Caribou Developers Meeting

Tuesday 4 Feb 2025, 16:00 → 17:30 Europe/Zurich

Younes Otarid (CERN)

16:00 → 16:10 General News

Speaker: Younes Otarid (CERN)

16:10 → 16:35 Hardware

Speakers: Eric Buschmann (Brookhaven National Laboratory (US)), Ryan St-Jean (Carleton University (CA)), Shaochun Tang (Brookhaven National Laboratory (US)), Tomas Vanat (Deutsches Elektronen-Synchrotron (DE)), Younes Otarid (CERN)

requirements_car_board_v2_0.xlsx

CaR board v2.0 specifications review:

• Pre-reviw notes
  • Anything with a strikethrough is set for deletion, upon acceptance from the team
  • Anything in dark orange is up for debate at the upcoming meeting
  • Yellow highlights can be ignored, they were notes from a previous level of review

   

• Review comments:
  
  • ID 11: rejected. 
    • We don't have these configuration and to this resoultion anymore. 
  • ID 17: rejected. 
    • We have unidirectional buffers  and series termiantion that could got to 50 Ohm
  • ID 19: accepted
    • 1 SFP+ should be enough 
    • QSFP+ could also be an option
  • ID 20: accepted
    • USB doens't take much space, and Mercury+ SoM already has interface for that. So unless there is a good reason drop it, we can keep it
    • Target 3.0
  • ID 22: on hold
    • It is fine as long as it doesn't take up too much resources or is complicated to implement
  • ID 30: rejected
    • More interesting is to sense the temperature on the chip board. But that is possible with I2C interface to the chip board. 
  • ID 34: rejected 
    • Should only focus on display port as adaptors to RJ45 can be made
  • ID 38: rejected
    • Can always use on board level shifters
  • ID 40: rejected
    • currents and voltages are now separate
  • ID 42: accepted
  • ID 44: accepted
  • ID 46: accepted
    • In the range of ~10mV
  • ID 48: accepted
  • ID 49: accepted
  • ID 55: rejected
    • about connecting the power supplies in parallel
    • Could be used in case we need higher currents, but current DACs already have limits of 1A up to 15V
  • ID 57: accepted
  • ID 58: rejected
    • Don't understand where the need is
  • ID 59: on hold
    • Need discussion
  • ID 60: on hold
    • Need discussion
  • ID  61: accepted
  • ID 62: accepted
  • ID 64: accepted
  • ID 65: accepted
    • Need to define the shutdown scheme
    • Possibility to use a QFN48 FPGA, with a flash SPI and I2C interface to configure the shutdown scheme
  • ID 67: accepted
  • ID 68: rejected
  • ID 69: accepted
  • ID 71: rejected
  • ID 72: accepted
  • ID 80: accepted
    • 1 channels for positive should be enough , and 2 for negative with 15mA (as it is tipically only needed for biasing)
  • ID 81: accepted
  • ID 84: accepted
    • We can also support Enclsutra XU8 with the notice that some features might not be available
  • ID 86: on-hold
    • Need discussion on the form factor to be used
  • ID 87: rejected
  • ID 88: rejected
  • ID 91: rejected
  • ID 92: rejected

16:35 → 17:00 Firmware

Speakers: Eric Buschmann (Brookhaven National Laboratory (US)), Tomas Vanat (Deutsches Elektronen-Synchrotron (DE)), Younes Otarid (CERN)

Younes:

• Worked on reorganisation of the Boreal firmware infrastructure:
  • Discarded single architecture for all supported devices. Too complex for what is really needed
  • Each project now has it's own Vivado project and custom architecture. This shall give more freedom and flexibility to users
  • Boreal will therefore only provide common modules/blocks, and an environment of automated scripts for creating, configuring, simulating and building the projects
  • A new boreal-modules Git project was created to openly distribute all common blocks to users. It is included as a submodule in the Boreal project.
  • Refer to the Boreal Gitlab to see latest structure. Documentation will follow shortly.

 

• Successfully closed on H2M support on ZCU102
  • Issue was that the petalinux image needed to be build with the hardware description file (XSA) extracted from Vivado
  • The reason is that by default the AXI interface on the Processing System in the default ZCU102 BSP provided by Xilinx is configured to be 128b. But on should configure thit to be 64b in the Vivado project and then use the XSA to build the Petalinux image
  • H2M can be controlled and configured on the ZCU102 board

 

• Next steps is to finalize the Peary generalization for multi-platform support

17:00 → 17:25 Software - OS

Speakers: Eric Buschmann (Brookhaven National Laboratory (US)), Mathieu Benoit (Oak Ridge National Laboratory (ORNL)), Simon Spannagel (Deutsches Elektronen-Synchrotron (DE)), Younes Otarid (CERN)

Younes:

• Worked on generalization of Peary class hierarchy using templates with the objective of supporting variable data width read/write regardless of the plateform used (32b or 64b).
• Refer to  my personal Peary branch for more information. A merge request will follow shortly.
• The new implementation was successfully exercised using the H2M firmware such that the H2M chip can be controlled and configured on both the ZC706 and ZCU102 boards
• Next step is abstracting the usse of the ZCU102 board in Peary. Still need to figure out best way to do so. 

17:25 → 17:30 AOB