March, 11, 2006 LCWS06, Bangalore, India Instrumentation of Very Forward Region BeamCal: 4 < θ < 28 mrad LumiCal: 26 < θ < 82 mrad PhotoCal: 100 < θ < 400 μrad

4. March, 11, 2006 LCWS06, Bangalore, India Mechanical design of LumiCal Segmented silicon sensors interspersed into the tungsten half disks Two half barrels to allow for mounting on closed beam pipe The blue bolts support the heavy part of the detector, tungsten half disks The red bolts carry only the sensors Holes for precision survey the sensors position

5. March, 11, 2006 LCWS06, Bangalore, India Mechanical design of BeamCal Diamond-Tungsten Sandwich Calorimeter Two half-barrels placed just outside the beam Thickness of the tungsten absorber disks is one radiation length Space for the sensors is ~0.5 mm

6. March, 11, 2006 LCWS06, Bangalore, India Space for electronics There is a limited space for the FE electronics for the 300 mm outer radius of LumiCal Bolts for tungsten & sensors support takes additional space Si sensors online positioning needs place also

7. March, 11, 2006 LCWS06, Bangalore, India Pads & Strips Design Both designs becomes very similar Different number of electronic channels: Strips ~8000, Pads ~25000 StripsPads

8. March, 11, 2006 LCWS06, Bangalore, India Digital Calorimeter Very high segmentation ~10 000 channels per plane = 300 k channels for calorimeter MC studies shows very good performance Preamplifiers, shapers, discrimators, zero suppression = simple (?) readout electronics (no need of ADC’s) Readout electronics have to be placed directly on the detector surface This is a brand new idea, details should be worked out

9. March, 11, 2006 LCWS06, Bangalore, India Channel occupancy & dynamic range The LumiCal mean channel occupancy is round 20% The BeamCal mean channel occupancy is even higher The zero suppression will not work efficiently The expected mean signal i one cell is ~25 MIPS For good event reconstruction we will need 14-bit ADC

10. March, 11, 2006 LCWS06, Bangalore, India Machine interface LumiCal can provide raw luminosity measurement after a bunch train – readout can be done in time slots between trains ~300 ms Precision luminosity can be calculated later (in order of seconds) BeamCal should provide beam diagnostics information ( σ x, σ y, σ z and Δσ x, Δσ y, Δσ z, x offset, y offset and Δx offset, Δy offset, Bunch rotation, N particles/bunch ) after a few bunches – readout should be done simultaneously with bunches every ~300 ns (~150 ns) Beamstrahlung Pair Analysis needs fast algorithm

11. March, 11, 2006 LCWS06, Bangalore, India Possible readout designs Analog transmission – simple, low heat dissipation, easy to maintenance – a huge number of cables, crosstalk, noise Digital transmission – small number of digital links, no crosstalk on the transmission – complicated, high heat dissipation

12. March, 11, 2006 LCWS06, Bangalore, India Analog transmission 25000 analog cables (coaxial or TP) needs ~30 mm round the LumiCal (can be reduced using multilayer flex PCB)

13. March, 11, 2006 LCWS06, Bangalore, India Digital Transmission LVDS link Current speed – 0.6 Gb/s Up to 10 m

14. March, 11, 2006 LCWS06, Bangalore, India Data stream for Forward Calorimeters 3000 bunches i one train – 1 ms Break between trains – 300 ms Number of channels – 25000 10 bit ADC Amount of data per train – 0.75 Gb Transmission during train – 750 Gb/s - ~1250 links Transmission during break – 2.25 Gb/s - ~4 links >>

15. March, 11, 2006 LCWS06, Bangalore, India Possible readout solutions ADC in each channel – expensive and space on chip consuming One ADC for a group of channels – cheaper, needs S/H, MUX and faster ADC Analog pipeline with ~3000 cells and 12 (14) bits accuracy over 300 ms is not a easy project, but the number of ADC can be reduced by a factor of ~300 Parallel conversion Multiplexed ADC Analog pipelined

16. March, 11, 2006 LCWS06, Bangalore, India Conclusions Number of channels and ADC resolution is essential to chose the readout solution Digital data transmission over the break between trains possible only for LumiCal The BeamCal should be read out simultaneously with bunches DAQ for the Forward Calorimeters has to read out each bunch crossing and the readout philosophy is different to the rest of Detector DAQ