Speaker
Description
The readout board for the ALICE TOF detector named DRM2 is now in the production phase: 88 boards are being produced (72 are needed in the experiment). Since the board will operate in a radiation environment (0.13 krad total dose expected in 10 years), a complete irradiation campaign at the component level was performed. We will focus on the Microsemi Igloo2 FPGA and two Avago optical transceivers radiation tests with a 100 MeV proton beam, available at the facility operated by INFN-TIFPA at the Centro di Protonterapia in Trento. This paper will also focus on the production test strategy.
Summary
In order to fulfill the new requirements on data acquisition performances given by the higher interaction rates both in proton-proton and in lead-lead collisions during RUN3 and RUN4 at ALICE, an upgraded version of the readout card has been designed and is now facing the production stage. The board features a few components which are in common with the previous-generation board, like an on-board ARM-processor mezzanine, and more up-to-date devices like the Microsemi Igloo2 FPGA and the GBTx ASIC from CERN. We launched an irradiation campaign at the Centro di Protonterapia in Trento with the goal to quantify the board components cross section to single event upsets (SEEs) as well as to evaluate the immunity from latchup and total irradiation damages up to a dose of a few krad (factor ~10-20 related to the expected dose in the real life experiment). In a previous irradiation campaign (in 2016) we tested several other components like static SRAM, clock and voltage drop regulators in addition to the Igloo2 FPGA (silicon revision 1) and the optical receivers FLTL8524P28BNL and FLTL8524P28BNV from Finisar. Worryingly, for the FPGA and these last two devices, we observed single event latchups.
Using a 100 MeV proton beam, we then studied in 2018 three Microsemi Igloo2 FPGA (using the latest available silicon revision number 3) and two Avago transceivers. For what concerns the FPGA, irradiated with a cumulative fluence up to 3 1011 p/cm2, we did not experience any latchup events . We also measured a cross section of single event upsets on the internal BRAM as low as 2 10-14 cm2 per bit. It is worth mentioning that all the FPGAs were still reprogrammable after receiving a dose of 3.75 krad even if, in one case, a certain annealing time was needed before the FPGA recovered the re-programming capability. Concerning the Avago optical transceivers, we used a sample of 4 AFBR-57R5AEZ and 4 AFBR-57R5APZ. No-latchup events and just one SEE on the former type were observed.. More detailed results on the radiation tests performed are given in this paper.
88 DRM2 boards are currently being produced at CAEN in order to provide for the 72 cards required for the experiment plus spares. The production test strategy is built upon a 3-stage approach: a first test is done at the manufacturer site (CAEN), followed by a second test performed at CERN using the same setup system as at CAEN. Before their insertion in the detector (planned during 2019), the boards are finally tested in the custom ALICE TOF VME crate where all real life functionalities (e.g. read-out of the full crate at sustained rate) are tested together. DRM2 boards successfully passing all the 3 steps will be then installed in the experimental crates while the commissioning stage in situ will start in late 2019.
This paper gives a thorough description of the production test setup initially built in Bologna and then replicated both at CAEN and at CERN.
