11-th CBM Collaboration Meeting. GSI Darmstadt Feb , A New Data Acquisition System based on Asynchronous Technique Yu. Bocharov, A. Gumenyuk, A. Klyuev, A. Simakov
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , To compare architectures Analog FIFO per channel vs. Dig.FIFO per ADC To estimate a data loss for systems based on architectures compared by a Monte Carlo modeling To define the ADC specifications Objectives
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , A New Readout System Architecture A New Readout System Architecture As an example – 2 ADC per 128 AFE channels Other variants – 1, 4 ADC PD – peak detector Main Feature Digital FIFO per ADC against Analog FIFO per channel
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Step 1 When a hit occurs in a channel PD locks the this channel in and sends EVENT signal to the control unit
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Step 2 The control unit writes a channel number and a time stamp into a Dual-port Memory/FIFO. Any type of arbiter may be used to prevent conflicts of writing
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Step 3 The High Speed ADC converts the outputs of channels which numbers are stored in FIFO
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Step 4 Converted data conjunctly with a channel number and a time stamp are transmitted to the external memory bus
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Step 5 When the conversion is finished a control unit initializes PD and corresponding MEM content and connects ADC to the next channel or switches it to a shutdown state if FIFO is empty
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Structure of the arbitration logic for analog de-randomizer (2007) Structure of the arbitration logic for analog de-randomizer (2007)
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Simulation of the synthesized arbitration logic Simulation of the synthesized arbitration logic
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Arbitration logic Area Estimation (Encounter, Faraday standard cells, UMC 0.18) Arbitration logic Area Estimation (Encounter, Faraday standard cells, UMC 0.18)
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Total of hits per cycle probability (%) for 128 (a), 64 (b), 32 (c) channel system at 5% channel occupancy for Poisson process Mean – 6.4, 3.2, 1.6 a b c
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , % MSPS CBM-XYTER data loss as a function of total ADC-channels throughput at the best (1) and worse (2) – numerical simulation
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , λ – Poisson distribution parameter, M – number of channels per ADC, θ – channel occupancy, μ – max number of channels may be A-D converted within one cycle Analytical estimation of a data loss for a new readout system Analytical estimation of a data loss for a new readout system
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , % MSPS Data loss of a new readout system as function of ADC 32 (1), 64(2), 128(3) channels per ADC – analytical and numerical
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Figure of merit (FOM) commonly used for ADC characterization P d – power dissipation ENOB – effective number of bits f s – sampling frequency (ENOB spec)
11-th CBM Collaboration Meeting. GSI Darmstadt Feb , Summary CBM-XYTER 120 MSPS, 128 mW, FOM < 10.7 pJ - 1 AD/chip 60 MSPS, 64 mW, FOM < 10.7 pJ - 2 AD/chip 30 MSPS, 32 mW, FOM < 10.7 pJ - 4 AD/chip Advantage – reduced ADC requirements New architecture 160 MSPS, 128 mW, FOM < 8.0 pJ - 1 AD/chip 105 MSPS, 64 mW, FOM < 6.1 pJ - 2 AD/chip 75 MSPS, 32 mW, FOM < 4.3 pJ - 4 AD/chip Advantage – elimination of 512 analog MEM cells ADC Pd = 1 mW/channel and ENOB = 6.6 bit (100 quantization levels) Max data loss level = 0.01%